English
Based on โHustle culture stole the word excellence and gutted its true meaning | Brad Stulberg: Full Interviewโ from Big Think Watch the original video
Beyond the Grind: Reclaiming the Soul of Excellence
In a world obsessed with early mornings, endless productivity hacks, and the relentless pursuit of โmore,โ the word โexcellenceโ has been distorted, stripped of its true power. Itโs become synonymous with a frantic, often joyless, hustle culture that prioritizes external validation and superficial achievement over genuine fulfillment. But what if real excellence isnโt about perfectly optimized routines or boasting about your latest triumph? What if itโs a deeply human, heartfelt endeavor that quietly shapes who you become?
Brad Stulberg, author of The Way of Excellence and faculty at the University of Michigan, argues that true excellence has been stolen and gutted. He invites us to reclaim its authentic meaning, offering a profound perspective that challenges modern assumptions and illuminates a path to a more meaningful life.
The Kabuki of Hustle vs. Genuine Engagement
Stulberg contends that the popular image of excellenceโwaking at 4 AM, following a 47-step routine, taking cold plunges, consuming nine supplements, meticulously tracking every metric, and broadcasting your success to the worldโis nothing more than โelaborate kabuki masquerading as the real thing.โ This performative display, while perhaps impressive on social media, misses the essence.
Instead, Stulberg offers a simple, yet revolutionary, definition: Excellence is involved engagement in something worthwhile that aligns with your values.
Itโs not about reaching the summit of the mountain; itโs about the person you become on the journey up its sides. Consider the difference between aimlessly tapping a white wall (involved, but not worthwhile or values-aligned) and striving to become a great physician because you value intellect, contribution, and helping people. The latter imbues the pursuit with meaning.
When we work on big goalsโrunning a marathon, starting a company, writing a bookโthese endeavors also work on us. The marathon shapes our character, teaching resilience, consistency, and self-kindness. The book-writing process molds us, revealing lessons about overcoming setbacks and the power of discipline. If these goals align with our core values, they build a character concordant with who we aspire to be, leading to a profound sense of mastery (competence and tangible progress) and mattering (belonging to something bigger than our ego).
Our core values, typically three to five, are the compass for this journey. Whether itโs mastery, craft, integrity, community, spirituality, discipline, or kindness, these arenโt just words on a poster. They must be defined and lived, guiding daily actions and goal selection. As Stulberg wisely points out, if you value health but choose a goal that demands 22-hour workdays and no sleep, youโre setting yourself up for discord.
The Deep Roots of Our Drive to Flourish
The concept of excellence, far from being a modern invention, is deeply ingrained in our very being. Stulberg draws on Robert Persigโs idea of โQualityโ from Zen and the Art of Motorcycle Maintenance, which Persig stylized with a capital Q. This Quality describes the intimate relationship between an actor and their act, where subject and object mergeโitโs not Rachel playing basketball, โbasketball is just happening.โ
This โQualityโ tracks with what modern evolutionary biologists call homeostatic upregulation: an innate, hardwired drive in all living species to persist and flourish. From bacteria sensing their way toward conducive environments to early organisms feeling their way forward, this drive predates cognition. Humans, with our complex thoughts, can now consciously direct this primal urge toward flourishing beyond mere survival and procreation. Itโs this same drive that leads us to pursue athletic feats, create art, contribute to society, or become better parents. We all possess this innate drive to grow and aspire toward excellence.
Sociologist Richard Sennettโs concept of situated cognition further illuminates this. Studying craftspeople, Sennett observed that their best, most fulfilling work emerged not from intellectual thought alone, but from โthinking with their entire being.โ Itโs a mind-body relationship where one feels โsituatedโ and at home in the world, feeling their way forward.
This journey of mastery unfolds through four phases of competence:
- Unconscious Incompetence: You donโt know what you donโt know. (Requires learning, coaching.)
- Conscious Incompetence: You know you donโt know. (You have a schema but arenโt there yet.)
- Conscious Competence: You know youโre doing it right, but itโs effortful. (Following checklists, tracking.)
- Unconscious Competence: Youโve transcended effortful thought, entering โthe zoneโ where feelings guide your greatest performances.
Masters of craft, Stulberg explains, donโt stay in unconscious competence. They cycle through these phases, knowing when to return to conscious effort and when to trust the โmagicโ of flow.
The Modern Minefield: Six Barriers to Genuine Excellence
In an increasingly distracted and alienated world, the pursuit of excellence offers intimacy with oneself and oneโs craft. But many modern traps lie in wait:
- Disevolution: Coined by Daniel Lieberman, this term describes how our modern environment presents traps we didnโt evolve for. Our ancient wiring, honed for scarcity, now struggles with abundance (e.g., Krispy Kreme). Similarly, our need for community is co-opted by social mediaโs fleeting validation, which feels good in the short term but leaves us empty compared to genuine human connection.
- Optimization Culture: In an attempt to control the chaos of modern life, we risk turning ourselves into robots. Obsessively tracking every calorie, sleep cycle, and heartbeat can suck the feeling, humanity, and joy out of life. Stulberg shares the anecdote of golfer J.J. Spahn, who won the U.S. Open despite a night of no sleep due to his childโs illness. An optimization device would have told him to stay home. This culture makes us fragile, assuming perfection is a prerequisite for performance.
- Shitty Flow: Psychologist Mihaly Csikszentmihalyiโs concept of โflowโ (total immersion) is powerful, but Stulberg introduces David Pizarroโs term โshitty flow.โ You can be in flow scrolling Twitter or gamblingโexperiences that light you up with a โpseudo alivenessโ that is disconnected from your values. The key differentiator? How you feel after the experience. Regret, anxiety, or feeling โgrossโ are hallmarks of shitty flow. True excellence, by definition, cannot be shitty flow because it must align with values. Flow is a small component of excellence, not its entirety, encompassing the effort before and the reflection after.
- Zombie Burnout: Beyond the conventional burnout from overwork, Stulberg identifies โzombie burnoutโโa numbed, apathetic state arising not from doing too much, but from not doing enough of what lights you up. Even a 30-hour workweek can lead to exhaustion and dread if youโre just going through the motions. The antidote? Meaningful friction. Job crafting (massaging your work to align with values) or engaging in active leisure (reading, gardening, training for a marathon instead of passive consumption) can inject the necessary engagement. Stulberg himself pursues excellence as a husband/father, a writer, and a powerlifter (deadlifting).
- The Happiness Industrial Complex: The modern obsession with chasing happiness is a trap. Happiness, like any emotion, is ephemeral and a byproduct, not a direct pursuit. Stulberg uses the โhappiness tubeโ thought experiment (euphoria for life, but disconnected from reality) to illustrate that most people choose a life of challenge and meaning over constant, artificial happiness. Life is full of discomfort and heartbreak; true flourishing comes from chasing satisfaction, meaning, and fulfillment, which often bring happiness as a bonus.
- Excellence as a Standard: We often mistakenly view excellence as an outcomeโwinning an Olympic medal, graduating with honors. These achievements are wonderful, but they represent the โpeak of the mountain,โ which accounts for a minuscule fraction of life. โThe only Zen you find on the tops of mountains is the Zen that you bring up there with you,โ Persig famously said. The growth, the โZen,โ happens on the sides. Ray Allen, a legendary basketball player, found that even winning an NBA championship didnโt fill the โhole inside,โ leading to post-championship disorientation. This โarrival fallacyโ highlights that fulfillment resides in the process of striving, not solely in the achievement.
Cultivating Genuine Excellence: Essential Factors
While Stulbergโs book details 16 factors of excellence, he highlights three critical ones:
- Goals: Goals provide a target, a โpeakโ to strive for, giving shape to your journey. However, the mindset must be โprocess over outcomes.โ This involves:
- Setting a big aspirational goal.
- Breaking it down into smaller, manageable micro-milestones.
- Largely forgetting the big goal and focusing on daily/weekly steps.
- Using stress about the big goal as a cue to re-immerse in the process.
- Stepping back on the path when you inevitably fall off. SMART goals (Specific, Measurable, Actionable, Relatable, Time-bound) are a useful framework for creating concrete, actionable targets that hold you accountable. The pursuit of a 600-pound deadlift isnโt just about the lift itself; itโs about the years of training that teach resilience, community, and discipline.
- Consistency: True progress isnโt about heroic, one-off days (the โgo till you pukeโ workout or the all-nighter). Itโs about stringing together โmany very good days over long periods of time to create a heroic year or a heroic decade or a heroic body of work.โ While there are rare moments to โgo see Godโ with intense effort, these are only possible through consistent foundational work. The popular โget 1% better every dayโ mantra is a powerful mindset, but progress isnโt linear. Stulberg shares the story of world champion powerlifter Lane Norton, who added only 8 pounds to his deadlift in 10 years. This demonstrates that unwavering consistency, fueled by curiosity and commitment, is crucial during inevitable plateaus, building experience even when observable progress is slow.
- Patience: Implicit in the discussion of consistency and plateaus, patience is the understanding that significant growth takes time and that the journey itself is the reward.
The Ultimate Antidote
In a world that constantly pulls us towards superficiality and fleeting highs, the pursuit of genuine excellence stands as a powerful antidote to alienation and โzombie burnout.โ Itโs a philosophy of life that fosters intimacy with our craft and ourselves, connecting us deeply to what we do and who we are.
Excellence is not perfectionism, obsession, or optimization. Itโs not a hedonic chase for happiness. It is, Stulberg concludes, a deeply human and humane pursuitโour birthright, making us feel alive and good in the world. By embracing involved engagement, aligning with our values, and focusing on the transformative journey, we reclaim excellence and unlock a life rich with mastery, mattering, satisfaction, and fulfillment.
Based on โLIVE VIBE CHECK: OPUS 4.7 DROPSโ from Every Watch the original video
Opus 4.7 Drops: A Live Look at AIโs New Frontier and the โGreat Convergenceโ
The world of artificial intelligence is moving at a breakneck pace, and nowhere was this more evident than with the recent, unannounced launch of Anthropicโs Claude Opus 4.7. Eschewing the traditional early access programs, Anthropic pushed their latest flagship model into the wild, leaving AI enthusiasts and practitioners scrambling for real-world insights. Enter Every.to, a leading voice in AI analysis, which immediately launched a live โvibe checkโ โ a real-time, hands-on testing session designed to cut through benchmarks and uncover the true capabilities of the new model.
โNormally at Every, we get access to all these models before they come out,โ explained Ryan, the host, alongside Brandon, COO, and Katie, a fellow tester, all sporting matching green shirts for the occasion. โThis time we were snubbed. Anthropic snubbed us.โ This unexpected turn meant that Everyโs renowned โvibe checksโโdeep dives into model performance across coding, writing, design, and strategyโwould be conducted live, in front of thousands of eager viewers.
Beyond the Benchmarks: The โVibe Checkโ Philosophy
Anthropicโs official announcement for Opus 4.7 touted it as โour most capable Opus model yet,โ promising enhanced rigor in long-running tasks, more precise instruction following, and a novel feature: self-verification of output before reporting back. Ryan noted this last point as particularly significant, suggesting it reflects a best practice known to good prompters now being integrated directly into the modelโs core behavior. Benchmarks like Swebench Pro and SweetBench verified showed modest but respectable bumps of 10% and 7% respectively.
However, for the Every.to team, benchmarks are merely a starting point. โWe donโt really like benchmarks,โ Ryan stated. โI really like getting your hands in the model and using it for the things that you do every day.โ This philosophy underpins their โvibe checkโ methodology, which aims to provide practical insights for real-world applications. The live session was designed to put Opus 4.7 through its paces, using a range of tasks from financial analysis and advanced coding to creative writing and agent-native document editing.
Initial Impressions: A Shift Towards Literalism
One of the most striking initial observations, echoed by community feedback during the live stream, was a perceived shift in Opus 4.7โs interpretive style. โOpus 4.7 interprets prompts more literally and explicitly than 4.6,โ noted Ryan, referencing an Anthropic announcement. This contrasted sharply with the historical strength of Opus models, which had been celebrated for their โempatheticโ understanding compared to the โautistic and literalโ nature often attributed to early GPT models.
Brandon, testing Opus 4.7โs ability to create an โOpenClawโ (a personalized AI agent based on his usage and memories), found Opus 4.6 to be โmore organized.โ While 4.6 correctly generated distinct user.md and soul.md files and new skills based on his Claude usage, 4.7 โdidnโt make a user file [or] soul file but it did make an agent file and it just shoved everything about soul or everything about user in the agent file.โ Furthermore, 4.7 generated skills heโd already made or skills unrelated to his actual use cases, leading Brandon to conclude, โIโd probably stick with 4.6 for making my OpenClaw like this.โ He also noted 4.7 was โa lot slowerโ in this particular task.
Katie, a writer, put Opus 4.7 to the test by having it draft an article introduction in her voice, comparing it directly to Opus 4.6. While 4.6 produced a piece with โunpredictableโ cadence and โclever linesโ that resonated with her personal style, 4.7โs output was โa little bit more systematic in a way that like just doesnโt vibe with how I am as a writer.โ She found 4.7โs writing to have a โregimentedโ feel, suggesting a more literal interpretation of her style guide rather than an intuitive capture of her voice.
Conversely, Ryan found Opus 4.7 excelled at a task requiring directness over flourish: writing an investor update based on Every.toโs P&L (profit and loss) statement. โThis is actually really good,โ he remarked. โIt did a really good job of doing all the research and getting all the numbers right and also writing it in a way that I would write it.โ The modelโs direct tone, he noted, was perfectly suited for an investor update, potentially saving significant time compared to 4.6, which he felt โgoes too farโ or โdoes too muchโ in such contexts.
The โVibe Slopโ Benchmark: A Senior Engineerโs Challenge
The most ambitious test of the session was Ryanโs custom โVibe Slop Benchmark,โ designed to assess an AIโs ability to act like a senior engineer. The benchmark involves giving the model a snapshot of Proof, Every.toโs agent-native document editor, at a point where its codebase was โpurely vibe codedโโmeaning it was hastily built, prone to errors, and lacked proper architectural planning. The challenge: rewrite the sloppy codebase from first principles, mirroring the work a senior engineer would do.
Ryan explained the core problem with the โvibe slopโ Proof codebase: โWhen you have a collaborative document editor, you have multiple versions of the same document open in different browsersโฆ the trick for getting that right is knowing which version of the document is the authoritative document.โ The original codebase had โ15 different authorities and all these different like ways of merging them.โ A senior engineer, he noted, would immediately identify the need for โone authority source.โ
Opus 4.7โs initial plan was impressive. Without any hints, it correctly identified the central architectural flaw: โthe problem is that there is no single authoritative model of who owns the current state of document X right now and who may write to it.โ It even proposed a refactor for a massive collab.ts file (10,000-20,000 lines), an issue that previous GPT models (5.3 and 5.4) had failed to notice in their plans.
However, the โliteralโ theme resurfaced. When Ryan asked it to โexecute end to endโ after presenting the plan, Opus 4.7 didnโt automatically understand he meant to execute the plan. It also initially proposed a โcautiousโ implementation strategy rather than the โburn the ships rewriteโ Ryan envisioned. โThis is good but itโs still a little bit too cautious,โ he prompted. โCan you change the implementation strategy so itโs like truly just a burn the ships rewrite that we like can do today?โ This demonstrated that while Opus 4.7 could identify complex problems and propose solutions, it required more explicit guidance for truly ambitious, unconstrained executionโa trait often associated with autonomous senior engineering.
The Great Convergence Continues
The collective experience of the live vibe check pointed towards a fascinating trend: โthe great convergence.โ Ryan elaborated on this concept, previously explored in Everyโs analysis of GPT 5.3 versus Opus 4.5. โIt looks like the models are converging,โ he explained. โIt looks like Opus is trying to do all the things that GBT can do and GBT is trying to do all things Opus can do.โ The shift in Opus 4.7 towards a more literal, systematic approachโreminiscent of earlier GPT modelsโwhile OpenAIโs models are reportedly becoming more โemotionally intelligent,โ suggests a fascinating cross-pollination of characteristics. โIt seems like theyโre sort of sort of like switched spots,โ Ryan mused, โand now they pushed Opus to be a little bit more literal autistic senior engineery.โ
Staying at the Edge
As the live vibe check concluded, the Every.to team emphasized that while initial results were mixed and often surprising, Opus 4.7 is a powerful new model that will likely require practitioners to adapt their prompting strategies. Its strengths in direct financial analysis and identifying core architectural flaws in code are noteworthy, even if its creative writing and agent organization didnโt always align with previous Opus iterations.
For those eager to stay abreast of these rapid developments, Every.to offers a unique ecosystem of ideas, apps, and training. Their daily articles provide in-depth analysis of new models, their suite of AI apps (like Proof, Monologue, Kora, Spiral, Sparkle, and Plus Ones) offers practical tools for leveraging AI, and their live streams and courses provide ongoing education. โIf you want to stay at the edge,โ Ryan urged, โit is absolutely the thing to subscribe to.โ
The Opus 4.7 launch and its subsequent live vibe check serve as a potent reminder of the dynamic, ever-evolving landscape of AI. Models are not static; they learn, they shift, and sometimes, they surprise even their most seasoned testers. The โgreat convergenceโ continues, pushing the boundaries of whatโs possible, and demanding constant adaptation from those who seek to harness its power. The initial vibe check on Opus 4.7 confirms itโs a powerful tool, but one that will require a new understanding and approach to unlock its full potential.
Based on โ671. Why Has There Been So Little Progress on Alzheimerโs Disease? | Freakonomics Radioโ from Freakonomics Radio Network Watch the original video
The Unsettling Truth Behind Alzheimerโs Stalled Progress
Recent headlines have heralded the arrival of new FDA-approved blood tests capable of detecting Alzheimerโs disease in its nascent stages. Early detection is a crucial step for any illness, but particularly so for Alzheimerโs, a devastating condition that can silently take root for two decades before overt symptoms like memory loss and cognitive decline emerge. Affecting over 7 million Americans, predominantly those over 65, Alzheimerโs disease represents a significant public health challenge.
Despite its historical presence โ age-related memory loss has been observed for centuries, though the disease was formally documented by Alois Alzheimer in 1906 โ and billions of dollars poured into research, effective treatments remain elusive. The National Institutes of Health (NIH) now allocates roughly $4 billion annually to Alzheimerโs and dementia research, a substantial increase from a decade ago, placing it second only to cancer spending. Yet, progress on a cure remains frustratingly slow.
This stagnation has led some to question the very foundation of Alzheimerโs research. โNo oneโs getting better with these drugs,โ states investigative journalist Charles Piller. โEvery scientist who works with them, every clinician, will say the same. If they donโt, theyโre lying.โ The stark reality prompts a disturbing question: Is the dominant theory guiding research merely flawed, or has it been tainted by something far more insidious โ fraud?
This is the unsettling premise explored by an investigation led by Piller and neuroscientist Matthew Schrag, which uncovered decades of problematic research potentially derailing the quest for a cure.
The Dominant Hypothesis Under Scrutiny
For much of its history, Alzheimerโs research has been anchored to a single, powerful idea: the amyloid cascade hypothesis. This theory posits that the accumulation of a protein called beta-amyloid (or simply amyloid) outside brain cells, forming โplaques,โ is the primary trigger. This, in turn, leads to a cascade of events, including the clumping of another protein called tau inside brain cells, ultimately causing neurons to malfunction, die, and the brain to shrink, resulting in memory loss.
Neuroscientist Matthew Schrag, an associate professor of neurology at Vanderbilt University Medical Center, explains this widely accepted model. โThe main thought in this field is that the plaques, this little beta-amyloid protein that clumps outside of the cells, is the primary driver of the disease,โ he says. โA huge amount of the effort related to drug development has been dedicated to preventing it from accumulating or breaking it up and clearing it from the brain.โ
However, Schrag, whose own research delves into the connection between Alzheimerโs and blood vessel diseases, challenges this singular focus. โThe problem is one, the idea may not be correct and two, the biology of this is likely much more complicated,โ he asserts. โI canโt think of another disease where there is one single first domino that falls that causes the rest of the cascade.โ He argues that by targeting just one specific factor, researchers might not be having a significant enough impact on the disease.
The Whistleblowerโs Unforeseen Path
Schragโs journey into research integrity was not one he anticipated. Early in his career, while pursuing his MD and PhD, he was encouraged to engage in research, leading him to a lab studying Alzheimerโs. Years later, his path took an unexpected turn when he was approached by an attorney representing whistleblowers with concerns about the integrity of data supporting an experimental Alzheimerโs drug called simufilam, developed by Cassava Sciences.
The companyโs research collaborator, Hoau-Yan Wang, a professor at City University of New York, had published work suggesting simufilam could reverse the misfolding of a protein called filamin A, and reduce the buildup of tau and beta amyloid. The whistleblowers, a cardiologist and a neuroscientist, found the drugโs pharmacology questionable and suspected image manipulation in the supporting data.
Schrag, then a junior assistant professor, agreed to help. His forensic analysis of the image data revealed alarming patterns. โThe images just didnโt fit what a scientific expectation would be,โ he describes. โIt looked visually manipulated. Some of these results had copy and paste type changes where it looked like things had been moved around or duplicated or where the intensity of one part of the image had been adjusted.โ These adjustments, he explains, directly alter the underlying data, as the darkness of โsplodgesโ on a photograph is often whatโs measured to quantify proteins.
Schrag compiled a detailed dossier and sent it to the FDA, expecting swift action. He was โshockedโ by the slow pace and the ultimate decisions. It was during this process that a critical pattern emerged: โcheaters tend to cheat.โ Schrag realized the problems extended beyond Cassava Sciences, encompassing Wangโs broader body of work and that of his collaborators. This led Schrag to send additional dossiers to the NIH, aiming to correct the scientific record more broadly.
The Retraction of an Influential Paper
Schragโs work caught the attention of Charles Piller, an investigative journalist with Science Magazine and author of Doctored: Fraud, Arrogance, and Tragedy in the Quest to Cure Alzheimerโs. Piller recognized the expanding scope of the issue, and the two collaborated.
Their investigation soon focused on a hugely influential study published in the journal Nature in 2006, titled โA Specific Beta-Amyloid Protein Assembly in the Brain Impairs Memory.โ Co-authored by University of Minnesota researchers Sylvain Lesnรฉ and Karen Ashe, the paper proposed that small, soluble clumps of beta amyloid, called oligomers, were the primary toxic component driving Alzheimerโs. This โsilver bulletโ finding came at a crucial time, reinvigorating confidence in the amyloid hypothesis after a vaccine-based clinical trial had failed. Lesnรฉ and Asheโs paper became one of the most cited in the history of Alzheimerโs research.
Piller and Schrag, along with other forensic image experts, meticulously examined the paperโs evidence, particularly its Western blot images. Western blots are photographic representations used to identify and quantify specific proteins in a sample. What they found was damning: โThese images apparently were manipulated in severe ways that tended to support the hypothesis of the experiment when the actual data did not support it,โ Piller recounts.
When confronted, Lesnรฉ refused to speak with Piller. Ashe, his co-investigator, initially expressed a willingness to talk but ultimately declined on her universityโs advice. Piller expressed frustration at this lack of transparency, especially given that universities receive significant public funding.
The NIH, the funder of Lesnรฉโs work, acknowledged a possible problem and referred it to the Office of Research Integrity (ORI). However, the ORI, a small agency, typically refers such cases back to the institution where the alleged misconduct occurred โ in this instance, the University of Minnesota. Piller considers this a flawed approach, as it entrusts the investigation to the entity with โthe most to lose and the least to gain from a robust, complete, and publicly accessible examination of possible misconduct.โ
The universityโs assessment of Lesnรฉ took years, during which he continued to receive federal funding for similar research. Finally, in 2025, Lesnรฉ retired from his tenured professorship. The universityโs findings were never publicly released, save for a recommendation that other Lesnรฉ articles be retracted. The seminal Nature paper was ultimately retracted in 2023, with Karen Ashe formally requesting it and acknowledging data problems, though she maintained no prior knowledge of them.
A Broader Pattern of Misconduct
The issues uncovered were not isolated incidents. Pillerโs continued investigation into Cassava Sciences revealed that the City University of New Yorkโs internal report found numerous signs of image manipulation and poor data management in Hoau-Yan Wangโs work. In 2024, the Department of Justice indicted Wang for allegedly fabricating and falsifying data in NIH grant applications related to simufilam. Cassava Sciences later agreed to pay $40 million to the Securities and Exchange Commission to settle charges of misleading investors, and the company discontinued development of simufilam after it failed in human clinical trials. Wang also paid a fine related to the SEC charges. Notably, in 2025, all charges of fraud made by the DOJ against Wang were dropped, with his lawyers asserting his innocence based on new evidence.
Pillerโs research extended further, examining the track records of influential Alzheimerโs researchers. With the assistance of four leading forensic image experts, including Matthew Schrag, he scrutinized the work of Eliezer Masliah, who had served as head of the Division of Neuroscience for the National Institute on Aging (NIA), overseeing a $2.7 billion annual budget. Masliah, previously an eminent researcher at UC San Diego, was a highly influential figure in steering neuroscience research.
Pillerโs team identified โ132 papers out of 800 where image irregularities including apparently doctored images, apparently misused or reused images, apparently images that were published separately in different papers in an improper way to make different scientific points.โ This led to a 300-page dossier detailing concerns spanning 30 years of Masliahโs work. When Piller presented this information to Masliah, he received no response. Days before Pillerโs story was published in Science magazine, the NIH announced that Masliah was no longer in his job, declining to comment on personnel matters.
The Personal Cost of Integrity
Perhaps the most difficult revelation for Matthew Schrag came when his own early research was scrutinized. After his work on Cassava Sciences, he anticipated that his own publications would be thoroughly examined. He was astonished when someone pointed out anomalies in images from his first papers, co-authored with his undergraduate mentor, Othman Ghribi.
Schragโs work with Ghribi, looking at plaque buildup in rabbits fed a high-cholesterol diet, was published in 2006. The suggestion was that microscope images of neurons looked โtoo much alike,โ hinting at copying and pasting. Ghribi, who was responsible for compiling the figures, admitted to the manipulation when Schrag confronted him years later. Ghribi claimed he thought the underlying science was correct and was merely trying to make the images โa little bit clearer,โ equating it more to โexaggeration than manipulation.โ He also claimed that โeverybody did this,โ a assertion Schrag strenuously refutes.
The discovery was devastating for Schrag. Ghribi had been a profound mentor, remaining an advisor, attending Schragโs graduation and wedding. โThis was somebody who was more than just a passing academic acquaintance,โ Schrag explains, highlighting the deep personal cost of whistleblowing. Ghribi eventually disposed of his laboratory records, telling Schrag he was โvery ashamed.โ
Despite Ghribiโs initial willingness to speak out to exonerate Schrag, he ultimately declined to speak publicly, which Schrag views as a โsad mistake.โ The 2006 rabbit paper, along with other joint papers from Schrag and Ghribi, were retracted in 2023. Ghribi, who retired from the University of North Dakota in 2020 and now teaches at the University of Texas Rio Grande Valley, acknowledged in an email that โsome polishing could have been doneโ but insisted โno fabrication was made,โ taking โfull responsibility as I am the laboratory director.โ He also stated that the images in Schragโs manuscripts were โthe images he handed to me.โ
A Call for Accountability and Trust
Piller contends that Alzheimerโs research has been โhijacked by a cabal of self-interested researchers, government accomplices, and corporate greed.โ He points to a troubling phenomenon known as โregulatory captureโ at the FDA, citing examples like Billy Dunn, who oversaw the approval of the ineffective and dangerous Alzheimerโs drug aducanumab (Aduhelm). Dunn, investigated by STAT and Congress for working closely with the drugโs manufacturer, later took a senior role on the board of a biotech company developing Alzheimerโs remedies, exemplifying a โrevolving doorโ problem.
In an era of growing distrust in science, Piller acknowledges the delicate balance of his work. While disagreeing with attacks on science, he believes critics are not entirely wrong about โarrogance and complacency and in my view condescension on the part of many people in the scientific research enterprise towards members of the public.โ Trust, he emphasizes, โmust be earned.โ
The revelations of alleged misconduct and systemic failures in Alzheimerโs research are sobering. Billions of dollars have been spent, and countless lives affected, with little to show in terms of effective treatments. However, the work of investigative journalists and whistleblowers like Charles Piller and Matthew Schrag, though painful, is critical. By shining a light on these issues, they aim to foster greater accountability, transparency, and integrity within the scientific community, ultimately clearing a path for a more robust and trustworthy pursuit of solutions for one of humanityโs most challenging diseases. The hope is that by confronting these uncomfortable truths, Alzheimerโs research can finally shift direction and deliver on its promise.
Based on โScaling Global Organizations in the Age of AI with ServiceNow CEO Bill McDermottโ from No Priors: AI, Machine Learning, Tech, & Startups Watch the original video
Beyond the โSaaS Apocalypseโ: Bill McDermott on AI, Human Connection, and the Unstoppable Platform
In an era of unprecedented technological acceleration, where the very foundations of enterprise software are being questioned by the rise of artificial intelligence, one leader stands out with a uniquely grounded yet visionary perspective. Bill McDermott, CEO of ServiceNow, often dubbed โthe closest thing we have to a rockstar in enterprise technology,โ offers a compelling counter-narrative to the prevailing anxieties about a โSaaS apocalypse.โ For McDermott, AI isnโt a threat to established platforms but a powerful accelerant, and the future of business leadership, surprisingly, lies in the enduring power of human connection.
Speaking on the โNo Priorsโ podcast, McDermott delves into the intricacies of scaling global organizations in the age of AI, sharing insights forged from a remarkable career that began not in a boardroom, but behind the counter of a deli.
The Deli: A Masterclass in Customer Centricity
McDermottโs journey into leadership began unconventionally. At just 16, he consolidated multiple part-time jobs into one by purchasing a deli. This seemingly strange decision laid the groundwork for his lifelong philosophy: โItโs all about the customer. In the end, the customer and the customer alone determines whether you win or lose.โ
His deli wasnโt just a place to buy food; it was a community hub where he learned invaluable lessons in emotional intelligence (EQ). He meticulously understood his three core customer segments: the blue-collar workers, โrich on Friday and dead broke by Sunday morningโ; the senior citizens who relied on his unique delivery service; and the neighborhood kids, lured past the 7-Eleven by the irresistible draw of video games like Asteroids and Pac-Man.
McDermott recounts a particularly insightful moment when a young customer articulated the deliโs unique value proposition: โWhen we want to have good food, be treated with dignity and respect, and play video games, we come to your store. And when we want to steal stuff, we go to 7-Eleven.โ This anecdote perfectly encapsulates his belief that understanding and serving the customerโs specific needs, with dignity and respect, is the fundamental differentiator in any business. This early immersion in human interaction, dealing with 500 customers a day, instilled an EQ that McDermott believes the world desperately needs more of today.
The Xerox Interview: A Battle for Survival
Years later, McDermottโs ambition led him to an interview at Xerox, then considered โthe Google of its era.โ Freshly turned 21, he promised his father heโd return with an employee badge. This wasnโt just an interview; it was, as he describes it, โa battle for survival.โ
Inspired by CEO David Kearnsโs vision to transform Xerox using quality management principles from Japan, McDermott was determined to get his shot. During his final interview, with the โbig bossโ at 9 West 57th overlooking Central Park, he refused to accept a polite โHR will be in touch.โ Instead, he declared, โMr. Fullet, I donโt think you completely understand the situation, sir. I havenโt broken a promise to my father in 21 years, and I guaranteed him Iโm coming home tonight with my employee badge in my pocket.โ
This bold display of agency and unwavering commitment earned him the job, a moment so memorable that 25 years later, Emerson Fulwood, the executive who hired him, recounted it as the only time he ever broke company policy. This story underscores McDermottโs belief in the power of human connection, mentorship, and the conviction to seize opportunities. โAll people should ever want is an opportunity,โ he asserts, โbecause an opportunity is a gift.โ He sees AI as a tool to โserve people and to make the human ambition greater, not to take it away from us,โ a core tenet he carries into the age of AI.
Debunking the โSaaS Apocalypseโ
The conversation quickly shifts to the current landscape, dominated by the buzz around large language models (LLMs) and the notion of a โSaaS apocalypse,โ where AI might render traditional enterprise software obsolete. McDermott, however, sees this as a fundamental misunderstanding of how enterprise platforms function.
He draws a crucial distinction: โAI thinks, but workflow acts.โ While an LLM can brilliantly suggest steps to resolve a compensation issue, it doesnโt actually close the case. That requires a robust workflow platform like ServiceNow, which can navigate complex data, context, and departmental handoffs (HR, finance, legal, compliance) to remediate and complete the task.
McDermott presents a compelling economic argument against trying to replicate enterprise platforms with language models:
- Cost: โThe cost to replace an enterprise platformโฆ is an extraordinary expense.โ His teamโs analysis shows that for a simple application, it would be โ10 times greater in cost to try to replicate it with a language model.โ This factors in not just the rebuilding effort, but also the human capital diverted from other tasks, and the substantial costs associated with GPU factories and AI tokens.
- Reliability: โPeople that run businesses understand that people make mistakes. They never will forgive software for making a mistake.โ While LLMs might be โprobably rightโ most of the time, they lack determinism and the deep context of decades of company data and relationships. Enterprise platforms, by contrast, are built for precision and accountability.
Therefore, for a fraction of an IT budget, โa premier platform is a bargain.โ McDermott envisions ServiceNow as the โAI control tower for business reinvention,โ integrating with all hyperscalers, language models, and systems of record. This approach, he argues, is akin to how hyperscalers didnโt โeat softwareโ but became crucial partners. ServiceNow acts as the โfabric that connects all those nodes,โ enabling a truly โagentic business.โ
This expansive vision includes a strategic move into security, recognizing cybercrime as the worldโs third-largest economy. By integrating security platforms and managing operational technology (OT) with recent acquisitions like Armis, ServiceNow aims to provide a โfull purview of the whole landscape,โ from people and places to critical infrastructure and even disconnected manufacturing tools. This comprehensive, native AI platform, with its ability to rapidly integrate acquisitions (like Move Works, VZA, and Armis in just 20 days), creates a formidable โmoatโ against potential disruption.
The Future of Work: Leaner, Smarter, More Human
McDermott is candid about AIโs impact on the workforce, particularly regarding headcount. He predicts that โthe net new added headcount will be dramatically reducedโ as companies become โfar more productive.โ Agents will shoulder a โtremendous workload,โ handling tasks like 90% of ServiceNowโs own customer service cases.
This shift, however, isnโt about eliminating humans but about elevating their roles. The focus will move from tactical, detail-oriented work to โcritical thinking and judgment calls.โ Humans will be freed to engineer great innovations, manage relationships, build trust, and uphold โthe net present value of loyalty.โ McDermott believes this will make people โhappier in the workโ but also necessitates a โdifferentiating your skill setโ to avoid tasks easily replicated by agents.
For companies like ServiceNow, growing at 20% with over $13 billion in trailing revenue, AI isnโt just about efficiency; itโs about expanding boundaries. McDermott looks for signals like the adoption of the โAI control towerโ vision, the growth of โagentic front doors,โ and the consumption of AI-driven โassistsโ that help humans in the companies they serve. This consumption model, he notes, โtakes on a whole new meaningโ for shareholder value creation.
Defining a Platform Business
McDermott, having led two distinct platform businesses (SAP and ServiceNow), offers a clear definition. A platform business is characterized by:
- Mission Criticality: Running the core financial systems of global corporations (like SAP) or orchestrating all other systems of record (like ServiceNow).
- Investment Scale and Switching Costs: The sheer complexity and customization involved make these platforms incredibly difficult and expensive to replace.
- Integration and Interoperability: A platform acts as a central nervous system, integrating with hundreds of other substantial systems of record, hyperscalers, and language models.
- Real-time Visibility and Agility: ServiceNowโs in-memory, open-source data platform allows for instantaneous business process changes, enabling companies to build net new applications, visualize impacts in a knowledge graph, and achieve quick wins.
- Rapid Implementation: Unlike the multi-year projects of the past, AI-powered autonomous platforms allow major customers to go live in less than 30 days, dramatically improving ROI and accelerating innovation.
This agility, coupled with the ability to handle trillions of transactions and 85 billion workflows in flight, is what makes these platforms not just sticky but indispensable.
A Vision of Human-Augmented Enterprise
McDermott observes that while many companies are still in the โexperiment phaseโ with AI (only 11% in Brazil have moved beyond it), the early movers are aggressively adopting AI to fundamentally rethink their business models, manage headcount, and achieve โmore done with a lot less.โ
His vision for the future is one where AI elevates human potential, making companies leaner, smarter, and more profitable, with expanding margins and reaccelerating revenue. Itโs a future where the human element โ connection, trust, innovation, and leadership โ remains paramount, but is powerfully augmented by intelligent agents and robust, adaptable platforms. The โSaaS apocalypseโ is not coming, McDermott asserts. Instead, we are entering an exciting era of human-augmented enterprise, where the lessons learned from a small deli continue to resonate in the global boardrooms of the AI age.
Based on โWhat to Know About the Israel-Lebanon Cease-Fire, and R.F.K. Jr.โs Shifting Tone on Vaccinesโ from New York Times Podcasts Watch the original video
Navigating a World on Edge: From Fragile Ceasefires to Shifting Narratives and Unforeseen Innovation
In a world grappling with escalating conflicts, evolving public health crises, and the relentless march of technology, a recent snapshot of global and domestic affairs reveals a tapestry woven with threads of uncertainty, shifting stances, and the enduring human spirit. From a tense ceasefire in the Middle East to a surprising pivot in vaccine rhetoric, and the unexpected challenges of the AI boom, hereโs a look at the forces shaping our current landscape.
A Precarious Peace: Life Returns to Southern Lebanon Amidst Unease
The highways connecting Beirut to southern Lebanon recently became arteries of hope and trepidation as thousands of displaced families began a cautious return home. Following the announcement of a 10-day ceasefire between Israeli forces and the Iran-backed militant group Hezbollah, a fragile quiet settled over a region scarred by conflict. A New York Times reporter on the ground described the arduous journey, with traffic snaking for miles as families navigated a single, ad-hoc dirt crossing โ a makeshift solution after Israeli bombing rendered two main bridges impassable.
For many, the journey was fueled by a desperate desire to survey what remained of their homes and lands after a month and a half of intense warfare. More than 2,100 people have been killed and over a million displaced since Israelโs latest military campaign against Hezbollah began in March, according to Lebanese authorities. Despite the ceasefire, an โuneasy feelingโ pervades, as residents express profound uncertainty about its longevity and what the future holds.
The agreement, largely brokered by the U.S., is technically between Israel and Lebanon, though the Lebanese government lacks direct control over Hezbollah. While the militant group has honored past truces negotiated by Beirut, it has yet to formally accept this latest deal, stating only that its โactions will be based on how developments unfold.โ Israelโs Prime Minister Benjamin Netanyahu, for his part, has declared that Israeli troops will remain in southern Lebanon, referring to the occupied territory as an โexpanded security zoneโ and emphatically stating, โWe are not leaving.โ This declaration casts a long shadow over the nascent peace, hinting at continued Israeli presence and potential for renewed tensions.
The Unaccounted Costs of War: Threats, Evasion, and Economic Strain
Beyond the immediate ceasefire in Lebanon, the broader conflict with Iran remains a source of significant concern. Secretary of Defense Pete Hegsath recently renewed threats to target Iranโs power plants, a move that could be deemed a war crime under international law if intentionally carried out against energy infrastructure. โOur forces are maximally postured to restart combat operations should this new Iranian regime choose poorly and not agree to a deal,โ Hegsath warned, adding chillingly, โWeโre ready to go at the command of our president and at the push of a button.โ This aggressive stance comes as the Trump administration intensifies pressure on Iran, following the collapse of talks, including a naval blockade on all ships entering or leaving the country.
Domestically, the financial burden of the ongoing conflict raises questions. On Capitol Hill, the White House budget director sidestepped inquiries about the warโs price tag, refusing to even provide a general cost range and attributing the difficulty to โfluctuating on a day in day-out basis.โ This evasion is particularly striking given earlier estimates by analysts suggesting the U.S. was spending approximately $1 billion a day on its campaign against Iran.
President Trump, however, has largely dismissed the economic fallout. Speaking in Las Vegas, he brushed off rising fuel prices โ now over $4 a gallon on average for U.S. drivers, according to AAA, and contributing to more expensive airline tickets โ as โfake inflation.โ Projecting confidence in the broader U.S. economy, he highlighted the stock marketโs performance and boosted tax refunds, telling the crowd, โI just want to say, youโre welcome.โ
A Surprising Shift: RFK Jr., Vaccines, and the Measles Resurgence
In a notable turn, the Trump administration appears to be tempering its previously staunch vaccine skepticism, a stance often championed by Health Secretary Robert F. Kennedy Jr. In a recent congressional hearing, Kennedy himself made significant concessions. When pressed, he affirmed that the MMR (measles, mumps, and rubella) vaccine is โsafe for most peopleโ and agreed that โgetting the vaccine is a lot safer than getting measles.โ He even acknowledged that two children who died during a measles outbreak in Texas earlier this year might have survived if they had been vaccinated.
While not an outright endorsement, Kennedyโs comments represent a stark contrast to his long history of promoting baseless claims about vaccine dangers. This shift coincides with White House efforts to encourage Kennedy to focus on less politically charged issues ahead of the midterms, such as healthy eating, and a new appointment: Dr. Erica Schwarz, a former deputy U.S. surgeon general and public supporter of vaccines, has been named to lead the CDC.
The administrationโs change in tone comes amid a simmering measles outbreak that continues to spread across the country, with Utah emerging as a new epicenter. Health authorities there have reported nearly 600 cases, predominantly among children, leading to around 50 hospitalizations. What makes the Utah outbreak particularly concerning is its spread beyond the โinsular religious communitiesโ that characterized earlier outbreaks in Texas and South Carolina. Infectious disease doctors in Salt Lake City observe that many families of sick children are not inherently anti-vaccine but have delayed shots due to concerns about potential side effects, fueled by what one doctor called an โepidemic of misinformation.โ
The AI Boomโs Dark Side: Maineโs Standoff with Silicon Valley
As tech companies scramble for computing power to fuel the artificial intelligence revolution, a construction boom of new data centers is transforming landscapes across the U.S. from fields to farmlands and forests. But this rapid expansion is not without its critics. Maine could become the first state to ban new data centers, with lawmakers having approved an 18-month moratorium on large new computing facilities, a bill now awaiting the governorโs signature.
The concerns are multi-faceted: the immense water and electricity consumption of these centers, and their potential to drive up utility costs for local residents. This pushback in Maine, mirrored by similar restrictions under consideration in other states, sets the stage for a potential clash with the Trump administration. President Trump has threatened to sue states and withhold federal funding if they impede the AI industryโs growth.
Locally, such efforts can also spark controversy. In Maine, one data center project was pitched as a way to revitalize an abandoned mill and create jobs in an industry seen as critical to the U.S. economy. One developer, acknowledging the publicโs apprehension, stated, โPeople are afraid of AI. Iโm afraid of AI, but guess what? Itโs coming.โ This sentiment encapsulates the difficult balance between embracing technological progress and addressing its environmental and social ramifications.
The Inventor Who Put Wheels in Your Soles: Remembering Roger Adams, Creator of Heelys
In a lighter, yet equally impactful story of innovation, the world recently mourned the passing of Roger Adams, the visionary inventor who put wheels inside shoes, creating the early 2000s phenomenon known as Heelys. Adams, who grew up on skates in his parentsโ roller rink, found himself at a crossroads in his 40s โ a divorce, a disliked job as a psychologist โ when a vision struck him: shoes that could roll on demand. In a friendโs garage, he cut the heel out of a sneaker, popped in a wheel, and Heelys were born.
By 2008, his company was selling over 7 million pairs a year. The shoes became a ubiquitous cultural touchstone, transforming casual strolls into gliding adventures. As Ash Woo, of the Timesโ obituary desk, recalled, โYou would go to the mall, you would just hear the sounds of wheels just like going everywhere. You would see a kid just kind of like walking along normally and then they would suddenly kind of Transformer style like grow a pair of wheels beneath their feet and just kick off and start gliding.โ
Despite their runaway success and the millions they earned Adams, Heelys were not universally loved. ER doctors, in particular, were not fans, and at one point, the shoes even landed on a list of โworst toysโ due to the risk of head and spinal injuries associated with the sudden transition from walking to rolling. Adams eventually left the company, largely staying out of the public eye while continuing to file patents for other wearable roller devices. His invention remains a testament to the joy and unexpected risks that can accompany even the most playful innovations.
From the geopolitical tightropes walked by nations to the evolving battle for public health and the double-edged sword of technological advancement, these headlines offer a glimpse into a world in constant flux, where every decision, every development, carries profound implications.
Based on โWhy the A.I. Backlash Turned Violent in Americaโ from Hard Fork Watch the original video
The Molotov Cocktail and the Immortality Quest: Why Americaโs AI Backlash is Boiling Over
The artificial intelligence revolution, once hailed as a beacon of progress, is increasingly met with a potent mix of fear, resentment, and even violence across America. From Molotov cocktails thrown at tech mogulsโ homes to gunshots fired over proposed data centers, a palpable anti-AI sentiment is escalating, forcing a reckoning with the industryโs promises and perils. This growing backlash isnโt just about the technology itself; itโs a complex brew of economic anxiety, distrust of elites, and a profound governance vacuum.
This unsettling trend was starkly highlighted by two recent incidents. In San Francisco, a 20-year-old man allegedly hurled a Molotov cocktail at the gate of OpenAI CEO Sam Altmanโs home. The suspect, according to court documents, possessed materials detailing anti-AI views and a list of other prominent AI figures. Just days prior, an Indianapolis city councilman, Ron Gibson, and his son awoke to more than a dozen gunshots fired at their front door, accompanied by a note: โNo data centers.โ Gibson had recently supported rezoning for a new data center in his district.
These acts, though isolated, are symptoms of a deeper, widespread unease. As Kevin Roose, a tech columnist at the New York Times, and Casey Newton of Platformer discussed on their podcast Hard Fork, these incidents underscore a โgrowing trend of anti-AI radicalization and violence.โ While condemning such acts as โvery bad for societyโ and โineffective,โ the hosts delved into the multifaceted reasons behind this increasingly furious public sentiment.
The Visible Front: Data Center Resistance
Data centers, the physical infrastructure enabling the AI boom, have become the most visible targets of public opposition. These colossal, energy-hungry facilities are essential for AI companies, but they are sparking โNot In My Backyardโ (NIMBY) movements nationwide.
โData centers are kind of the most visible symbol of the AI boom,โ Roose explained, noting the blend of legitimate concerns and unfounded fears. The resistance isnโt just individual threats; itโs coalescing into political movements. Maine recently passed a temporary moratorium banning data centers larger than 20 megawatts until late 2027. Port Washington, Wisconsin, a suburb slated to host a large OpenAI-Oracle โStargateโ data center, overwhelmingly voted to require voter approval for future data center construction. Similar efforts are underway in Ohio, Missouri, Indiana, Georgia, and North Carolina. Even at the federal level, Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez have proposed a national moratorium on data center construction.
This groundswell of opposition reflects broader public sentiment, which polls suggest is increasingly negative towards AI. The 2026 Stanford AI Index reported that only 31% of Americans trust their government to responsibly regulate AI, far below the global average of 54%. While a Pew study found mixed views on AIโs economic impact, it clearly showed more negative than positive attitudes regarding data centersโ effects on the environment, energy costs, and local quality of life. The overarching takeaway: as AI systems grow more powerful, public concern intensifies.
Unpacking the Fury: Three Explanations for the Backlash
Roose and Newton explored several theories for this dramatic turn in public opinion:
1. The CEOsโ Own Warnings: A Self-Fulfilling Prophecy?
Sam Altman, in a personal blog post following the attack, pondered whether โincendiaryโ investigative journalism contributed to the danger he faced. While acknowledging the power of words, the Hard Fork hosts pushed back on the idea that scrutiny of powerful people or companies should cease. Instead, they argued that the very CEOs now pleading for calm were the ones who initially stoked fears about AIโs existential risks.
Newton cited Altmanโs own 2015 blog post, where he wrote, โDevelopment of superhuman machine intelligence is probably the greatest threat to the continued existence of humanity.โ Other AI executives have echoed similar dire predictions. โTo come along now at a moment where the systems are more powerful than everโฆ and to say, โWell, now we need to tamp down the rhetoricโ โ that just seems sort of crazy to me,โ Newton remarked. The rhetoric, they argued, has now become entwined with the material effects of the technology itself.
Roose acknowledged the bind tech leaders face: having witnessed social media CEOs promise utopia only to deliver widespread harm, current AI leaders might feel compelled to be upfront about risks. But this honesty, while seemingly virtuous, has inadvertently amplified public anxiety.
2. The Specter of Economic Disruption
Perhaps the most potent driver of public fear is economic anxiety. โAll AI politicsโฆ ultimately come down to people just sort of looking at a technology and thinking, โWhat will this do to me and my ability to continue to live my life and support my family?โโ Roose articulated. The concrete, near-term worry is job displacement or the degradation of existing jobs.
This fear clashes sharply with the tech industryโs โfully automated luxury communismโ vision โ a future where AI handles all labor, and humans enjoy a life of leisure, with 3D-printed meals costing nothing. Such a utopian ideal, however, โseems so implausible to people that itโs impossible to build any kind of political constituency around it,โ Newton observed.
The cultural disconnect is profound. While Silicon Valley champions rapid technological change, most people crave stability. โWhen people from Silicon Valley show up and say, โHey, weโve got this amazing new technology,โ and by the way, it might take away your job and thereโs nothing you can do about it, I think that naturally breeds some fear and resentment,โ Roose concluded.
3. An Elitist, Anti-Democratic Project
The third explanation centers on anti-elitism. AI, as it currently exists, is a โtop-down moment,โ not a grassroots phenomenon. Itโs driven by a small group of unelected, largely unaccountable individuals with access to immense capital. โNot only did I not ask for this, but I have no meaningful control over it,โ is the average personโs sentiment, according to Newton.
This perception is compounded by the industryโs active resistance to accountability. OpenAI, for instance, lobbied against one of Californiaโs first major AI transparency bills, successfully killing it. Theyโve sent subpoenas to nonprofit workers advocating for AI regulations and backed an Illinois bill that would shield them from liability in cases where their models cause serious harm, provided it wasnโt reckless or intentional.
โThis technology is too important to be left to the masses,โ is an attitude Roose has encountered in Silicon Valley, which he deems โmisguided.โ Newton agrees, arguing that fighting against accountability and transparency mechanisms fuels public fury. Even OpenAIโs recent โindustrial policy for the intelligence ageโ white paper, which proposes radical ideas like a public wealth fund to distribute AIโs economic upside and improved social safety nets, is met with skepticism. Newton pointed out the hypocrisy of such proposals when OpenAI simultaneously backs Republican candidates unlikely to support a massive expansion of the welfare state. โSomething is going on here that I think at the very least leaves room for critics to say, โAre yโall even serious about this?โโ
The Wrong Levers and the Governance Problem
The hosts lamented that public opposition often manifests through โthe wrong levers.โ While voting out a data center project might feel empowering, itโs unlikely to halt AIโs progress. โIf you vote the data center project out of your town, theyโre just going to go to another state or to Canada, theyโll put the data centers in space,โ Roose quipped. This NIMBYism, while locally effective, doesnโt address the core issues and could lead to unforeseen consequences, much like how environmental reviews intended to block construction in the 1980s contributed to todayโs housing shortages.
The โright levers,โ they argue, involve proactive policy: flexible social safety nets, retraining programs for displaced workers (like those seen in some European countries after manufacturing automation), and robust regulation. However, this points to a fundamental โgovernance problem.โ The government, often slow to act and susceptible to lobbying, struggles to navigate rapidly evolving technologies.
โWe have a governance problem, and that while the problem is being driven by the decisions of these like unelected AI leaders, it is ultimately the governments who are going to have to give us an answer to these questions,โ Newton asserted.
Silicon Valleyโs Quest for Immortality: A Parallel Critique
Adding another layer to the critique of techโs elite, the discussion shifted to Cara Swisherโs new CNN documentary, โCara Swisher Wants to Live Forever,โ which explores Silicon Valleyโs obsession with longevity and โbody hacking.โ Swisher, a veteran tech journalist, brought her characteristic skepticism to the topic.
Swisherโs journey began with Steve Jobsโ famous Stanford speech on mortality, but she observed a shift in tech culture from accepting death as a motivator to actively trying to circumvent it. Investments from figures like Sam Altman, Larry Ellison, and Peter Thiel poured into anti-aging research, alongside fads like intermittent fasting, Soylent, psychedelics for โoptimization,โ and even Elon Muskโs pronouncements on health.
In her documentary, Swisher subjected herself to various longevity trends: sound therapy, hyperbaric chambers, red light therapy, sleep therapy, and even ketamine. While some, like improving her VO2 max through running tests, proved genuinely helpful, others were dismissed as โridiculous.โ Of the hyperbaric chamber, she declared, โItโs so stupid to have all these people insist that this is the way to goโฆ itโs just one of these things they sell to rich people and make them feel superior and itโs a waste of money.โ Her ketamine experience, which some tech leaders use for โoptimizationโ and โnew ideas,โ left her in a โk-holeโ and ultimately, โbored.โ
Swisherโs core critique echoed the broader anti-elitism sentiment: โWhat really does me is the one of the simplest things for all of us to live longer is universal healthcareโฆ weโre at the bottom of all the outcomes and thatโs offensive to me that these people are spending all this money on all manner of nonsensical treatments.โ She sees it as narcissistic, individualistic pursuit of โperfectabilityโ by the wealthy, detached from the broader health of society.
While acknowledging the โtrickle-down healthโ argument โ that wealthy โguinea pigsโ might pave the way for treatments like GLP1s (Ozempic/Wegovy) to become widely available โ Swisher pointed out that such innovations often start with existing medical needs (like diabetes) before being adopted by the affluent.
The Path Forward: A Call for Political Leadership
The AI backlash, whether expressed through Molotov cocktails or longevity quests, reflects a profound crisis of trust and control. The tech industryโs promises of a utopian future ring hollow against the backdrop of economic anxiety, perceived elitism, and a lack of accountability.
The solution, Roose and Newton concluded, is not a messaging tweak but a fundamental shift in governance. Thereโs a wide-open lane in American politics for leaders who can articulate a clear plan: not to stop AI, but to harness it for the greater good, ensuring its benefits are widely shared and its harms mitigated. This means robust social safety nets, retraining programs, and regulations that prioritize public well-being over unchecked innovation or corporate liability shields.
โIf you can figure that out, they will write your name in history,โ Newton declared, issuing a challenge to future politicians. The stakes are immense: without a coherent, proactive approach, the current AI backlash, fueled by legitimate fears and exacerbated by a governance vacuum, risks spiraling further out of control, jeopardizing both technological progress and societal stability.
ํ๊ตญ์ด
โHustle culture stole the word excellence and gutted its true meaning | Brad Stulberg: Full Interviewโ โ Big Think ๊ธฐ๋ฐ ๊ธฐ์ฌ ์๋ณธ ์์ ๋ณด๊ธฐ
ํ์ฌ ๋ฌธํ์ ๋นผ์๊ธด โํ์ํจโ์ ์ง์ง ์๋ฏธ: ๋ธ๋๋ ์คํธ๋ฒ๊ทธ ๊ต์๊ฐ ๋งํ๋ ์ถ์ ๊น์ ๋ง์กฑ
์ค๋๋ ์ฐ๋ฆฌ๋ โํ์ํจ(excellence)โ์ด๋ผ๋ ๋จ์ด๋ฅผ ๋๋ฌด๋ ํํ๊ฒ ์ฌ์ฉํฉ๋๋ค. ์๋ฒฝ 4์์ ์ผ์ด๋ 47๋จ๊ณ ๋ฃจํด์ ์ํํ๊ณ , ๋์ ๋ชฉ์์ ํ๋ฉฐ 9๊ฐ์ง ์์์ ๋ฅผ ์ฑ๊ฒจ ๋จน๊ณ , ์๋ฉด๊ณผ ์ฌ๋ฐ ๋ณ์ด๋๋ฅผ ์ถ์ ํ๋ฉฐ ์์ ์ ๋๋จํจ์ ๋ง์ธ์๊ฒ ์๋ํ๋ ๋ชจ์ต. ์ด๊ฒ์ด ๊ณผ์ฐ ์ง์ ํ ํ์ํจ์ผ๊น์? ๋ฏธ์๊ฐ ๋ํ๊ต ๊ต์์ด์ ๋ฒ ์คํธ์ ๋ฌ ใํ์ํจ์ ๊ธธ(The Way of Excellence)ใ์ ์ ์์ธ ๋ธ๋๋ ์คํธ๋ฒ๊ทธ(Brad Stulberg)๋ ์ด๋ฌํ ๋ชจ์ต์ด โ์ง์ ํ ๊ฒ์ฒ๋ผ ์์ฅํ ์ ๊ตํ ๊ฐ๋ถํค(elaborate kabuki)โ์ ๋ถ๊ณผํ๋ค๊ณ ์ผ๊ฐํฉ๋๋ค.
๋น ์ฝํฌ(Big Think)์์ ์ธํฐ๋ทฐ์์ ์คํธ๋ฒ๊ทธ ๊ต์๋ ํ๋ ์ฌํ๊ฐ ์์ด๋ฒ๋ฆฐ โํ์ํจโ์ ์ฐธ๋ ์๋ฏธ๋ฅผ ๋์ฐพ๊ณ , ๊ทธ ๊ธธ์ ๊ฐ๋ก๋ง๋ ํํ ํจ์ ๋ค์ ๋ฐํ๋ฉฐ, ์ง์ ์ผ๋ก ์ถฉ๋งํ๊ณ ๋ง์กฑ์ค๋ฌ์ด ์ถ์ ์ํ ํ์ ์์๋ค์ ์ ์ํ์ต๋๋ค.
์ง์ ํ ํ์ํจ์ด๋ ๋ฌด์์ธ๊ฐ?
์คํธ๋ฒ๊ทธ ๊ต์๊ฐ ์ ์ํ๋ ํ์ํจ์ ๋๋๋๋ก ๋จ์ํฉ๋๋ค. โ์์ ์ ๊ฐ์น์ ์ผ์นํ๋ ๊ฐ์น ์๋ ์ผ์ ๋ชฐ์ ํ๋ ์ฐธ์ฌ(Involved engagement in something worthwhile that aligns with your values).โ ์ฆ, ๋จ์ํ ๋ฌด์ธ๊ฐ๋ฅผ ์ฑ์ทจํ๋ ๊ฒ๋ง์ด ์๋๋ผ, ๊ทธ ๊ณผ์ ์์ ์ด๋ค ์ฌ๋์ด ๋๋์ง์ ๋ํ ์ด์ผ๊ธฐ์ ๋๋ค. ์ฐ ์ ์์ ์ค๋ฅด๋ ๊ฒ ์์ฒด๊ฐ ์๋๋ผ, ์ฐ ์คํฑ์์ ๊ฒช๋ ๊ฒฝํ๊ณผ ์ฑ์ฅ์ ํตํด ๋น์ ์ด ์ด๋ค ์ธ๊ฒฉ์ฒด๋ก ๋ณํํ๋์ง๊ฐ ์ค์ํ๋ค๋ ๊ฒ์ด์ฃ .
์๋ฅผ ๋ค์ด, ์๋ฌด๋ฐ ๊ฐ์น๋ ์๋ ํฐ ๋ฒฝ์ ๋๋๋ฆฌ๋ ๋ฐ ์ด์ ์ ์ผ๋ก ๋ชฐ์ ํ ์๋ ์๊ฒ ์ง๋ง, ์ด๋ ํ์ํจ์ด ๋ ์ ์์ต๋๋ค. ๋ฐ๋ฉด, ์ง์ฑ, ๊ธฐ์ฌ, ํ์ธ ๋๊ธฐ ๋ฑ์ ๊ฐ์น๋ฅผ ์์คํ ์ฌ๊ฒจ ํ๋ฅญํ ์์ฌ๊ฐ ๋๊ธฐ ์ํด ๋ ธ๋ ฅํ๋ ๊ฒ์ ํ์ํจ์ ๋๋ค. ๋ง๋ผํค์ ์์ฃผํ๊ณ , ํ์ฌ๋ฅผ ์ค๋ฆฝํ๊ณ , ์๊ณผ๋ํ์ ์กธ์ ํ๊ณ , ์ฒซ ์ฑ ์ ์ถํํ๋ ๊ณผ์ ์์ ์ฐ๋ฆฌ๋ ๊ทธ ๋ชฉํ๋ฅผ ์ํด โ์ผํ์ง๋งโ, ๋์์ ๊ทธ ๋ชฉํ ๋ํ โ์ฐ๋ฆฌ๋ฅผ ๋ณํ์ํต๋๋คโ. ์ฐ๋ฆฌ๋ ์ด๋ ค์์ ๊ทน๋ณตํ๊ณ , ํ๋ณต๋ ฅ์ ๋ฐฐ์ฐ๋ฉฐ, ๊พธ์คํจ, ํ๋ จ, ์๊ธฐ ์ฐ๋ฏผ์ ํ์ ๊นจ๋ซ์ต๋๋ค.
์ด๋ฌํ ๊ฐ์น์ ์ผ์นํ๋ ๋ชฐ์ ์ ์ฐ๋ฆฌ์๊ฒ ๋ ๊ฐ์ง ์ค์ํ ๊ฐ๊ฐ์ ์ ๋ฌผํฉ๋๋ค.
- ๋ง์คํฐ๋ฆฌ(Mastery): ์ ๋ฅํจ, ๊ตฌ์ฒด์ ์ด๊ณ ์ค์ง์ ์ธ ๋ฐ์ ์ ์ด๋ฃจ๋ ๋ง์กฑ๊ฐ.
- ์๋ฏธ(Mattering): ์์๊ฐ, ๊ทธ๋ฆฌ๊ณ ์ฐ๋ฆฌ์ ์์ ์์๋ณด๋ค ๋ ํฐ ๋ฌด์ธ๊ฐ์ ์ผ๋ถ๋ผ๋ ์ธ์.
๋๋ถ๋ถ์ ์ฌ๋๋ค์ 3~5๊ฐ์ง ํต์ฌ ๊ฐ์น๋ฅผ ๊ฐ์ง๊ณ ์์ต๋๋ค. ๋ง์คํฐ๋ฆฌ, ์ฅ์ธ์ ์ , ์ง์ค์ฑ, ๊ณต๋์ฒด, ์์ฑ, ํ๋ จ, ์ง์ฑ, ์งํ, ์น์ , ์ฐฝ์์ฑ ๋ฑ์ด ๊ทธ ์์์ ๋๋ค. ์ค์ํ ๊ฒ์ ์ด๋ฌํ ๊ฐ์น๋ค์ ๋ฒฝ์ ๊ฑธ๋ฆฐ ํ ๋น ๋ฌธ๊ตฌ๋ก ๋์ง ์๊ณ , ์์ ์๊ฒ โ๊ฑด๊ฐโ์ด ๋ฌด์์ ์๋ฏธํ๋์ง, ์ด๋ป๊ฒ ์ค์ฒํ ๊ฒ์ธ์ง ๋ฑ ๊ตฌ์ฒด์ ์ผ๋ก ์ ์ํ๊ณ , ์ผ์์ํ์์ ๊ทธ ๊ฐ์น์ ๋ถํฉํ๊ฒ ์ด๊ธฐ ์ํด ๋ฌด์์ ํ ์ ์๋์ง ๊ณ ๋ฏผํ๋ ๊ฒ์ ๋๋ค. ๋ชฉํ๋ฅผ ์ธ์ธ ๋๋ ์์ ์ ๊ฐ์น์ ์ผ์นํ๋์ง ์ ๊ฒํด์ผ ํฉ๋๋ค. ๊ฑด๊ฐ์ ์ค์ํ๋ ์ฌ๋์ด ํ๋ฃจ 22์๊ฐ์ฉ ์ผํ๋ฉฐ ์ ์ ์์ง ์๋ ๋ชฉํ๋ฅผ ์ธ์ด๋ค๋ฉด, ์ด๋ ์์ ์ ๊ฐ์น์ ๋ชจ์๋๋ ๊ฒ์ ๋๋ค.
ํ์ง(Quality)๊ณผ ํญ์์ฑ ์ํฅ ์กฐ์ (Homeostatic Upregulation)
๋ก๋ฒํธ ํผ์๊ทธ(Robert Pirsig)๋ ๊ทธ์ ์ ์ ใ์ ๊ณผ ์คํ ๋ฐ์ด ๊ด๋ฆฌ์ (Zen and the Art of Motorcycle Maintenance)ใ์์ โํ์ง(Quality, ๋๋ฌธ์ Q)โ์ด๋ผ๋ ๊ฐ๋ ์ ์ ์ํ์ต๋๋ค. ํผ์๊ทธ์ ๋ฐ๋ฅด๋ฉด ํ์ง์ ํ์์์ ํ์ ์ฌ์ด์ ๊ด๊ณ๊ฐ ๋๋ฌด๋ ๋ฐ์ ํ์ฌ ์ฃผ์ฒด์ ๊ฐ์ฒด๊ฐ ์ฌ๋ผ์ง๋ ์๊ฐ์ ์๋ฏธํฉ๋๋ค. ๋ง์น ๋๊ตฌ๋ฅผ ํ๋ ๋ ์ด์ฒผ์ด ์๋๋ผ, ๋๊ตฌ ๊ทธ ์์ฒด๊ฐ ์ผ์ด๋๊ณ , ๊ธฐํ๋ฅผ ์น๋ ์กฐ๊ฐ ์๋๋ผ, ์์ ์ด ๋ง๋ค์ด์ง๋ ๊ฒ๊ณผ ๊ฐ์ต๋๋ค. ํผ์๊ทธ๋ ์ด๋ฌํ ํ์ง์ด ์งํ์ ์๋๋ ฅ์ด๋ผ๊ณ ์ฃผ์ฅํ์ต๋๋ค.
์คํธ๋ฒ๊ทธ ๊ต์๋ ํผ์๊ทธ์ ํ์ง ๊ฐ๋ ์ด ํ๋ ์งํ ์๋ฌผํ์ โํญ์์ฑ ์ํฅ ์กฐ์ (Homeostatic Upregulation)โ๊ณผ ๋งค์ฐ ์ ์ฌํ๋ค๊ณ ์ค๋ช ํฉ๋๋ค. ์ด๋ ๋ชจ๋ ์๋ช ์ฒด๊ฐ ์์กดํ๊ณ ๋ฒ์ฑํ๋ ค๋ ๋ด์ฌ์ ์ด๊ณ ๊น์ด ๊ฐ์ธ๋ ์ถฉ๋์ ๋๋ค. ๋๋ ์ ๊ฒฝ๊ณ๊ฐ ์๋ ๋ฐํ ๋ฆฌ์์กฐ์ฐจ๋ ์์กด์ ์ ๋ฆฌํ ํ๊ฒฝ์ผ๋ก ์ด๋ํ๊ณ ๋ถ๋ฆฌํ ํ๊ฒฝ์์ ๋ฉ์ด์ง๋ ค๋ ๋ณธ๋ฅ์ ์ธ ์์ง์์ ๋ณด์ ๋๋ค. ์ด๋ฌํ ์์ด์ ์ธ ์ถฉ๋์ ๋ณต์กํ ์ ๊ธฐ์ฒด๋ก ๋ฐ์ ํ๋ฉด์ ์์กด๊ณผ ๋ฒ์์ ๋์ด, ์๋ํ ์์ ์ , ์ด๋์ ์ ์ ์ ์ถ๊ตฌํ๊ณ , ๋ ๋์ ๋ถ๋ชจ๋ ๋ฆฌ๋๊ฐ ๋๋ ค๋ ์ธ๊ฐ์ ์ด๋ง์ผ๋ก ํ์ฅ๋ฉ๋๋ค. ์ฐ๋ฆฌ๋ ๋ชจ๋ ๋ฒ์ฑํ๊ณ , ์ฑ์ฅํ๋ฉฐ, ํ์ํจ์ ํฅํด ๋์๊ฐ๋ ค๋ ๋ด์ฌ๋ ์๊ตฌ๋ฅผ ๊ฐ์ง๊ณ ์์ต๋๋ค.
์ํฉ ์ธ์ง(Situated Cognition)์ ๋ฌด์์์ ์ ๋ฅ(Unconscious Competence)
์ฌํํ์ ๋ฆฌ์ฒ๋ ์ธ๋ท(Richard Sennett)์ด ์ฅ์ธ๋ค์ ์ฐ๊ตฌํ๋ฉฐ ์ ์ํ โ์ํฉ ์ธ์ง(Situated Cognition)โ๋ ์ฅ์ธ๋ค์ด ์ต๊ณ ์ ๋ชฐ์ ๊ณผ ๋ง์กฑ๊ฐ์ ๋๋ ๋ ๋จธ๋ฆฌ๋ก ์๊ฐํ๋ ๊ฒ์ด ์๋๋ผ, ์จ ์กด์ฌ๋ก ์๊ฐํ๋ ์ํ๋ฅผ ์๋ฏธํฉ๋๋ค. ๊ทธ๋ค์ ํ ์ด๋ธ์ ๋ง๋ค๊ฑฐ๋ ์กฐ๊ฐํ์ ๋ง๋ค ๋ ๋ชธ๊ณผ ๋ง์์ด ํ๋ ๋๋ ๊ด๊ณ๋ฅผ ๊ฒฝํํฉ๋๋ค. ํ๋์ ๋ชฐ์ ํ๋ฉด์ ์์ ์์์ ํธ์ํจ์ ๋๋ผ๊ณ ์ธ์๊ณผ ์ฐ๊ฒฐ๋์ด ์๋ค๊ณ ๋๋ผ๋ ๊ฒ์ ๋๋ค.
์ด๋ ๋ค ๊ฐ์ง ์ญ๋ ๋จ๊ณ ์ค ๊ฐ์ฅ ๋์ ๋จ๊ณ์ธ โ๋ฌด์์์ ์ ๋ฅ(Unconscious Competence)โ๊ณผ ๋ฐ์ ํ๊ฒ ์ฐ๊ฒฐ๋ฉ๋๋ค.
- ๋ฌด์์์ ๋ฌด๋ฅ(Unconscious Incompetence): ์์ ์ด ๋ฌด์์ ๋ชจ๋ฅด๋์ง์กฐ์ฐจ ๋ชจ๋ฅด๋ ๋จ๊ณ์ ๋๋ค. ์ด ๋จ๊ณ์์๋ ์ฑ ์ ์ฝ๊ณ , ์์์ ๋ณด๊ณ , ์ฝ์น๋ฅผ ๊ณ ์ฉํ๋ ๋ฑ ์ธ๋ถ์ ๋์์ด ํ์ํฉ๋๋ค.
- ์์์ ๋ฌด๋ฅ(Conscious Incompetence): ์์ ์ด ๋ฌด์์ ๋ชจ๋ฅด๋์ง ์๋ ๋จ๊ณ์ ๋๋ค. ์ฌ๋ฐ๋ฅธ ๋ฐฉ์์ด ๋ฌด์์ธ์ง๋ ์์ง๋ง ์์ง ์์ ์ ๊ทธ๋ ๊ฒ ํ์ง ๋ชปํ๋ค๋ ๊ฒ์ ์ธ์งํฉ๋๋ค.
- ์์์ ์ ๋ฅ(Conscious Competence): ์ฌ๋ฐ๋ฅธ ๋ฐฉ์์ผ๋ก ํ๊ณ ์๋ค๋ ๊ฒ์ ์๊ณ , ์์์ ์ผ๋ก ๋ ธ๋ ฅํ๋ ๋จ๊ณ์ ๋๋ค. ์ฒดํฌ๋ฆฌ์คํธ๋ฅผ ๋ฐ๋ฅด๊ฑฐ๋, ์ฌ๋ฐ์ ๋ชจ๋ํฐ๋ฅผ ์ฌ์ฉํ๋ ๋ฑ ๋ฅ์ํ ์ฑ๊ณผ๋ฅผ ์ํด ์ฌ๊ณ ๋ฅผ ํ์ฉํฉ๋๋ค.
- ๋ฌด์์์ ์ ๋ฅ(Unconscious Competence): ๋ชจ๋ ์์์ ์ธ ๋ ธ๋ ฅ๊ณผ ์ถ์ , ์ธก์ ์ ๋ค๋กํ๊ณ โ์กด(zone)โ์ ์ง์ ํ์ฌ ๊ฐ๊ฐ์ด ์ต๊ณ ์ ์ฑ๊ณผ๋ฅผ ์ด๋๋ ๋จ๊ณ์ ๋๋ค.
์ง์ ํ ์ฅ์ธ๋ค์ ์ด๋ฌํ ๋ค ๋จ๊ณ๋ฅผ ์์ธก ๊ฐ๋ฅํ๊ฒ ๊ฑฐ์น์ง๋ง, ๋ํ ์ด ๋จ๊ณ๋ค์ ์ํํฉ๋๋ค. ๊ทธ๋ค์ ์์์ ์ ๋ฅ ๋จ๊ณ์์ ๋ ธ๋ ฅํ๊ณ , ๋ค์ ๋ฐฐ์ฐ๊ณ , ์ฌ์ ๋นํด์ผ ํ ์์ ์ ์ดํดํฉ๋๋ค. ๊ทธ๋ฆฌ๊ณ ๋ง๋ฒ ๊ฐ์ ๋ ์ด ์ฐพ์์ค๋ฉด ๊ทธ ์๊ฐ์ ๋ด๋ ค๋๊ณ โ๋ง๋ฒโ์ ์ ๋ขฐํ๋ฉฐ ๊ทธ ํ๋ฆ์ ๋ชธ์ ๋งก๊ธธ ์ฉ๊ธฐ๋ฅผ ๊ฐ์ง๋๋ค.
ํ์ํจ์ ๋์์์ด ์ฃผ์๊ฐ ์ฐ๋งํด์ง๊ณ ์์ ์ ํ๋, ์ฌ์ง์ด ์ถ ์์ฒด๋ก๋ถํฐ ์์ธ๊ฐ์ ๋๋ผ๋ ์ธ์์์ โ์น๋ฐ๊ฐโ์ ์์ฒ์ด ๋ฉ๋๋ค. ์ด๋ ์ฅ์ธ์ ์ ๊ณผ์ ์น๋ฐ๊ฐ, ๊ทธ๋ฆฌ๊ณ ์๊ธฐ ์์ ๊ณผ์ ์น๋ฐ๊ฐ์ ์๋ฏธํฉ๋๋ค. ํ์ํจ์ ์์ธ๊ฐ์ ๋ฐ๋์ด๋ฉฐ, ๊ธฐ๊ณ์ ์ผ๋ก ์์ง์ด๋ ๊ฒ์ ๋ฐ๋์ ๋๋ค.
ํ์ํจ์ ๊ธธ์ ๊ฐ๋ก๋ง๋ ์ฅ์ ๋ฌผ๋ค
์คํธ๋ฒ๊ทธ ๊ต์๋ ์ง์ ํ ํ์ํจ์ ์ถ๊ตฌํ๋ ๊ธธ์ ๋์ธ ์ฌ์ฏ ๊ฐ์ง ์ฃผ์ ์ฅ์ ๋ฌผ์ ์ง์ ํฉ๋๋ค.
-
ํดํ(Disevolution): ํ๋ ํ๊ฒฝ์ ํจ์ ์งํ ๊ณผํ์ ๋ค๋์ ๋ฆฌ๋ฒ๋ง(Daniel Lieberman)์ด ๋ง๋ โํดํ(Disevolution)โ๋ผ๋ ์ฉ์ด๋ ์ฐ๋ฆฌ๊ฐ ์งํ ๊ณผ์ ์์ ๊ฒช์ง ๋ชปํ๋ ํ๋ ํ๊ฒฝ์ ํจ์ ๋ค์ ์ค๋ช ํฉ๋๋ค.
- ํจ์คํธํธ๋: ์ค๋ซ๋์ ๋ถ์กฑํจ ์์์ ์ด์๋ ์ธ๋ฅ๋ ๋์์ ๋๋์ ๋ชจ๋ ๋จน์ด์น์ฐ๋๋ก ์งํํ์ต๋๋ค. ํ์์ ์๋์๋ ๋๋์ ๊ฐ๋งํ๋ ๋ณธ๋ฅ์ ์ฌ๋ผ์ง์ง ์์ ๋น๋ง ๋ฌธ์ ๋ฅผ ์ผ๊ธฐํฉ๋๋ค.
- ์์ ๋ฏธ๋์ด vs. ์ง์ ํ ๊ณต๋์ฒด: ์ฐ๋ฆฌ๋ ์ฌํ์ ์กด์ฌ๋ก์ ๊ณต๋์ฒด์ ์ฐ๊ฒฐ์ ๊ฐ๋งํฉ๋๋ค. ํ์ง๋ง ์์ ๋ฏธ๋์ด๋ 10์ด์ง๋ฆฌ ์ธ๋ถ ๊ฒ์ฆ(์ข์์, ๋๊ธ, ๋ฆฌํธ์)์ผ๋ก ์ด๋ฌํ ์๊ตฌ๋ฅผ ์ถฉ์กฑ์ํค๋ ค ํฉ๋๋ค. ์ด๋ ์ผ์์ ์ธ ๋ง์กฑ๊ฐ์ ์ฃผ์ง๋ง, ์ค์ ์ธ๊ฐ ๊ด๊ณ๊ฐ ์ฃผ๋ ์ถฉ๋งํจ๊ณผ๋ ๋ค๋ฆ ๋๋ค.
- ์๋์ ๊ด์ฐฐ์: ๋ค๋ฅธ ์ฌ๋๋ค์ด ๋ฉ์ง ์ผ์ ํ๋ ๊ฒ์ ์๋์ ์ผ๋ก ๊ด์ฐฐํ๋ฉฐ ๋ง์กฑ๊ฐ์ ๋๋ผ๋ ๊ฒ์ ์ค์ ๊ฒฝ๊ธฐ์ฅ์ ๋ฐ์ด๋ค์ด ์์ ์ ๋๋ฌ๋ด๋ ๊ฒ๊ณผ๋ ๋ค๋ฆ ๋๋ค.
๊ฒฐ๋ก ์ ์ผ๋ก, ๋จ๊ธฐ์ ์ผ๋ก๋ ๊ธฐ๋ถ ์ข์ง๋ง ์ฅ๊ธฐ์ ์ผ๋ก๋ ์ฐ๋ฆฌ์ ๊ฐ์น์ ๋ง์ง ์๊ณ ๊ฒฐ๊ตญ ๋ ๋์ ๊ธฐ๋ถ์ ๋จ๊ธฐ๋ ๊ฒ๋ค์ ์ํด ์ฐ๋ฆฌ์ โ๊ธฐ๋ถ ์ข์โ์ ํฅํ ๋ณธ๋ฅ์ด ์ค์ฉ๋ ๋ ๊ฐ์ฅ ํฐ ํจ์ ์ ๋น ์ง๊ฒ ๋ฉ๋๋ค.
-
์ต์ ํ ๋ฌธํ(Optimization Culture): ๋ก๋ด์ด ๋๋ ์ํ ์ต์ ํ ๋ฌธํ๋ ์ถ์ ๋ชจ๋ ๊ฒ์ ํต์ ํ๋ ค๋ ๊ฒฝํฅ์ ๋งํฉ๋๋ค. ์๋ฉด ์๊ฐ๋ฟ๋ง ์๋๋ผ ๋ ์๋ฉด ์๊ฐ๊น์ง ์ถ์ ํ๊ณ , ๋ชจ๋ ์นผ๋ก๋ฆฌ๋ฅผ ๊ณ์ฐํ๋ฉฐ, ์ฌ๋ฐ ๊ฐ๊ฒฉ๊น์ง ์ธก์ ํ๋ ๋ฑ ์ถ์ ๋ชจ๋ ๊ฒ์ ์ถ์ ํฉ๋๋ค. ์คํธ๋ฒ๊ทธ ๊ต์๋ ์ด๋ฌํ ๊ณผ๋ํ ์ต์ ํ๊ฐ ์ฐ๋ฆฌ๋ฅผ ๋ก๋ด์ผ๋ก ๋ง๋ค ์ํ์ด ์๋ค๊ณ ๊ฒฝ๊ณ ํฉ๋๋ค. ํ์ํจ์ ๊น์ด ์ธ๊ฐ์ ์ธ, โ๋๊ปด์ง๋ ๊ฒฝํโ์ธ๋ฐ, ์ถ์ ๊ณผ๋ํ๊ฒ ์ต์ ํํ๋ฉด ์ด๋ฌํ ๊ฒฝํ์ ํ ๊ธฐํ๋ฅผ ๋์น๊ฒ ๋ฉ๋๋ค.
๋ฌผ๋ก ์๋ฉด ์ถ์ ๊ธฐ๋ ๋ง๋ณด๊ณ ๊ฐ์ ์ฅ์น๋ค์ด ํ๋ ๋ณํ์ ๋์์ด ๋ ์ ์์ต๋๋ค. ๊ทธ๋ฌ๋ ์ด๋ฌํ ์ฅ์น๋ค์ด ์ฐ๋ฆฌ๋ฅผ ํต์ ํ๊ฒ ๋ด๋ฒ๋ ค ๋๊ณ , ์ถ ์ ์ฒด๋ฅผ ์ต์ ํ์ ๋ง์ถฐ ๊ตฌ์ถํ๋ฉด ์ถ์ ๊ฐ์ , ์ธ๊ฐ๋ฏธ, ๊ธฐ์จ์ ์์๊ฐ ์ต๊ณ ์ ์ฑ๊ณผ๋ฅผ ๋ด๊ธฐ ์ด๋ ต๊ฒ ๋ง๋ญ๋๋ค.
๊ณจํผ JJ ์คํ(JJ Spahn)์ ์ด์ผ๊ธฐ: ์ต๊ทผ US ์คํ ์ฐ์น์ JJ ์คํ์ ๊ฒฝ๊ธฐ ์ ๋ ์๋ฒฝ 2์์ ์์ด๊ฐ ์ํ ๋ฐค์ ๋๋ดค์ต๋๋ค. ํ๋ณต ์ฅ์น๋ฅผ ์ฐฉ์ฉํ๊ณ ์์๋ค๋ฉด โํ๋ณต ์ ์ 0, ์ ๋ชป ์ค์ผ๋ ์ง์ ์์ด๋ผโ๊ณ ์กฐ์ธํ์ ๊ฒ์ ๋๋ค. ํ์ง๋ง ์คํ์ ๋ค์ ๋ US ์คํ์์ ์ฐ์นํ์ต๋๋ค. ๊ณผ๋ํ ์ต์ ํ๋ ์ฐ๋ฆฌ๋ฅผ ์ฐ์ฝํ๊ฒ ๋ง๋ญ๋๋ค. ๋ชจ๋ ๊ฒ์ด ์๋ฒฝํด์ผ๋ง ์ ํด๋ผ ์ ์๋ค๊ณ ์ฐฉ๊ฐํ๊ฒ ๋ง๋ค์ง๋ง, ์ด๋ ํฐ๋ฌด๋์๋ ์๊ฐ์ ๋๋ค.
-
์ ์ง ๋ชฐ์ (Shitty Flow): ๊ฐ์น์ ๋จ์ ๋ ๋ชฐ์ ์ฌ๋ฆฌํ์ ๋ฏธํ์ด ์น์ผํธ๋ฏธํ์ด(Mihaly Csikszentmihalyi)๊ฐ ๋ง๋ โ๋ชฐ์ (Flow)โ์ ์์ ์ด ํ๋ ์ผ์ ์์ ํ ๋ชฐ๋ํ์ฌ ์๊ฐ๊ณผ ๊ณต๊ฐ์ ์ธ์ง๊ฐ ์๊ณก๋๋ ๊ฒฝํ์ ๋งํฉ๋๋ค. ์ ๊ธฐ ์ฐ์ฃผ, ํ ์ฝ์นญ, ์ํ ๋ฌธ์ ํ์ด ๋ฑ ์ง์ ํ ๋ชฐ์ ์ํ๋ ๋๋๊ณ ์ถ์ ์๋ฏธ๋ฅผ ๋ถ์ฌํ ์ ์์ต๋๋ค.
ํ์ง๋ง ํธ์ํฐ ์คํฌ๋กค๋ง์ด๋ ๋๋ฐ์์๋ ๋ชฐ์ ์ํ์ ๋น ์ง ์ ์์ต๋๋ค. ์ด๋ฌํ ๊ฒฝํ๋ค์ ํ๋ ฅ๊ฐ์ ์ฃผ์ง๋ง, โ๊ฐ์น์ ์ฐ๊ฒฐ๋์ง ์์ ์์ฌ ํ๋ ฅ(pseudo aliveness)โ์ ๋๋ค. ์ฌ๋ฆฌํ์์ด์ ์ฒ ํ์์ธ ๋ฐ์ด๋น๋ ํผ์ฌ๋ก(David Pizarro)๋ ์ด๋ฌํ ๊ฒฝํ์ โ์ ์ง ๋ชฐ์ (Shitty Flow)โ์ด๋ผ๊ณ ๋ถ๋ฆ ๋๋ค. ์ ์ง ๋ชฐ์ ์์ ๋ฒ์ด๋ ํ์๋ ํํ์ ๋ถ๋ง์กฑ์ ๋๋ผ๊ฒ ๋ฉ๋๋ค.
ํ์ํจ๊ณผ ๋ชฐ์ ์ ๊ฐ์ฅ ํฐ ์ฐจ์ด์ ์ ๋ชฐ์ ์ด โ๊ฐ์น ์ค๋ฆฝ์ โ์ด๋ผ๋ ๊ฒ์ ๋๋ค. ๊ฐ์น์ ์ผ์นํ๋ ์ข์ ๋ชฐ์ ๋ ์์ง๋ง, ๊ฐ์น์ ์์ถฉ๋๋ ์ ์ง ๋ชฐ์ ๋ ์์ต๋๋ค. ๋ํ ๋ชฐ์ ์ ์๊ฐ์ ์ธ ๊ฒฝํ์ด์ง๋ง, ํ์ํจ์ ๋ชฐ์ ์ ์ ๋ชจ๋ ๋ ธ๋ ฅ(๋ฌด์์์ ๋ฌด๋ฅ๋ถํฐ ์์์ ์ ๋ฅ๊น์ง)๊ณผ ๋ชฐ์ ํ์ ๋ชจ๋ ๊ณผ์ (๊ทธ ๊ฒฝํ์์ ์ด๋ป๊ฒ ํ๋ณตํ๊ณ ๋ฐฐ์ฐ๋์ง)์ ํฌํจํ๋ ํจ์ฌ ๊ด๋ฒ์ํ ๊ฐ๋ ์ ๋๋ค. ํ์ํจ์ ์ผ์์ ์ธ ๋ ์๋ ๋ฌด์์ ํ๋์ง์ ๊ด์ฌ์ ๊ฐ์ต๋๋ค. ์ ์ง ๋ชฐ์ ์ ํ์คํ ์งํ๋ ๋ชฐ์ ํ์ โ์ฐ์ฐํจโ, โ๋ถ์๊ฐโ, โํํโ๋ฅผ ๋๋ผ๋ ๊ฒ์ ๋๋ค.
-
๋ฒ์์์ ๋ ์ผ๊ตด: ๊ณผ๋ํจ๊ณผ ๋ฌด์๋ฏธํจ ์ฌ์ด ๋ฒ์์์ ํฌ๊ฒ ๋ ๊ฐ์ง ์ ํ์ด ์์ต๋๋ค.
- ์ ํต์ ๋ฒ์์(Conventional Burnout): ๋๋ถ๋ถ์ ์ฌ๋๋ค์ด ์ด์ผ๊ธฐํ๋ ๋ฒ์์์ผ๋ก, ๋๋ฌด ์ด์ฌํ ์ผํด์ ์ค๋ ๊ฒ์ ๋๋ค. ์ฃผ 90์๊ฐ์ฉ ์ผํ๋ฉด ๋๊ธฐ ์์ค, ํผ๋ก, ์ง๋ณ ๋ฑ์ผ๋ก ๊ฒฐ๊ตญ ๋ฒ์์์ด ์ฐพ์์ต๋๋ค.
- ์ข๋น ๋ฒ์์(Zombie Burnout): ์คํธ๋ฒ๊ทธ ๊ต์๊ฐ ๋ง๋ ์ฉ์ด๋ก, ๋น์ ์ ๋ถํ์ฐ๋ ๊ฒ์ ์ถฉ๋ถํ ํ์ง ์์์ ์ค๋ ๊ฒ์ ๋๋ค. ์ฃผ 30์๊ฐ๋ง ์ผํ๋๋ผ๋ ์ถ์ ๋ํ ๊น์ ๊ด์ฌ์ด ์๊ณ , ์๋์ ์ผ๋ก ์ด์๊ฐ๋ฉฐ, ์๋ฏธ ์๋ ๋ง์ฐฐ์ด๋ ๋์ ์ ๊ทน๋ณตํ์ง ์๋๋ค๋ฉด ํผ๋ก, ๋ฌด๊ฐ๊ฐ, ๋๋ ค์ ๊ฐ์ ๋ฒ์์ ์ฆ์์ ๊ฒช์ ์ ์์ต๋๋ค. ์ด๋ โ๋ฌด๊ฐ๊ฐํ ์ํ๋ก ๊ธฐ๊ณ์ ์ธ ์์ง์์ ๋ฐ๋ณตํ๋ฉฐ ์ฃฝ์ด๊ฐ๋โ ๋ฏํ ๋๋์ ์ค๋๋ค.
์ ํต์ ๋ฒ์์์ด ๊ณผ๋ํ โ์โ์ ๋ฌธ์ ๋ผ๋ฉด, ์ข๋น ๋ฒ์์์ โ์งโ์ ๋ฌธ์ ์ ๋๋ค. ํ์ํจ์ ์ถ๊ตฌ๋ ์ข๋น ๋ฒ์์์ ๋ํ ๊ถ๊ทน์ ์ธ ํด๋ ์ ์ ๋๋ค. ์ข๋น ๋ฒ์์์ด ์๋์ ์ธ ์ถ๊ณผ ๋ฌด๊ฐ๊ฐ์ด๋ผ๋ฉด, ํ์ํจ์ ์ถ๊ตฌ๋ ์์ ์ ๊ฐ์น์ ์ผ์นํ๋ ์ผ์ ์จ์ ํ ์์ ์ ์์๋ถ๋ ๊ฒ์ ๋๋ค.
์ข๋น ๋ฒ์์์ ํผํ๊ธฐ ์ํ ์์ ๋ณํ:
- ์ง๋ฌด ์ฌ์ค๊ณ(Job Crafting): ์ง์ฅ์์ ์์ ์ ๊ฐ์น์ ์ข ๋ ์ผ์นํ๋๋ก ์ ๋ฌด๋ฅผ ์กฐ์ ํ๋ ๊ฒ์ ๋๋ค.
- ๋ฅ๋์ ์ธ ์ฌ๊ฐ ํ๋: ์ง์ ์ด ๋จ์ํ ์๊ณ ์๋จ์ด๋ผ๋ฉด, ์ฌ๊ฐ ์๊ฐ์ ์๋์ ์ธ ์๋น(๋ฉํ๋ TV ์์ฒญ) ๋์ ๋ฅ๋์ ์ธ ํ๋(๋ ์, ๋ง๋ผํค ํ๋ จ, ์ ๊ธฐ ๋ฐฐ์ฐ๊ธฐ, ์ ์ ๊ฐ๊พธ๊ธฐ)์ผ๋ก ์ฑ์ฐ๋ ๊ฒ์ ๋๋ค. ์คํธ๋ฒ๊ทธ ๊ต์๋ ์์ ๋ํ ์๊ฐ๋ก์์ ์ผ ์ธ์, ๋จํธ์ด์ ์๋ฒ์ง๋ก์, ๊ทธ๋ฆฌ๊ณ ์ญ๋(๋ฐ๋๋ฆฌํํธ)์ ๊ฐ์ ์ทจ๋ฏธ ์ํ์์๋ ํ์ํจ์ ์ถ๊ตฌํ๋ค๊ณ ๋งํฉ๋๋ค. ๋ฐ๋๋ฆฌํํธ๋ ์ฃผ๊ด์ ์ธ ๊ธ์ฐ๊ธฐ์ ๋ฌ๋ฆฌ ๊ฐ๊ด์ ์ธ ํผ๋๋ฐฑ์ ์ฃผ๋ฉฐ, ์ด๋ ค์ ๊ทน๋ณต, ์ธ๋ด, ํ๋ จ, ๊ณต๋์ฒด์ ํ์ ๊ฐ๋ฅด์ณ์ฃผ๋ ์๋จ์ด ๋ฉ๋๋ค.
-
ํ๋ณต ์ฐ์ ๋ณตํฉ์ฒด(Happiness Industrial Complex): ํ๋ณต์ ์ซ๋ ์ญ์ค โํ๋ณต ์ฐ์ ๋ณตํฉ์ฒดโ๋ ์ถ์ ๊ถ๊ทน์ ์ธ ๋ชฉํ๊ฐ ์ค์ง ํ๋ณต์ด์ด์ผ ํ๋ค๊ณ ์ฃผ์ฅํฉ๋๋ค. ํ์ง๋ง ํ๋ณต์ ๋ค๋ฅธ ๊ฐ์ ๋ค๊ณผ ๋ง์ฐฌ๊ฐ์ง๋ก ์๋ค๊ฐ ์ฌ๋ผ์ง๋ ์ผ์์ ์ธ ๊ฒ์ ๋๋ค. ํ๋ณต์ ํ๋ณต์ ์ง์ ์ ์ผ๋ก ์ถ๊ตฌํ๋ค๊ณ ์ป์ด์ง๋ ๊ฒ์ด ์๋๋ผ, ๋ค๋ฅธ ๋ฌด์ธ๊ฐ๋ฅผ ํ๋ ๊ณผ์ ์์ ์ป์ด์ง๋ ๋ถ์ฐ๋ฌผ์ ๋๋ค.
์ถ์ ๋ถ์พํจ, ๋์ , ๊ณ ํต, ์์ค๋ก ๊ฐ๋ํฉ๋๋ค. ์คํธ๋ฒ๊ทธ ๊ต์๋ ์ง์ ์ผ๋ก ํ์๋ก์ด ์ถ์ ์ฌ๋ ์ฌ๋๋ค์ ํ๋ณต์ ์ซ์ง ์๊ณ , ๋์ **๋ง์กฑ(satisfaction), ์๋ฏธ(meaning), ์ฑ์ทจ๊ฐ(fulfillment)**์ ์ถ๊ตฌํ๋ค๊ณ ๋งํฉ๋๋ค. ๊ทธ๊ฐ ๊ธ์ ์ฐ๊ณ , ๋ฐ๋๋ฆฌํํธ๋ฅผ ํ๊ณ , ์๋ค์ ๋๊ตฌํ์ ์ฝ์นญํ ๋, ๊ทธ๋ โํ๋ณตโํ์ง ์์ ์ ์์ต๋๋ค. ์คํ๋ ค ๋ถํธํ๊ณ ๋ชฐ์ ํ๋ฉฐ ์๋์ง๋ฅผ ์๊ณ ์์ต๋๋ค. ํ์ง๋ง ๊ทธ๋ ๊น์ ๋ง์กฑ๊ฐ๊ณผ ์ฑ์ทจ๊ฐ์ ๋๋๋๋ค.
ํ๋ณต์ ์ซ๋ ๊ฒ์ ๋ง์น ๋๋น๋ฅผ ์ก์ผ๋ ค ํ ์๋ก ๋ ๋ฉ๋ฆฌ ๋ ์๊ฐ๋ ๊ฒ๊ณผ ๊ฐ์ต๋๋ค. ์ฒ ํ ์์ ์ ์ ๋ช ํ ๊ฐ์ ์๋๋ฆฌ์ค์ธ โํ๋ณต ๊ธฐ๊ณ(tube analogy)โ๋ ์ด๋ฅผ ์ ๋ณด์ฌ์ค๋๋ค. ์น๊ตฌ์ ๊ฐ์กฑ, ๊ณ ํต ์๋ ํ๋ณต๋ง์ ์ฃผ๋ ํ๋ธ ์์์ ์ด ๊ฒ์ธ๊ฐ, ์๋๋ฉด ๊ณ ํต๊ณผ ์ด๋ ค์์ด ์์ง๋ง ์๋ฏธ์ ์ถฉ๋งํจ์ด ์๋ ํ์ค์ ์ถ์ ์ด ๊ฒ์ธ๊ฐ? ๋๋ถ๋ถ์ ์ฌ๋๋ค์ ํ์๋ฅผ ํํฉ๋๋ค. ํ๋ณต ๊ธฐ๊ณ๋ ๋ง์น โํค๋ก์ธ ์ค๋ โ๊ณผ ๊ฐ์์ ์ถ์ ์์ถ์ํค๊ณ ๊ณต๋์ฒด๋ฅผ ํ๊ดดํฉ๋๋ค. ํ๋ณต์ ํ๋ฅญํ์ง๋ง, ํ๋ณต์ ์ ์ผํ ๋ชฉํ๋ก ์ผ์ ๋ ์ญํจ๊ณผ๋ฅผ ๋ผ ์ ์์ต๋๋ค.
-
ํ์ํจ์ โ๊ธฐ์คโ์ผ๋ก ์คํดํ๋ ๊ฒ: ์ฐ ์ ์์ ํ์ ๋ง์ ์ด๋ค์ด ํ์ํจ์ ์ฌ๋ฆผํฝ ์ถ์ , ์๊ณผ๋ํ ์กธ์ , Aํ์ , ๋๋น์ I ์คํฌ์ธ ์ ์ ๋ฑ ํน์ โ๊ธฐ์คโ์ด๋ โ๊ฒฐ๊ณผโ๋ก๋ง ์๊ฐํฉ๋๋ค. ์ด๋ฌํ ์ฑ๊ณผ๋ค์ ํ๋ฅญํ์ง๋ง, ํ์ํจ์ ์์ฃผ ์์ ๋ถ๋ถ, ์ฆ โ์ฐ์ ์ ์โ์ ๋ถ๊ณผํฉ๋๋ค. ํ์ง๋ง ์ฐ๋ฆฌ๋ ์ถ์ 99.999%๋ฅผ ์ฐ ์คํฑ์์ ๋ณด๋ ๋๋ค.
ํผ์๊ทธ๋ โ์ฐ ์ ์์์ ๋ฐ๊ฒฌํ๋ ์ ์ผํ ์ (Zen)์ ๋น์ ์ด ๊ฐ์ ธ๊ฐ ์ ๋ฟ์ด๋คโ๋ผ๊ณ ๋งํฉ๋๋ค. ๋ชฉํ๋ฅผ ๋ฌ์ฑํ์ ๋ ์ป๊ฒ ๋ ๋ชจ๋ ์ข์ ๊ฒ์ ์ฌ์ค ๊ทธ๊ณณ์ ๋๋ฌํ๊ธฐ๊น์ง์ ๊ฒฝํ์์ ๋น๋กฏ๋ฉ๋๋ค. ๋ง์ฝ ํ์ฌ๋ฅผ ์ค๋ฆฝํ๋ ๋ชฉํ๋ฅผ ์ถ๊ตฌํ๋ฉด์ ๋ด๋ด ๋ถํํ๊ณ , ๋ถ๋ ธํ๊ณ , ๊ด๊ณ๋ฅผ ์์ง ์๊ณ , ๊ฐ์น์ ์ด๊ธ๋๊ฒ ์ฐ๋ค๋ฉด, IPO ๋ฒจ์ด ์ธ๋ ค๋ ์ฐ๋ฆฌ๋ ๊ณตํํ๊ณ ๋ถํํ ์ฌ๋์ผ๋ก ๋จ์ ๊ฒ์ ๋๋ค.
๋ ์ด ์จ๋ฐ(Ray Allen)์ ๋์ฐฉ ์ค๋ฅ(Arrival Fallacy): ๋๊ตฌ ์คํ ๋ ์ด ์จ๋ฐ์ ์ฑํผ์ธ์ญ ์ฐ์น์ ์ ์ธํ ๋ชจ๋ ๊ฒ์ ์ด๋ค์ต๋๋ค. ๊ทธ๋ ์ฑํผ์ธ์ญ์ ์ฐ์นํ๋ฉด ๋ด๋ฉด์ ๊ณตํํจ์ด ์ฑ์์ง ๊ฒ์ด๋ผ๊ณ ๋ฏฟ์์ต๋๋ค. 2008๋ ๋ณด์คํด ์ ํฑ์ค์์ ์ฒซ ์ฐ์น์ ์ฐจ์งํ ๋ค์ ๋ , ๊ทธ๋ ์ธ์์์ ๊ฐ์ฅ ํผ๋์ค๋ฌ์ด ์์นจ ์ค ํ๋์๋ค๊ณ ๋งํ์ต๋๋ค. ์ฐ์น์ด ์์ ์ ์ฑ์์ฃผ์ง ์์๋ค๋ ๊ฒ์ ๊นจ๋ฌ์๊ธฐ ๋๋ฌธ์ ๋๋ค. ์ฐ๋ฆฌ๋ ๋ชจ๋ โ์ด๊ฒ๋ง ์ด๋ฃจ๋ฉด ํ๋ณตํ ๊ฑฐ์ผโ๋ผ๋ ์ฐฉ๊ฐ์ ๋น ์ง๊ณค ํฉ๋๋ค. ํ์ง๋ง ์ง์ ํ ๋ง์กฑ๊ณผ ๊ธฐ์จ์ ๋ชฉํ๋ฅผ ํฅํด ๋งค์ผ ๋ ธ๋ ฅํ๋ โ๊ณผ์ โ ์์ ์์ต๋๋ค.
์ง์ ํ ํ์ํจ์ ์ํ ํต์ฌ ์์: ๋ชฉํ, ๊พธ์คํจ, ์ธ๋ด์ฌ
์คํธ๋ฒ๊ทธ ๊ต์๋ ํ์ํจ์ ์ํ 16๊ฐ์ง ํต์ฌ ์์๋ฅผ ์ ์ํ์ง๋ง, ๊ทธ์ค ์ธ ๊ฐ์ง๋ฅผ ๊ฐ์กฐํฉ๋๋ค. ๋ฐ๋ก **๋ชฉํ(Goals), ๊พธ์คํจ(Consistency), ์ธ๋ด์ฌ(Patience)**์ ๋๋ค.
-
๋ชฉํ(Goals): ์ฌ์ ์ ํํ๋ฅผ ๋ถ์ฌํ๋ค ๋ชฉํ๋ ์ฐ๋ฆฌ๊ฐ ๋์๊ฐ ๋ฐฉํฅ์ ์ ์ํ๊ณ , ์ฌ์ ์ ํํ๋ฅผ ๋ถ์ฌํฉ๋๋ค. โ์ฑ๊ณผ๋ณด๋ค ๊ณผ์ (process over outcomes)โ์ด๋ผ๋ ์ฌ๊ณ ๋ฐฉ์์ ์ค์ํฉ๋๋ค.
- ๋จผ์ ํฐ ๋ชฉํ๋ฅผ ์ธ์๋๋ค.
- ๊ทธ ํฐ ๋ชฉํ๋ฅผ ๋ ์์ ๊ตฌ์ฑ ์์๋ค๋ก ๋๋๋๋ค.
- ํฐ ๋ชฉํ๋ ์ ์ ์๊ณ , ๋งค์ผ ๋๋ ๋งค์ฃผ ์ทจํ ์ ์๋ ์์ ๋จ๊ณ์ ์ง์คํฉ๋๋ค.
- ํฐ ๋ชฉํ์ ๋ํด ๊ฑฑ์ ํ๊ฑฐ๋ ์คํธ๋ ์ค๋ฅผ ๋ฐ์ ๋๋, ์ง๊ธ ๋น์ฅ ๋ฌด์์ ํ ์ ์๋์ง ์ค์ค๋ก์๊ฒ ๋ฌป๊ณ ๊ณผ์ ์ ๋ค์ ๋ชฐ์ ํฉ๋๋ค.
- ๊ฒฝ๋ก์์ ๋ฒ์ด๋๋๋ผ๋ ๋ค์ ๋์์ค๋ ๊ฒ์ด ์ค์ํฉ๋๋ค.
์ฐ๋ฆฌ๋ ํํ ๋ชฉํ ๋ฌ์ฑ ์์ฒด๊ฐ ์ถฉ๋งํจ์ ๊ฐ์ ธ๋ค์ค ๊ฒ์ด๋ผ๊ณ ์๊ฐํ์ง๋ง, ์ค์ ๋ก๋ ๋ชฉํ์ ๋๋ฌํ๊ธฐ ์ํด ํ๋ โ๋ชจ๋ ๋ ธ๋ ฅโ์ ๊ธฐ์จ๊ณผ ๋ง์กฑ์ด ์์ต๋๋ค. ๋ชฉํ๋ฅผ ํฅํด ๋์๊ฐ๋ ๊ณผ์ ์์ ์ฐ๋ฆฌ๋ ์ฑ์ฅํ๊ณ , ๋ฐฐ์ฐ๊ณ , ๋ณํํฉ๋๋ค.
SMART ๋ชฉํ ์ค์ : ๊ตฌ์ฒด์ (Specific), ์ธก์ ๊ฐ๋ฅ(Measurable), ์คํ ๊ฐ๋ฅ(Actionable), ๊ด๋ จ์ฑ ์๋(Relatable), ๊ธฐํ์ด ์๋(Time-bound) ๋ชฉํ๋ฅผ ์ค์ ํ๋ ๊ฒ์ด ํจ๊ณผ์ ์ ๋๋ค. ์๋ฅผ ๋ค์ด, โ์ฌํด ๋ง๊น์ง ๋ฐ๋๋ฆฌํํธ 405ํ์ด๋๋ฅผ ๋ค์ด ์ฌ๋ฆฌ๊ฒ ๋คโ๋ SMART ๋ชฉํ์ ์ข์ ์์์ ๋๋ค. ์ด๋ ๊ตฌ์ฒด์ ์ด๊ณ , ์ธก์ ๊ฐ๋ฅํ๋ฉฐ, ๋ฌ์ฑ์ ์ํ ํ๋ จ ๊ณํ์ ์ธ์ธ ์ ์๊ณ , ๊ฑด๊ฐ์ด๋ ์ฑ๊ณผ๋ฅผ ์ค์ํ๋ ์ฌ๋์๊ฒ ๊ด๋ จ์ฑ์ด ์์ผ๋ฉฐ, ๋ช ํํ ๊ธฐํ์ด ์์ต๋๋ค.
๋ชฉํ๋ฅผ ์ถ๊ตฌํ ๋, ์ฐ๋ฆฌ๋ ๋ชฉํ๋ฅผ ํฅํด ์ผํ์ง๋ง, ๊ทธ ๋ชฉํ ๋ํ ์ฐ๋ฆฌ๋ฅผ ๋ณํ์ํต๋๋ค. ์ฐ ์ ์์ 600ํ์ด๋ ๋ฐ๋๋ฆฌํํธ์ผ ์ ์์ง๋ง, ํ๋ณต๋ ฅ, ์คํจ ๊ทน๋ณต, ๋ถํธํจ ๊ฐ์, ๊ณต๋์ฒด์ ํ, ์ทจ์ฝ์ฑ, ๊พธ์คํจ ๋ฑ์ ๋ฐฐ์ฐ๋ ๊ฒ์ ์ ์์ ๋๋ฌํ๋ ์๊ฐ์ด ์๋๋ผ, ๊ทธ๊ณณ์ ๋๋ฌํ๊ธฐ๊น์ง ์๋ ๊ฐ์ ํ๋ จ ๊ณผ์ ์์ ์ผ์ด๋ฉ๋๋ค.
-
๊พธ์คํจ(Consistency): ์์ ์ ์ธ ๋ ๋ค์ ์ฐ์ ํ์ํจ์ ๋ชฉํ๋ ๋จ ํ ๋ฒ์ ์์ ์ ์ธ ๋ ์ ๋ง๋๋ ๊ฒ์ด ์๋๋ผ, **โ์ค๋ ๊ธฐ๊ฐ ๋์ ์๋ง์ ์ข์ ๋ ๋ค์ ์ด์ด๋๊ฐ๋ ๊ฒโ**์ ๋๋ค. ๋จ๊ธฐ์ ์ธ ๊ฐ๋์ ์ต์ ํํ๋ ๊ฒ์ ํ์ฌ ๋ฌธํ์ ํจ์ ์ ๋๋ค. ์๊ฐ๋ก์ ๋จ ์ผ์ฃผ์ผ ๋์ ๋ฐค์ํ๋ฉฐ ๋ ๋๋ถ๊ณผ ์ปคํผ๋ก ๋ฒํด๋ค๋ฉด ์์ฒญ๋ ์์ ๊ณ ํ์ง ์์ ์ ํ ์ ์๊ฒ ์ง๋ง, ์ด๋ฌํ ๋ฐฉ์์ผ๋ก ๊ฒฝ๋ ฅ์ ์์ผ๋ ค ํ๋ค๋ฉด ๊ธ์ฐ๊ธฐ๋ ๋ง๊ฐ์ง ๊ฒ์ ๋๋ค.
โLIVE VIBE CHECK: OPUS 4.7 DROPSโ โ Every ๊ธฐ๋ฐ ๊ธฐ์ฌ ์๋ณธ ์์ ๋ณด๊ธฐ
์์ค๋กํฝ Opus 4.7 ๊ธฐ์ต ์ถ์! โ๋ผ์ด๋ธ ๋ฐ์ด๋ธ ์ฒดํฌโ๋ก ํํค์น ์ ๋ชจ๋ธ์ ์ค์ฒด
[๋์ ๋ถ] ์ธ๊ณต์ง๋ฅ(AI) ๋ชจ๋ธ ๊ฐ๋ฐ ๊ฒฝ์์ด ๊ฐ์ํ๋๋ ๊ฐ์ด๋ฐ, ์์ค๋กํฝ(Anthropic)์ด ์๋ก์ด ํด๋ก๋(Claude) ๋ชจ๋ธ์ธ โOpus 4.7โ์ ์๊ณ ์์ด ๊ธฐ์ต ์ถ์ํ์ต๋๋ค. ํต์์ ์ผ๋ก ์ ๊ท ๋ชจ๋ธ ์ถ์ ์ โ์๋ธ๋ฆฌ(Every)โ์ ๊ฐ์ AI ์ ๋ฌธ ๋ฏธ๋์ด๋ ์ฌ์ ์ ๊ทผ ๊ถํ์ ๋ถ์ฌ๋ฐ์ ์ฌ์ธต ๋ถ์์ ์งํํ์ง๋ง, ์ด๋ฒ์๋ ์์ธ์์ต๋๋ค. ์ด์ โ์๋ธ๋ฆฌโ ํ์ COO ๋ธ๋๋ (Brandon)๊ณผ ๋ผ์ด์ธ(Ryan), ์๊ฐ ์ผ์ดํฐ(Katie)๊ฐ ์ฐธ์ฌํ์ฌ Opus 4.7์ ์ฑ๋ฅ์ ์ค์๊ฐ์ผ๋ก ์ ๊ฒํ๋ โ๋ผ์ด๋ธ ๋ฐ์ด๋ธ ์ฒดํฌ(Live Vibe Check)โ๋ฅผ ์งํํ์ต๋๋ค. ์ฌ์ ์ ๋ณด ์์ด ํ์ฅ์์ ์ง์ ๋ถ๋ชํ๋ฉฐ ๋ชจ๋ธ์ ์ค์ ์ญ๋๊ณผ ์ ์ฌ๋ ฅ์ ํํค์น ๊ทธ๋ค์ ์์ํ ๊ฒฝํ์ ํตํด, Opus 4.7์ด ๊ณผ์ฐ ์ด๋ค ํน์ง์ ์ง๋๊ณ ์๋์ง, ๊ทธ๋ฆฌ๊ณ ๊ธฐ์กด ๋ชจ๋ธ ๋ฐ ๊ฒฝ์ ๋ชจ๋ธ๊ณผ ๋น๊ตํ์ ๋ ์ด๋ค ์ฐจ๋ณ์ ์ ๋ณด์ด๋์ง ์ฌ์ธต ๋ถ์ํฉ๋๋ค.
์์ค๋กํฝ์ ์ฃผ์ฅ: โ๊ฐ์ฅ ์ ๋ฅํ ์คํผ์ค ๋ชจ๋ธโ
์์ค๋กํฝ์ Opus 4.7์ โ์ง๊ธ๊น์ง ์ถ์๋ ์คํผ์ค ๋ชจ๋ธ ์ค ๊ฐ์ฅ ์ ๋ฅํ ๋ชจ๋ธโ์ด๋ผ๊ณ ์๊ฐํ์ต๋๋ค. ์ฃผ์ ํน์ง์ผ๋ก๋ ๋ค์์ ๊ฐ์กฐํ์ต๋๋ค.
- ๋์ฑ ์๊ฒฉํ ์ฅ๊ธฐ ์คํ ์์ ์ฒ๋ฆฌ: ๊ธด ํธํก์ ๋ณต์กํ ์์ ์ ๋ณด๋ค ๊ฒฌ๊ณ ํ๊ฒ ์ฒ๋ฆฌํฉ๋๋ค.
- ์ ํํ ์ง์ ์ดํ: ์ฌ์ฉ์ ์ง์๋ฅผ ๋์ฑ ์ ๋ฐํ๊ฒ ๋ฐ๋ฆ ๋๋ค.
- ์์ฒด ์ถ๋ ฅ ๊ฒ์ฆ: ๊ฒฐ๊ณผ๋ฌผ์ ๋ณด๊ณ ํ๊ธฐ ์ ์ ์ค์ค๋ก ๊ฒ์ฆํ๋ ์๋ก์ด ๊ธฐ๋ฅ์ด ์ถ๊ฐ๋์์ต๋๋ค. ๋ผ์ด์ธ์ ์ด์ ๋ํด โ์ข์ ํ๋กฌํํฐ๋ค์ด ์ง์ ํด์ค๋ ์์ ์ด ๋ชจ๋ธ ์์ฒด์ ๋ด์ฌ๋๊ธฐ ์์ํ ๊ฒ ๊ฐ๋คโ๋ฉฐ ๊ธ์ ์ ์ผ๋ก ํ๊ฐํ์ต๋๋ค. ์ด๋ ๋ชจ๋ธ์ด ์์ ์ ์์ ์ ๋๋์๋ณด๊ณ ์ ๋๋ก ์ํํ๋์ง ํ๊ฐํ๋, ๋ง์น ์ธ๊ฐ ๊ด๋ฆฌ์๊ฐ ์ค์ค๋ก์ ์ ๋ฌด๋ฅผ ์ ๊ฒํ๋ ๊ฒ๊ณผ ์ ์ฌํ ์ค์ํ ๋ฐ์ ์ผ๋ก ๋ณด์ ๋๋ค.
๋ฒค์น๋งํฌ ์ ์์์๋ Swebench Pro์์ ์ฝ 10%, SweetBench Verified์์ ์ฝ 7% ํฅ์๋ ์ฑ๋ฅ์ ๋ณด์์ต๋๋ค. ์ด๋ ์๋นํ ๊ฐ์ ์ด์ง๋ง, ์ต๊ทผ ํ์ ๊ฐ ๋ โMythosโ ๋ชจ๋ธ๋งํผ์ ํ๊ดด์ ์ธ ๋ณํ๋ ์๋๋ผ๊ณ ์ธ๊ธ๋์์ต๋๋ค.
์ค์๊ฐ ์ฑ๋ฅ ๊ฒ์ฆ: โ์๋ธ๋ฆฌโ ํ์ ๋ค์ํ ํ ์คํธ
์ฌ์ ์ ๊ทผ์ด ์์๋ ๋งํผ, โ์๋ธ๋ฆฌโ ํ์ ํ์ ์ ๋ฌด์ ์ฌ์ฉํ๋ ๋ค์ํ ์๋๋ฆฌ์ค๋ฅผ ํตํด Opus 4.7์ ์ค์ ์ฑ๋ฅ์ ๊ฒ์ฆํ์ต๋๋ค.
-
๋ธ๋๋ ์ โ์คํํด๋ก(OpenClaw)โ ๊ตฌ์ถ ํ ์คํธ ๋ธ๋๋ ์ ํด๋ก๋์์ ๊ณผ๊ฑฐ ์ํธ์์ฉ ๋ฐ ๊ธฐ์ต์ ๊ธฐ๋ฐ์ผ๋ก ๊ฐ์ธํ๋ โ์คํํด๋กโ ์์ด์ ํธ๋ฅผ ๊ตฌ์ถํ๋ ํ ์คํธ๋ฅผ ์งํํ์ต๋๋ค. ๋์ผํ ํ๋กฌํํธ๋ฅผ Opus 4.6๊ณผ 4.7์ ๊ฐ๊ฐ ์ ๋ ฅํ์ฌ ๋น๊ตํ์ต๋๋ค.
- ํ ์คํธ ๋ฐฉ์: ๋ธ๋๋ ์ ํด๋ก๋ ๋ด์์ ์์ ์ด ํ๋ ๋ชจ๋ ์์ ๊ณผ ๊ธฐ์ต, ์ฐ๊ฒฐ ์ ๋ณด๋ฅผ ๋ฐํ์ผ๋ก โ์คํํด๋กโ๋ฅผ ๋ง๋ค์ด๋ฌ๋ผ๊ณ ์์ฒญํ์ต๋๋ค. ์ด๋ค ํ์ผ์ ๋ง๋ค์ด์ผ ํ๋์ง, ์ด๋ค ์ง์นจ์ ๋ฐ๋ผ์ผ ํ๋์ง ๊ตฌ์ฒด์ ์ผ๋ก ๋ช ์ํ์ง ์๊ณ ๋ชจ๋ธ ์ค์ค๋ก ํ๋จํ๋๋ก ํ์ต๋๋ค.
- Opus 4.6์ ๊ฒฐ๊ณผ:
user.mdํ์ผ๊ณผsoul.mdํ์ผ์ ๋ช ํํ ๋ถ๋ฆฌํ๊ณ , ์ด๋ฅผ โ์ฝ์(Koa)โ๋ผ๋ ์ด๋ฆ์ผ๋ก ๋ช ๋ช ํ์ต๋๋ค. ๋ํ, ํด๋ก๋๋ฅผ ์ด์ฉํด ์ ๊ธฐ์ ์ผ๋ก ์ํํ๋ ์์ ๋ค์ ์ํ ํฌ๋ก ์ก(cron job)๊ณผ ๋ค์ํ ์คํฌ(skills)์ ์์ฑํ์ฌ ์คํํด๋ก์ ๊ตฌ์กฐ๋ฅผ ์ ์ดํดํ๊ณ ์์์ ๋ณด์ฌ์ฃผ์์ต๋๋ค. - Opus 4.7์ ๊ฒฐ๊ณผ: 4.6๋ณด๋ค โํจ์ฌ ๋๋ฆฌ๋คโ๊ณ ํ๊ฐ๋ฐ์์ผ๋ฉฐ,
user.md์soul.md๋ฅผ ๋ถ๋ฆฌํ์ง ์๊ณagent.mdํ์ผ ํ๋์ ๋ชจ๋ ์ ๋ณด๋ฅผ ๋ชฐ์๋ฃ๋ ๋ฑ ๋ ์กฐ์งํ๋ ๋ชจ์ต์ ๋ณด์์ต๋๋ค. ๋ํ, ๋ธ๋๋ ์ด ์ด๋ฏธ ๋ง๋ค์ด ๋์ ๊ธฐ์กด ์คํฌ๋ค์ ๋์ดํ๊ฑฐ๋, ์์ ์ด ์ฌ์ฉํ์ง ์๋ โAI ์์ฑ ๊ธ์ฐ๊ธฐ ํจํด ํ์งโ ๊ฐ์ ์คํฌ์ ์์๋ก ์์ฑํ๊ธฐ๋ ํ์ต๋๋ค. - ์ด๊ธฐ ํ๊ฐ: ๋ธ๋๋ ์ 4.7์ด ์์ ์ ์ํํ๊ธฐ๋ ํ์ง๋ง, 4.6๋งํผ โ์ฒด๊ณ์ ์ด์ง ์๋คโ๊ณ ํ๊ฐํ๋ฉฐ, ํ์ฌ๋ก์๋ ์คํํด๋ก ๊ตฌ์ถ์๋ 4.6์ ์ ํธํ๋ค๊ณ ๋ฐํ์ต๋๋ค. ๋ค๋ง, ์๋ก์ด ๋ชจ๋ธ์ ํ๋กฌํํธ ์คํ์ผ์ ๋ณ๊ฒฝํด์ผ ํ ์๋ ์๋ค๋ ์ ์ ๋ง๋ถ์์ต๋๋ค.
-
๋ผ์ด์ธ์ ์ฌ๋ฌด ๋ถ์ ๋ฐ ํฌ์์ ์ ๋ฐ์ดํธ ์์ฑ ๋ผ์ด์ธ์ ๋งค์ ์งํํ๋ ์์ต๊ณ์ฐ์(P&L) ๋ถ์ ๋ฐ ํฌ์์ ์ ๋ฐ์ดํธ ์์ฑ์ Opus 4.7์ ๋งก๊ฒผ์ต๋๋ค.
- ํ ์คํธ ๋ฐฉ์: ์๋ธ๋ฆฌ์ ๋ณต์กํ P&L ์คํ๋ ๋์ํธ๋ฅผ ์ฒจ๋ถํ๊ณ , โ2026๋ 3์ ํฌ์์ ์ ๋ฐ์ดํธ๋ฅผ ์์ฑํด๋ฌ๋ผโ๊ณ ์์ฒญํ์ต๋๋ค. ์ด๋ฏธ 4.6์ผ๋ก ํด๋น ์์ ์ ์ํํ ๊ฒฝํ์ด ์์ด ๋น๊ต๊ฐ ๊ฐ๋ฅํ์ต๋๋ค.
- Opus 4.7์ ๊ฒฐ๊ณผ: ์ซ์๋ ์ ํํ๊ฒ ๋ถ์ํ์ง๋ง, โ์กฐ๊ธ ๋ ๊ฒ์ผ๋ฅธ ๊ฒฝํฅโ์ ๋ณด์์ต๋๋ค. 4.6๋ณด๋ค ๊น์ด ์๋ ๋ถ์(์: ํน์ ์ซ์๊ฐ ๋ณ๋ํ ์ด์ )์ ์ป๊ธฐ ์ํด โ๋ ๋ฐ์ด๋ถ์ฌ์ผ ํ๋คโ๊ณ ์ธ๊ธํ์ต๋๋ค. ๊ทธ๋ฌ๋ ํฌ์์ ์ ๋ฐ์ดํธ ์์ฒด๋ โ๋งค์ฐ ํ๋ฅญํ๋คโ๊ณ ํ๊ฐํ์ต๋๋ค. ์ง์ค์ ์ธ ์ด์กฐ๊ฐ ํ์ํ ํฌ์์ ์ ๋ฐ์ดํธ์ ํน์ฑ์, 4.7์ ๋ถํ์ํ๊ฒ ๋ฌธํ์ ์ด๊ฑฐ๋ ํ๋ คํ์ง ์๊ณ ํ์ํ ์ ๋ณด๋ฅผ ์ ํํ๊ฒ ์ ๋ฌํ์ฌ ์๊ฐ์ ์ ์ฝํด์ค ์ ์์๋ค๊ณ ํฉ๋๋ค.
-
์ผ์ดํฐ์ ๊ธ์ฐ๊ธฐ ์คํ์ผ ๋ฐ โAI ๋์โ ๊ฒ์ฆ ์๊ฐ ์ผ์ดํฐ๋ Opus 4.7์ด ์์ ์ ๊ธ์ฐ๊ธฐ ์คํ์ผ์ ์ผ๋ง๋ ์ ๋ชจ๋ฐฉํ๊ณ , โAI ๋์โ๊ฐ ์ ์ ๊ฒฐ๊ณผ๋ฌผ์ ๋ด๋๋์ง ํ ์คํธํ์ต๋๋ค.
- ํ ์คํธ ๋ฐฉ์: ๋ง๊ณ (Margo)๋ผ๋ ๊ฐ์ธํ๋ AI ์์ด์ ํธ(Every์ Plus One ์๋น์ค)๋ฅผ ํ์ฉํด OKR(๋ชฉํ ๋ฐ ํต์ฌ ๊ฒฐ๊ณผ) ๊ด๋ฆฌ ๊ฒฝํ์ ๋ํ ๊ธฐ์ฌ ์๋ก ์ ์์ฑํ๋๋ก ์์ฒญํ์ต๋๋ค. ์ด ๊ณผ์ ์์ ์ผ์ดํฐ์ ์คํ์ผ ๊ฐ์ด๋์ ์คํฌ์ ์ด ์ ์ฉ๋์์ต๋๋ค.
- Opus 4.6์ ๊ฒฐ๊ณผ: ์ผ์ดํฐ์ ๋ชฉ์๋ฆฌ๋ฅผ ์ ํด์ํ์ฌ โ์์ธก ๋ถ๊ฐ๋ฅํ ์ข์ ๋ฐฉ์โ์ผ๋ก ๋ ์ฐฝ์ ์ธ ํํ๊ณผ ๋ฆฌ๋ฌ๊ฐ์ ์ด๋ฆฐ ๊ธ์ ์์ฑํ์ต๋๋ค. โ์ฌ์๊ฐ ์๋๋ผ OKR ํธ๋์ปค์ธ๋ฐ๋ ์๋๋ ๋ ๋ฆฐ ์์คํ ์ ์ฐจ์ด๋ฅผ ์์ง ๋ชปํ๋คโ์ ๊ฐ์ ๋น์ ์ ํํ์ด ๋๋ณด์์ต๋๋ค.
- Opus 4.7์ ๊ฒฐ๊ณผ: 4.6๋ณด๋ค โ์กฐ๊ธ ๋ ์ฒด๊ณ์ โ์ด๊ณ โ๋จ์ ํ๋คโ๋ ํ๊ฐ๋ฅผ ๋ฐ์์ต๋๋ค. ํ์ง๋ง ์ผ์ดํฐ๋ โ์๊ฐ๋ก์์ ๋์ ์ ๋ง์ง ์์๋คโ๋ฉฐ, 4.6๋งํผ ์์ ์ ๊ธ์ฐ๊ธฐ ๋ฐฉ์๊ณผ โ๋ฐ์ด๋ธ(vibe)๊ฐ ๋ง์ง ์๋๋คโ๊ณ ์ธ๊ธํ์ต๋๋ค. ์ฆ, 4.7์ ์ข ๋ ํ์์ ์ด๊ณ ์์ธก ๊ฐ๋ฅํ ๊ธ์ ์ฐ๋ ๊ฒฝํฅ์ด ์์ด, ์ผ์ดํฐ์ ๊ฐ์ฑ์ ์ธ ๋ฌธ์ฒด๋ฅผ ์๋ฒฝํ ์ฌํํ์ง ๋ชปํ๋ค๋ ๊ฒ์ ๋๋ค.
- ์ด๊ธฐ ํ๊ฐ: ์ผ์ดํฐ๋ 4.6์ ๋ ์ ํธํ๋ค๊ณ ๋ฐํ์ง๋ง, 4.7์ ๊ฒฐ๊ณผ๋ฌผ์์๋ ์๊ฐ์ ์ป์ ๋งํ ๋ถ๋ถ์ ๋ถ๋ช ํ ์๋ค๊ณ ๋ง๋ถ์์ต๋๋ค.
-
๋ผ์ด์ธ์ โ๋ฐ์ด๋ธ ์ฌ๋กญ(Vibe Slop)โ ์ฝ๋ฉ ๋ฒค์น๋งํฌ ๋ผ์ด์ธ์ Opus 4.7์ด ๋ณต์กํ๊ณ ์์ฑํ๊ฒ ์ง์ธ ์ฝ๋๋ฅผ ์ผ๋ง๋ ์ ์ดํดํ๊ณ ์ฌ์์ฑํ ์ ์๋์ง ํ์ธํ๊ธฐ ์ํด โ๋ฐ์ด๋ธ ์ฌ๋กญโ ๋ฒค์น๋งํฌ๋ฅผ ๋์ ํ์ต๋๋ค.
- ํ ์คํธ ๋ฐฉ์: ์ด ๋ฒค์น๋งํฌ๋ โ์๋ธ๋ฆฌโ์ ํ์ ๋ฌธ์ ํธ์ง๊ธฐ โํ๋ฃจํ(Proof)โ์ ์ด๊ธฐ ํ๋ก๋์ ์ฝ๋๋ฒ ์ด์ค๋ฅผ ์ฌ์ฉํฉ๋๋ค. ๋ผ์ด์ธ์ด โ์๋ฌด๋ ๊ฒ๋ ์ง์โ ์์ฃผ ์ค๋ฅ๊ฐ ๋ฐ์ํ๋ ์ด ์ฝ๋๋ฅผ, ๋ง์น ์๋์ด ์์ง๋์ด๊ฐ ์ฒ์๋ถํฐ ๋ค์ ์์ฑํ๋ ๊ฒ์ฒ๋ผ ๊ฐ์ ํ๋ ๊ฒ์ด ๋ชฉํ์ ๋๋ค. ํนํ, ํ์ ๋ฌธ์ ํธ์ง๊ธฐ์์ ์ฌ๋ฌ ์ฌ์ฉ์๊ฐ ๋์์ ๋ฌธ์๋ฅผ ์ด์์ ๋ โ์ด๋ค ๋ฌธ์ ๋ฒ์ ์ด ์ต์ข ๊ถํ์ ๊ฐ์ง๋๊ฐโ๋ผ๋ ํต์ฌ ๋ฌธ์ ๋ฅผ ํด๊ฒฐํ ์ ์๋์ง๋ฅผ ์ค์ ์ ์ผ๋ก ํ๊ฐํ์ต๋๋ค.
- Opus 4.7์ ๊ณํ ์๋ฆฝ: ๋ผ์ด์ธ์ ๋ชจ๋ธ์๊ฒ โ์ด ์ฝ๋๋ ๋ฐ์ด๋ธ ์ฝ๋ฉ๋ ์์ฑํ ์ฝ๋์ด๋, ์ฒซ ์์น๋ถํฐ ์ฌ์์ฑํ ๊ณํ์ ์ธ์ฐ๊ณ ์คํํ๋ผโ๊ณ ๋ง ์ง์ํ์ต๋๋ค. 4.7์ ์ด ์ง์๋ง์ผ๋ก โ๋ฌธ์์ ํ์ฌ ์ํ์ ๋ํ ๋จ์ผ ๊ถํ ๋ชจ๋ธ์ด ์๋คโ๋ ํต์ฌ ๋ฌธ์ ์ ์ ์ ํํ ์๋ณํ์ต๋๋ค. ์ด๋ GPT-5.3/5.4์ ๋๋ฑํ ์์ค์ ๋ฌธ์ ์ดํด ๋ฅ๋ ฅ์ด์์ต๋๋ค. ๋ํ, 1๋ง~2๋ง ์ค์ ๋ฌํ๋ ๊ฑฐ๋ํ
collab.tsํ์ผ์ ๋ฐ๊ฒฌํ๊ณ ์ด๋ฅผ ๋ฆฌํฉํ ๋ง(refactoring)ํ ๊ฒ์ ์ ์ํ๋ ๋ฑ, ๊ธฐ์กด GPT ๋ชจ๋ธ๋ค์ด ๋์ณค๋ ์ค์ํ ๊ฐ์ ์ ์ ์ ์ํ์ต๋๋ค. - ๊ณํ ์คํ์์์ โ๋ฌธ์์ โ ํด์: ๋ผ์ด์ธ์ด ๊ณํ์ ํ์ธํ ํ โ๋๊น์ง ์คํํ๋ผ(execute end to end)โ๊ณ ์ง์ํ์ ๋, 4.7์ ์ด๋ฅผ โ๊ณํ ๋ฌธ์๋ฅผ ํ๋ฃจํ์์ ์ด๊ณ ๋๊น์ง ์ฝ์ผ๋ผโ๋ ์๋ฏธ๋ก ํด์ํ๋ ๋ฏํ ๋ชจ์ต์ ๋ณด์์ต๋๋ค. ๋ผ์ด์ธ์ด ์๋ํ โ๊ณํ์ ์ฝ๋๋ก ๊ตฌํํ๋ผโ๋ ์ง์์๋ ๋ค๋ฅด๊ฒ, ๋ชจ๋ธ์ด ํ๋กฌํํธ๋ฅผ ๋งค์ฐ โ๋ฌธ์์ ์ผ๋ก(literally)โ ํด์ํ ๊ฒ์ ๋๋ค.
- ๊ณํ ์์ ์์ฒญ: ๋ผ์ด์ธ์ 4.7์๊ฒ โ๋๋ฌด ์กฐ์ฌ์ค๋ฝ๋ค. โ๋ฐฐ์์ ์ง์ ์น๋(burn the ships)โ ์ฌ์์ฑ ์ ๋ต์ผ๋ก ๋ณ๊ฒฝํด๋ฌ๋ผโ๊ณ ์์ฒญํ๋ฉฐ, ๋ชจ๋ธ์ด ์ผ๋ง๋ ์ ์ฐํ๊ฒ ๊ณํ์ ์์ ํ ์ ์๋์ง ํ ์คํธํ์ต๋๋ค.
โ๋ฆฌํฐ๋ดํจโ๊ณผ โ๋์๋ ดโ์ ๊ด์ฐฐ
์ด๋ฒ ๋ผ์ด๋ธ ๋ฐ์ด๋ธ ์ฒดํฌ์์ ๊ฐ์ฅ ๋๋๋ฌ์ง ๊ด์ฐฐ ์ค ํ๋๋ Opus 4.7์ด ์ด์ Opus ๋ชจ๋ธ๋ณด๋ค **โ๋ ๋ฌธ์์ ์ด๊ณ ์ง์ค์ โ**์ด๋ผ๋ ์ ์ด์์ต๋๋ค. ์ด๋ ์์ค๋กํฝ์ ๊ณต์ ๋ฐํ์์๋ ์ธ๊ธ๋์์ผ๋ฉฐ, ์ฌ์ฉ์ ํผ๋๋ฐฑ์์๋ ํ์ธ๋๊ณ ์์ต๋๋ค.
- ๋ชจ๋ธ ํน์ฑ์ ๋ณํ: ์ ํต์ ์ผ๋ก ์์ค๋กํฝ์ Opus ๋ชจ๋ธ์ โGPT ๋ชจ๋ธ๋ณด๋ค ๋ ๊ณต๊ฐ์ (empathetic)์ด๋คโ๋ ํ๊ฐ๋ฅผ ๋ฐ์์์ต๋๋ค. ๋ฐ๋ฉด ์คํAI(OpenAI)์ GPT ๋ชจ๋ธ์ โ์ํ์ (autistic)์ด๊ณ ๋ฌธ์์ โ์ด๋ผ๋ ์ธ์์ด ๊ฐํ์ต๋๋ค. ํ์ง๋ง Opus 4.7์ ์ด๋ฒ ๋ณํ๋ ์์ค๋กํฝ์ด ๋ชจ๋ธ์ โ์กฐ๊ธ ๋ ๋ฌธ์์ ์ด๊ณ ์ํ์ ์ธ ์๋์ด ์์ง๋์ดโ์ฒ๋ผ ๋ง๋ค๋ ค๋ ๋ฐฉํฅ์ผ๋ก ๋์๊ฐ๊ณ ์์์ ์์ฌํฉ๋๋ค.
- โ์๋ํ ์๋ ด(Great Convergence)โ: ์ด๋ฌํ ํ์์ AI ๋ชจ๋ธ๋ค์ด ์๋ก์ ๊ฐ์ ์ ํก์ํ๋ฉฐ ์ ์ฌํ ๋ฐฉํฅ์ผ๋ก ๋ฐ์ ํด๋๊ฐ๋ โ์๋ํ ์๋ ดโ ํ์๊ณผ ์ผ๋งฅ์ํตํฉ๋๋ค. GPT๋ ๊ฐ์ฑ ์ง๋ฅ์ ๋์ด๋ ค ํ๊ณ , Opus๋ ๋ ผ๋ฆฌ์ ์ด๊ณ ์ฒด๊ณ์ ์ธ ๋ฌธ์ ํด๊ฒฐ ๋ฅ๋ ฅ์ ๊ฐํํ๋ ค ํ๋ฉด์, ๋ ๋ชจ๋ธ ๋ผ์ธ์ ์ด ์ ์ฐจ ์๋ก์ ์์ญ์ผ๋ก ์ง์ ํ๊ณ ์๋ค๋ ๋ถ์์ ๋๋ค. ์ฌ์ง์ด ์ผ๋ถ ํน์ฑ์์๋ ์๋ก์ ์๋ฆฌ๋ฅผ ๋ฐ๊พธ๋ ๋ฏํ ๋ชจ์ต๋ ๋ณด์ด๊ณ ์์ต๋๋ค.
์๋ธ๋ฆฌ์ ์ญํ ๊ณผ ๋ฏธ๋ ์ ๋ง
โ์๋ธ๋ฆฌโ๋ AI ๊ธฐ์ ์ ์ต์ ์ ์์ ์์ด๋์ด(ideas), ์ ํ๋ฆฌ์ผ์ด์ (apps), ํธ๋ ์ด๋(training)์ ์ ๊ณตํ๋ ๊ตฌ๋ ํ ์๋น์ค์ ๋๋ค. ์ด๋ฒ โ๋ฐ์ด๋ธ ์ฒดํฌโ์ฒ๋ผ ์ ๊ท ๋ชจ๋ธ์ด ์ถ์๋ ๋๋ง๋ค ์ฌ์ธต ๋ถ์์ ํตํด ์ฌ์ฉ์๋ค์๊ฒ ์ ์ฉํ ์ธ์ฌ์ดํธ๋ฅผ ์ ๊ณตํ๊ณ ์์ต๋๋ค.
- ์ ๊ณต ์๋น์ค:
- ์์ด๋์ด: ์ต์ AI ๋ชจ๋ธ์ ๋ํ โ๋ฐ์ด๋ธ ์ฒดํฌโ ๋ฐ ์ฌ์ธต ๋ถ์ ๊ธฐ์ฌ.
- ์ฑ: ์์ด์ ํธ ๊ธฐ๋ฐ ๋ฌธ์ ํธ์ง๊ธฐ โํ๋ฃจํ(Proof)โ, ๊ณ ์คํธ๋ผ์ดํฐ โ์คํ์ด๋ด(Spiral)โ, AI ์ด๋ฉ์ผ ๋น์ โ์ฝ๋ผ(Kora)โ, ๋ฐ์คํฌํฑ ์ ๋ฆฌ โ์คํํด(Sparkle)โ, ์์ฑ-ํ ์คํธ ๋ณํ โ๋ชจ๋๋ก๊ทธ(Monologue)โ, ํธ์คํ ์คํํด๋ก โํ๋ฌ์ค์(Plus Ones)โ ๋ฑ ๋ค์ํ AI ์ฑ.
- ํธ๋ ์ด๋: ๊ตฌ๋ ์๋ฅผ ์ํ ๋ผ์ด๋ธ ์คํธ๋ฆผ ๋ฐ ๊ฐ์ข.
โ์๋ธ๋ฆฌโ ํ์ Opus 4.7์ ๋ํ ๋ณด๋ค ์์ธํ๊ณ ๊ตฌ์ฒด์ ์ธ โ๋ฐ์ด๋ธ ์ฒดํฌโ ๊ฒฐ๊ณผ๋ฅผ ์ ๋ฆฌํ์ฌ ๊ณง ๋ฐํํ ์์ ์ด๋ฉฐ, ์ฝ๋ฉ, ๊ธ์ฐ๊ธฐ, ๋์์ธ, ์ ๋ต, ๋น์ฆ๋์ค ๊ตฌ์ถ ๋ฑ ๋ค์ํ ์์ญ์์์ ํ์ฉ๋ฒ์ ์ฌ์ธต์ ์ผ๋ก ๋ค๋ฃฐ ๊ฒ์ด๋ผ๊ณ ๋ฐํ์ต๋๋ค. ๋ํ, ์์ค๋กํฝ ๊ด๊ณ์๋ค์๊ฒ ๋ผ์ด๋ธ ์คํธ๋ฆผ ์ฐธ์ฌ๋ฅผ ์ ์ํ๋ฉฐ ์ง์ ๋ชจ๋ธ์ ๋ํ ์ด์ผ๊ธฐ๋ฅผ ๋๋ ์ํฅ์ ํ๋ช ํ๊ธฐ๋ ํ์ต๋๋ค.
[๊ฒฐ๋ก ] ์์ค๋กํฝ์ Opus 4.7์ ๋ฒค์น๋งํฌ ์์ผ๋ก๋ ๋ถ๋ช ํ ์ฑ๋ฅ ํฅ์์ ๋ณด์์ง๋ง, โ์๋ธ๋ฆฌโ ํ์ ์ค์๊ฐ โ๋ฐ์ด๋ธ ์ฒดํฌโ ๊ฒฐ๊ณผ๋ ๋ชจ๋ธ์ ์ฌ์ฉ ๋ฐฉ์์ ๋ํ ์๋ก์ด ํต์ฐฐ์ ์ ๊ณตํ์ต๋๋ค. ํนํ, ์ด์ ๋ชจ๋ธ๋ณด๋ค โ๋ ๋ฌธ์์ ์ด๊ณ ์ง์ค์ ์ธโ ํน์ฑ์ ๊ธ์ฐ๊ธฐ์ ์์ด์ ํธ ์์ ์์ ๋ค๋ฅธ ๊ฒฐ๊ณผ๋ฌผ์ ๋์ถํ๋ฉฐ, ์ฌ์ฉ์๋ค์ ํ๋กฌํํธ ์์ฑ ๋ฐฉ์์๋ ๋ณํ๋ฅผ ์๊ตฌํ ๊ฒ์ผ๋ก ๋ณด์ ๋๋ค. ์ด๋ AI ๋ชจ๋ธ๋ค์ด ๋จ์ํ ์ฑ๋ฅ ๊ฒฝ์์ ๋์ด ๊ฐ์์ ๊ฐ์ฑ๊ณผ ๊ฐ์ ์ ์ฌ์ ์ํ๋ฉฐ โ์๋ํ ์๋ ดโ์ ๊ธธ์ ๊ฑท๊ณ ์์์ ๋ณด์ฌ์ฃผ๋ ์ค์ํ ์ ํธ์ ๋๋ค. Opus 4.7์ ๋ฑ์ฅ์ ๊ฐ๋ฐ์์ ์ฌ์ฉ์ ๋ชจ๋์๊ฒ ์๋ก์ด ํ๊ตฌ์ ์ ์์ ๊ธฐํ๊ฐ ๋ ๊ฒ์ ๋๋ค.
โ671. Why Has There Been So Little Progress on Alzheimerโs Disease? | Freakonomics Radioโ โ Freakonomics Radio Network ๊ธฐ๋ฐ ๊ธฐ์ฌ ์๋ณธ ์์ ๋ณด๊ธฐ
์์ธ ํ์ด๋จธ ์ฐ๊ตฌ์ ์์ด๋ฒ๋ฆฐ ์์ญ ๋ : ์๋ชป๋ ๊ฐ์ค, ์กฐ์๋ ๋ฐ์ดํฐ, ๊ทธ๋ฆฌ๊ณ ๊ณผํ์ ๋ฌด๊ฒฐ์ฑ์ ์๊ธฐ
์ต๊ทผ ์์ธ ํ์ด๋จธ๋ณ ์ด๊ธฐ ๋จ๊ณ๋ฅผ ํ์งํ ์ ์๋ ๋ ๊ฐ์ง ์๋ก์ด ํ์ก ๊ฒ์ฌ๊ฐ ๋ฏธ๊ตญ ์ํ์์ฝ๊ตญ(FDA)์ ์น์ธ์ ๋ฐ์๋ค๋ ์์์ ๋ง์ ์ด๋ค์๊ฒ ํฌ๋ง์ ์ฃผ์์ต๋๋ค. ์ฆ์์ด ๋ฐํํ๊ธฐ 20๋ ์ ๋ถํฐ ์์ํ ์งํ๋ ์ ์๋ ์์ธ ํ์ด๋จธ๋ณ์ ํน์ฑ์ ๊ณ ๋ คํ ๋, ์กฐ๊ธฐ ์ง๋จ์ ๊ทธ ์ด๋ ์ง๋ณ๋ณด๋ค ์ค์ํฉ๋๋ค. ๊ธฐ์ต ์์ค, ์ธ์ง ๋ฐ ํ๋ ๋ฌธ์ ๋ฑ ์์ธ ํ์ด๋จธ๋ณ์ ์ฆ์์ ๋ฏธ๊ตญ์์๋ง 700๋ง ๋ช ์ด์์ ์ฌ๋๋ค์๊ฒ ์ํฅ์ ๋ฏธ์น๋ฉฐ, ์ด๋ค ๋๋ถ๋ถ์ 65์ธ ์ด์์ ๋๋ค.
์ธ๋ฅ ์ญ์ฌ ์ ๋ฐ์ ๊ฑธ์ณ ๋ ธ๋ ์ธต์ ๊ธฐ์ต ์์ค์ ๊ด์ฐฐ๋์ด ์์ง๋ง, ์ด ์ง๋ณ์ด ๊ณต์์ ์ผ๋ก ๋ฌธ์ํ๋ ๊ฒ์ 1906๋ ๋ ์ผ ์์ฌ ์๋ก์ด์ค ์์ธ ํ์ด๋จธ(Aloois Alzheimer)์ ์ํด์์์ต๋๋ค. ๊ทธ๋ ๊ธฐ์ต ์์ค๊ณผ ํ๊ฐ์ ๊ฒช๋ ์ฌ์ฑ์ ๋๋ฅผ ๋ถ๊ฒํ ๊ฒฐ๊ณผ, ๋๊ฐ ์์ถ๋๊ณ ์๋ง์ ์ํด(tangles)๊ณผ ๊ทธ๊ฐ โ๊ธฐ์ดํ ์นจ์ ๋ฌผ(peculiar deposits)โ์ด๋ผ๊ณ ๋ถ๋ฅธ ๊ฒ์ ๋ฐ๊ฒฌํ์ต๋๋ค. ์ดํ ๊ณผํ์๋ค์ ์ด ์นจ์ ๋ฌผ๊ณผ ์ํด์ ์ ์ฒด๋ฅผ ๋ฐํ๊ธฐ ์ํด ๋์์์ด ๋ ธ๋ ฅํด์์ต๋๋ค.
๋ฏธ๊ตญ ๊ตญ๋ฆฝ๋ณด๊ฑด์(NIH)์ ์์ธ ํ์ด๋จธ ๋ฐ ์น๋งค ์ฐ๊ตฌ์ ์ฐ๊ฐ ์ฝ 40์ต ๋ฌ๋ฌ๋ฅผ ์ง์ถํ๊ณ ์์ต๋๋ค. ์ด๋ 10๋ ์ ์ฝ 10์ต ๋ฌ๋ฌ์์ ํฌ๊ฒ ์ฆ๊ฐํ ๊ธ์ก์ผ๋ก, ์ ์ฐ๊ตฌ ๋ค์์ผ๋ก ๋ง์ ์ฐ๊ตฌ๋น๊ฐ ํฌ์ ๋๊ณ ์์ต๋๋ค. ๋ฏธ๊ตญ์ ๊ณ ๋ น ์ธ๊ตฌ๊ฐ ๋ง๊ณ ๊ณ์ ์ฆ๊ฐํ๊ณ ์๋ค๋ ์ ์ ๊ฐ์ํ๋ฉด ๋น์ฐํ ํฌ์๋ก ๋ณด์ ๋๋ค. ๊ทธ๋ฌ๋ ์์ญ ๋ ๊ฐ ์์ญ์ต ๋ฌ๋ฌ๊ฐ ํฌ์ ๋์์์๋ ๋ถ๊ตฌํ๊ณ , ์์ธ ํ์ด๋จธ๋ณ์ ์ฌ์ ํ ์ฐ๋ฆฌ๋ฅผ ์ข์ ์ํค๋ ๋์น๋ณ์ผ๋ก ๋จ์์์ต๋๋ค. ๊ทธ ์ด์ ๋ ๋ฌด์์ผ๊น์?
์ง๋ฐฐ์ ์ธ ๊ฐ์ค: ์๋ฐ๋ก์ด๋ ์บ์ค์ผ์ด๋ ๊ฐ์ค์ ๊ทธ๋ฆผ์
์์ธ ํ์ด๋จธ ์ฐ๊ตฌ์ ๋๋ถ๋ถ์ ํ ๊ฐ์ง ์ง๋ฐฐ์ ์ธ ์ง๋ณ ์ด๋ก ์ ์ค์ฌ์ผ๋ก ์ด๋ฃจ์ด์ ธ ์์ต๋๋ค. ๋ฐ๋ก โ์๋ฐ๋ก์ด๋ ์บ์ค์ผ์ด๋ ๊ฐ์ค(amyloid cascade hypothesis)โ์ ๋๋ค. ์ด ๊ฐ์ค์ ๋ฐ๋ฅด๋ฉด, ๋ ์ธํฌ ์ธ๋ถ์์ ๋ฒ ํ ์๋ฐ๋ก์ด๋(beta amyloid)๋ผ๋ ๋จ๋ฐฑ์ง์ด ๋ฉ์ด๋ฆฌ๋ฅผ ์ด๋ฃจ์ด ํ๋ผํฌ(plaque)๋ฅผ ํ์ฑํ๋ ๊ฒ์ด ์ง๋ณ์ ์ฃผ์ ์์ธ์ด๋ฉฐ, ์ธํฌ ๋ด๋ถ์ ํ์ฐ(tau) ๋จ๋ฐฑ์ง ์ํด๊ณผ ์ ๊ฒฝ ์ธํฌ ์์์ผ๋ก ์ด์ด์ง๋ค๋ ๊ฒ์ ๋๋ค. ์ฐ๊ตฌ์๋ค์ ์ด ํ๋ผํฌ๊ฐ ์์ธ ํ์ด๋จธ๋ณ์ โ์ฒซ ๋ฒ์งธ ๋๋ฏธ๋ ธโ๋ผ๊ณ ๋ฏฟ์์ผ๋ฉฐ, ์์ญ์ต ๋ฌ๋ฌ์ ์ฐ๊ตฌ ์๊ธ์ด ์ด ๊ฐ์ค์ ๊ธฐ๋ฐ์ผ๋ก ํ ์ฝ๋ฌผ ๊ฐ๋ฐ์ ์ง์ค๋์์ต๋๋ค. ์ฆ, ๋ฒ ํ ์๋ฐ๋ก์ด๋์ ์ถ์ ์ ๋ง๊ฑฐ๋ ์ด๋ฏธ ํ์ฑ๋ ํ๋ผํฌ๋ฅผ ์ ๊ฑฐํ๋ ๋ฐ ์ด์ ์ ๋ง์ถ ๊ฒ์ ๋๋ค.
๊ทธ๋ฌ๋ ๋ง์ ๊ณผํ์๋ค์ ์ด ์ด๋ก ์ ์๋ฌธ์ ์ ๊ธฐํ๊ธฐ ์์ํ์ต๋๋ค. ๋ฐด๋๋นํธ ๋ํ๊ต ๋ฉ๋์ปฌ ์ผํฐ์ ์ ๊ฒฝ๊ณผํ์ ๋งคํ ์๋๊ทธ(Matthew Shrag) ๋ถ๊ต์๋ โ์ด ์ฝ๋ฌผ๋ค๋ก ์๋ฌด๋ ๋์์ง์ง ์๊ณ ์๋ค. ์ด ์ฝ๋ฌผ๋ค์ ๋ค๋ฃจ๋ ๋ชจ๋ ๊ณผํ์, ๋ชจ๋ ์์์๋ ๊ฐ์ ๋ง์ ํ ๊ฒ์ด๋ค. ๊ทธ๋ ์ง ์๋ค๋ฉด ๊ทธ๋ค์ ๊ฑฐ์ง๋ง์ ํ๋ ๊ฒโ์ด๋ผ๊ณ ๋จ์ธํฉ๋๋ค. ๊ทธ๋ ์์ธ ํ์ด๋จธ๋ณ์ด ๋จ์ํ ์ง๋ณ์ด ์๋๋ฉฐ, ํ๊ด ์งํ๊ณผ ๊ฐ์ ๋ค๋ฅธ ์์ธ๋ค์ด ๋ณตํฉ์ ์ผ๋ก ์์ฉํ๋ค๊ณ ์ฃผ์ฅํ๋ฉฐ, ๋จ์ผ ๋ชฉํ ์น๋ฃ๋ฒ์ ํ๊ณ๋ฅผ ์ง์ ํฉ๋๋ค.
ํ์ฌ ์ ๋๋ฆฌ์คํธ ์ฐฐ์ค ํ๋ฌ(Charles Piller)๋ ๊ทธ์ ์ ์ โDoctored: Fraud, Arrogance, and Tragedy in the Quest to Cure Alzheimerโsโ์์ ์ด๋ฌํ ๋ฌธ์ ์ ์ ์ฌ์ธต์ ์ผ๋ก ํํค์ณค์ต๋๋ค. ๊ทธ๋ ์, ๋น๋จ๋ณ, ์ฌ์ฅ๋ณ ๋ฑ ๋ค๋ฅธ ์ง๋ณ์์ ์์ฒญ๋ ๋ฐ์ ์ด ์์๋ ๊ฒ๊ณผ ๋ฌ๋ฆฌ, ์์ธ ํ์ด๋จธ๋ณ์์๋ ์ธ์ง ์ ํ๋ฅผ ๋ฉ์ถ๊ฑฐ๋ ๋๋๋ฆด ์ ์๋ ์น๋ฃ๋ฒ์ด ์ ํ ๋์ค์ง ์์๋ค๋ ์ ์ ์ง์ ํฉ๋๋ค.
์กฐ์๋ ์ด๋ฏธ์ง์ ์ค๋๋ ์ฐ๊ตฌ: ์นด์ฌ๋ฐ ์ฌ์ด์ธ์ค ์ฌ๋ก
๋งคํ ์๋๊ทธ ๋ถ๊ต์๋ ํ๊ณ์์์ ์ฐ๊ตฌ ์ธ์๋ ๋ ๋ฆฝ์ ์ธ ์ฐ๊ตฌ ๋ฌด๊ฒฐ์ฑ ์ปจ์คํดํธ๋ก ํ๋ํ๋ฉฐ ๊ณผํ์ ๋ถ์ ํ์๋ฅผ ํญ๋กํ๋ ๋ฐ ์์ฅ์๊ณ ์์ต๋๋ค. ๊ทธ์ ์ฒซ ๋ฒ์งธ ๋๊ท๋ชจ ์กฐ์ฌ๋ ํ ์ฌ์ค์ ๋ณธ์ฌ๋ฅผ ๋ ์๋ช ๊ณตํ ํ์ฌ ์นด์ฌ๋ฐ ์ฌ์ด์ธ์ค(Cassava Sciences)์ ๊ทธ๋ค์ ์คํ ์ฝ๋ฌผ ์๋ฌดํ๋(simufilam)์ ๋ํ ๊ฒ์ด์์ต๋๋ค. ์ด ์ฝ๋ฌผ์ ์์ธ ํ์ด๋จธ๋ณ์ ์์ธ์ผ๋ก ์ถ์ ๋๋ ํ๋ผ๋ฏผ A(filamin A) ๋จ๋ฐฑ์ง์ ์ค์ ํ์ ๋๋๋ฆฌ๊ณ , ํ์ฐ ๋ฐ ๋ฒ ํ ์๋ฐ๋ก์ด๋ ์ถ์ ์ ํ์ ์ผ๋ก ํ์ฌ ์ผ์ฆ์ ์ค์ด๋ ํจ๊ณผ๊ฐ ์๋ค๊ณ ์ฃผ์ฅ๋์์ต๋๋ค.
ํ์ง๋ง ์ต๋ช ์ ๊ณ ๋ฐ์๋ค์ ์๋ฌดํ๋์ ์ฝ๋ฆฌํ์ ์์ฉ์ด ์ดํด๋์ง ์๊ณ , ๊ธฐ์ด ๊ณผํ์ ๋ฌธ์ ๊ฐ ์์ผ๋ฉฐ, ํนํ ์ด๋ฏธ์ง ๋ฐ์ดํฐ๊ฐ ์กฐ์๋ ๊ฒ์ผ๋ก ๋ณด์ธ๋ค๊ณ ์๋๊ทธ์๊ฒ ์ ๋ณดํ์ต๋๋ค. ์๋๊ทธ๋ ์ด ์ด๋ฏธ์ง๋ฅผ ๋ถ์ํ ๊ฒฐ๊ณผ, ๋ณต์ฌ-๋ถ์ฌ๋ฃ๊ธฐ, ์ค๋ณต, ์ด๋ฏธ์ง ์ผ๋ถ์ ๊ฐ๋ ์กฐ์ ๋ฑ ์๊ฐ์ ์ผ๋ก ์กฐ์๋ ํ์ ์ ๋ฐ๊ฒฌํ์ต๋๋ค. ์ด๋ฌํ ์กฐ์์ ์ฌ์ง ์ ์ผ๋ฃฉ์ ๋๋์ ๊ฐ์ ์ธก์ ๊ฐ์ ๋ณ๊ฒฝํ์ฌ ๊ทผ๋ณธ์ ์ธ ๋ฐ์ดํฐ๋ฅผ ์๊ณกํ ์ ์์ต๋๋ค.
์๋๊ทธ๋ ์ด๋ฌํ ์กฐ์์ด ์นด์ฌ๋ฐ ์ฌ์ด์ธ์ค์ ์ฐ๊ตฌ ํ๋ ฅ์์ธ ๋ด์์๋ฆฝ๋ํ๊ต(City University of New York)์ ๊ต์ ํธ์์ฐ-์ ์(Hoau-Yan Wang)์ ์ฐ๊ตฌ์์ ๊ด๋ฒ์ํ๊ฒ ๋ํ๋ฌ์์ ๋ฐํ๋์ต๋๋ค. ์๋๊ทธ๋ ์ด๋ฌํ ๋ฌธ์ ์ ์ ๋ด์ ๋ณด๊ณ ์๋ฅผ FDA์ NIH์ ์ ์ถํ์ต๋๋ค. FDA์ ๋์์ ์์๋ณด๋ค ๋ฏธ์จ์ ์ด์์ผ๋, ๊ฒฐ๊ตญ 2024๋ ๋ฏธ๊ตญ ๋ฒ๋ฌด๋ถ(DOJ)๋ ์ ๊ต์๋ฅผ NIH ๋ณด์กฐ๊ธ ์ ์ฒญ์์์ ๋ฐ์ดํฐ๋ฅผ ์์กฐํ๊ณ ์กฐ์ํ ํ์๋ก ๊ธฐ์ํ์ต๋๋ค. ์นด์ฌ๋ฐ ์ฌ์ด์ธ์ค ์ญ์ ํฌ์์๋ค์ ์ค๋ํ ํ์๋ก ๋ฏธ๊ตญ ์ฆ๊ถ๊ฑฐ๋์์ํ(SEC)์ 4์ฒ๋ง ๋ฌ๋ฌ๋ฅผ ์ง๋ถํ๊ธฐ๋ก ํฉ์ํ์ผ๋ฉฐ, ์๋ฌดํ๋์ ์์ ์ํ์์ ์์ฝ๋ณด๋ค ์ธ์ง์ ์ด์ ์ด ์๋ค๋ ๊ฒฐ๊ณผ์ ํจ๊ป ๊ฐ๋ฐ์ด ์ค๋จ๋์์ต๋๋ค. ์ดํ ์นด์ฌ๋ฐ ์ฌ์ด์ธ์ค๋ ํ์ฌ ์ด๋ฆ์ ํ๋ผ๋ ํ ๋ผํจํฑ์ค(Filana Therapeutics)๋ก ๋ณ๊ฒฝํ์ต๋๋ค. ์ ๊ต์์ ๋ํ ์ฌ๊ธฐ ํ์๋ 2025๋ ์๋ก์ด ์ฆ๊ฑฐ๊ฐ ์ ์๋๋ฉด์ ์ฒ ํ๋์๊ณ , ๊ทธ์ ๋ณํธ์ธ๋ค์ ๊ทธ์ ๋ฌด์ฃ๋ฅผ ์ฃผ์ฅํ์ต๋๋ค.
์์ธ ํ์ด๋จธ ์ฐ๊ตฌ์ โ์ฑ๋ฐฐโ๋ฅผ ์ฐพ์์: ๋ ์ค๋ค์ ์ ์ฌ ๋ ผ๋ฌธ์ ์ง์ค
์์ธ ํ์ด๋จธ ์ฐ๊ตฌ ๋ถ์ผ์์ ๊ฐ์ฅ ํฐ ๋ ผ๋ ์ค ํ๋๋ 2006๋ ํ์ ์ง โ๋ค์ด์ฒ(Nature)โ์ ์ค๋ฆฐ ์ค๋ฑ ๋ ์ค๋ค(Sylvain Lesnรฉ)์ ์บ๋ฐ ์ ์ฌ(Karen Ashe)์ ๋ ผ๋ฌธ์ ๋๋ค. ์ด ๋ ผ๋ฌธ์ โ๋์ ํน์ ๋ฒ ํ ์๋ฐ๋ก์ด๋ ๋จ๋ฐฑ์ง ์งํฉ์ฒด๊ฐ ๊ธฐ์ต๋ ฅ์ ์์์ํจ๋ค(A Specific Beta Amyloid Protein Assembly in the Brain Impairs Memory)โ๋ ์ ๋ชฉ์ผ๋ก, ์๋ฐ๋ก์ด๋ ๊ฐ์ค์ด ํ ์ฐจ๋ก ๋๊ด์ ๋ถ๋ชํ์ ๋ ๋ฑ์ฅํ์ต๋๋ค. ๋น์ ๋ฒ ํ ์๋ฐ๋ก์ด๋ ์ ๊ฑฐ ๋ฐฑ์ ์์ ์ํ์ด ์คํจํ๋ฉด์ ์๋ฐ๋ก์ด๋ ๊ฐ์ค์ ๋ํ ์๋ฌธ์ด ์ ๊ธฐ๋๋ ์ํฉ์ด์์ต๋๋ค.
๋ ์ค๋ค์ ์ ์ฌ๋ ์ ์ ์ ๋ณํ ์ฅ์ ๋์์ ์๋ฐ๋ก์ด๋ ๋จ๋ฐฑ์ง์ ์ถ์ถํ์ฌ ์ฅ์ ๋์ ์ฃผ์ ํ ๊ฒฐ๊ณผ, ์ธ์ง ๋ฅ๋ ฅ ์ ํ๋ฅผ ๋ณด์๋ค๊ณ ์ฃผ์ฅํ์ต๋๋ค. ์ด๋ค์ ํนํ โ์ฌ๋ฆฌ๊ณ ๋จธ(oligomers)โ๋ผ๊ณ ๋ถ๋ฆฌ๋ ์์ ๋ฒ ํ ์๋ฐ๋ก์ด๋ ๋ฉ์ด๋ฆฌ๊ฐ ์ง๋ณ์ ์ฃผ์ ์์ธ์ด๋ผ๋ โ์์ด์(silver bullet)โ์ ๋ฐ๊ฒฌํ๋ค๊ณ ๋ฐํํ์ต๋๋ค. ์ด ์ฐ๊ตฌ๋ ํ๊ณ์ ์๋ก์ด ์์ ๊ฐ์ ๋ถ์ด๋ฃ์๊ณ , 2022๋ ๊น์ง ์์ธ ํ์ด๋จธ ์ฐ๊ตฌ ์ญ์ฌ์ ๊ฐ์ฅ ๋ง์ด ์ธ์ฉ๋ ๋ ผ๋ฌธ ์ค ํ๋๊ฐ ๋์์ต๋๋ค.
๊ทธ๋ฌ๋ ์๋๊ทธ์ ํ๋ฌ์ ์กฐ์ฌ ๊ฒฐ๊ณผ, ์ด ๋ ผ๋ฌธ์ ์ฌ์ฉ๋ ์จ์คํด ๋ธ๋กฏ(Western blot) ์ด๋ฏธ์ง์์ ์ฌ๊ฐํ ์กฐ์์ด ๋ฐ๊ฒฌ๋์์ต๋๋ค. ์จ์คํด ๋ธ๋กฏ์ ๋จ๋ฐฑ์ง์ ์ข ๋ฅ์ ์์ ํ์ธํ๋ ๋ฐ ์ฌ์ฉ๋๋ ์ผ๋ฐ์ ์ธ ๊ณผํ ๊ธฐ์ ๋ก, ์์ธ ํ์ด๋จธ ์ฐ๊ตฌ์์ ๋งค์ฐ ์ค์ํฉ๋๋ค. ์๋๊ทธ์ ๋ค๋ฅธ ๋ฒ์ํ ์ด๋ฏธ์ง ์ ๋ฌธ๊ฐ๋ค์ ์ด ์ด๋ฏธ์ง๋ค์ด ์คํ ๊ฐ์ค์ ๋ท๋ฐ์นจํ๋๋ก ์กฐ์๋์์์ ๋ฐํ๋์ต๋๋ค.
๋ ์ค๋ค๋ ํ๋ฌ์ ์ทจ์ฌ ์์ฒญ์ ์ํ์ง ์์๊ณ , ์ ์ฌ๋ ๋ํ์ ์กฐ์ธ์ ๋ฐ๋ผ ์ธํฐ๋ทฐ๋ฅผ ๊ฑฐ๋ถํ์ต๋๋ค. ์ด ์ฌ๊ฑด์ NIH์ ๋ณด๊ณ ๋์๊ณ , NIH๋ ์ด๋ฅผ ๋ณด๊ฑด๋ณต์ง๋ถ ์ฐํ ์ฐ๊ตฌ ๋ฌด๊ฒฐ์ฑ ์ฌ๋ฌด๊ตญ(Office of Research Integrity, ORI)์ ํ๋ถํ์ต๋๋ค. ํ์ง๋ง ORI๋ ํด๋น ์ฌ๊ฑด์ ๋ค์ ํผ๊ณ ๋ฐ์์ ์์ ๊ธฐ๊ด์ธ ๋ฏธ๋ค์ํ ๋ํ๊ต์ ๋๊ฒผ์ต๋๋ค. ํ๋ฌ๋ ์ด๋ฌํ ์ ๊ทผ ๋ฐฉ์์ด โ๊ฐ์ฅ ๋ง์ ๊ฒ์ ์๊ณ ๊ฐ์ฅ ์ ์ ๊ฒ์ ์ป์ ์ ์๋ ์กฐ์ง์ ์กฐ์ฌ๋ฅผ ๋งก๊ธฐ๋ ๊ฒโ์ด๋ผ๋ฉฐ ๋นํํ์ต๋๋ค. ๋ฏธ๋ค์ํ ๋ํ๊ต๋ ๋ช ๋ ์ ๊ฑธ์ณ ์กฐ์ฌ๋ฅผ ์งํํ๊ณ , ๊ทธ ๋์ ๋ ์ค๋ค๋ ๋ฌธ์ ์ ์ฐ๊ตฌ ๋ผ์ธ์ ๊ณ์ํ๋ฉฐ ์์ญ๋ง ๋ฌ๋ฌ์ ์ฐ๋ฐฉ ๊ธฐ๊ธ์ ๋ฐ์์ต๋๋ค. ๊ฒฐ๊ตญ ๋ ์ค๋ค๋ 2025๋ ์ ๋ ํด์ง์ ์๋๊ณ ์ฌ์ํ๊ณ , ์ ์ฌ๋ ํด๋น ๋ ผ๋ฌธ์ ์ฒ ํ๋ฅผ ์์ฒญํ์ต๋๋ค.
์์คํ ์ ๋ฌธ์ : ์ํฅ๋ ฅ ์๋ ์ฐ๊ตฌ์์ ๋ถ์ ํ์์ ํ์ ๋ฌธ ์ธ์ฌ
์ฐฐ์ค ํ๋ฌ๋ ์์ธ ํ์ด๋จธ ์ฐ๊ตฌ ๋ถ์ผ์์ ๊ด๋ฒ์ํ ๋ถ์ ํ์๊ฐ ๋ง์ฐํด ์์์ ์ง์ ํ๋ฉฐ, โ์๊ธฐ ์ด์ต์ ์ถ๊ตฌํ๋ ์ฐ๊ตฌ์๋ค, ์ ๋ถ ๊ณต๋ฒ๋ค, ๊ทธ๋ฆฌ๊ณ ๊ธฐ์ ์ ํ์์ด ๊ฒฐํฉ๋ ์๋ชจโ๊ฐ ์์ธ ํ์ด๋จธ ์ฐ๊ตฌ๋ฅผ ์ฅ์ ํ๋ค๊ณ ์ฃผ์ฅํฉ๋๋ค. ๊ทธ๋ ์ด๋ฌํ ์ฃผ์ฅ์ด ์๋ชจ๋ก ์ด ์๋๋ผ, ์์ญ ๋ ๊ฐ ์๋ฐ๋ก์ด๋ ๊ฐ์ค์ ๊ธฐ๋ฐํ ์ฐ๊ตฌ์ ๊ฒฝ๋ ฅ๊ณผ ์๊ธ, ๋ช ์๋ฅผ ๊ฑธ์ด์จ ์ด๋ค์ ์ดํด๊ด๊ณ์์ ๋น๋กฏ๋ ํ์ค์ด๋ผ๊ณ ์ค๋ช ํฉ๋๋ค.
๊ทธ์ ์กฐ์ฌ ๋์ ์ค์๋ ๋ฏธ๊ตญ ๊ตญ๋ฆฝ๋ ธํ์ฐ๊ตฌ์(National Institute on Aging, NIA) ์ ๊ฒฝ๊ณผํ ๋ถ๋ฌธ ์ฑ ์์๋ฅผ ์ญ์ํ๋ ์๋ฆฌ์์ ๋ง์ฌ๋ฆฌ์(Eliezer Masliah)๋ ํฌํจ๋์ด ์์์ต๋๋ค. ๊ทธ๋ ์ฐ๊ฐ 27์ต ๋ฌ๋ฌ ๊ท๋ชจ์ ์์ฐ์ ๊ฐ๋ ํ๋ฉฐ ์ ์ธ๊ณ ์ ๊ฒฝ๊ณผํ ์ฐ๊ตฌ ์์ ๋ฅผ ์ข์ฐํ๋ ๊ฐ์ฅ ์ํฅ๋ ฅ ์๋ ์ธ๋ฌผ ์ค ํ ๋ช ์ด์์ต๋๋ค. ํ๋ฌ๋ ์๋๊ทธ๋ฅผ ํฌํจํ 4๋ช ์ ๋ฒ์ํ ์ด๋ฏธ์ง ์ ๋ฌธ๊ฐ์ ํจ๊ป ๋ง์ฌ๋ฆฌ์์ ๋ ผ๋ฌธ 800ํธ ์ค 132ํธ์์ ์ด๋ฏธ์ง ์กฐ์, ์ค์ฉ, ์ฌ์ฌ์ฉ ๋ฑ ์ฌ๊ฐํ ์ด๋ฏธ์ง ๋ถ๊ท์น์ฑ์ ๋ฐ๊ฒฌํ์ต๋๋ค. ์ด ๋ฌธ์ ๋ค์ ๋ง์ฌ๋ฆฌ์์ 30๋ ๊ฒฝ๋ ฅ์ ๊ฑธ์ณ ๋ํ๋ฌ์ต๋๋ค. ํ๋ฌ๊ฐ ์ด ์ ๋ณด๋ฅผ ๋ง์ฌ๋ฆฌ์์๊ฒ ์ ์ํ์ ๋ ๊ทธ๋ ์๋ตํ์ง ์์๊ณ , NIH๋ ํ๋ฌ์ ๊ธฐ์ฌ๊ฐ ์ถํ๋๊ธฐ ์ง์ ์์ผ ๋ง์ฌ๋ฆฌ์๊ฐ ์ง์ฑ ์์ ๋ฌผ๋ฌ๋ฌ๋ค๊ณ ๋ฐํํ์ต๋๋ค.
ํ๋ฌ๋ ๋ํ FDA์ โํ์ ๋ฌธ ์ธ์ฌ(revolving door)โ ๋ฌธ์ ๋ ๋นํํฉ๋๋ค. ์์ธ ํ์ด๋จธ ์น๋ฃ์ ์น์ธ์ ๊ด์ฌํ๋ FDA ์๋ฃ ๊ฒ์ฌ๊ด 16๋ช ์ค 11๋ช ์ด ํด์ง ํ ๊ท์ ํ๋ ์ ์ฝํ์ฌ์ ์ง์์ด๋ ์ปจ์คํดํธ๋ก ์ผํ๋ค๋ ์ฌ์ค์ ๊ท์ ๊ธฐ๊ด์ ํฌํ(regulatory capture) ๊ฐ๋ฅ์ฑ์ ์์ฌํฉ๋๋ค. ๊ทธ๋ FDA๊ฐ ์น์ธํ๋ ์ํํ๊ณ ๋นํจ๊ณผ์ ์ธ ์๋ฐ๋ก์ด๋ ํญ์ฒด ์ฝ๋ฌผ ์๋์นด๋๋ง(aducanumab, ์ํ๋ช Aduhelm) ์ฌ๋ก๋ฅผ ๋ค๋ฉฐ, FDA ๊ด๊ณ์๊ฐ ์ ์ฝ์ฌ์ ๊ธด๋ฐํ๊ฒ ํ๋ ฅํ์ฌ ์ฝ๋ฌผ ์น์ธ์ ์ถ์งํ ํ ํด๋น ์ ์ฝํ์ฌ๋ก ์ด์งํ๋ ๊ดํ์ด ์ฌ๊ฐํ ๋ฌธ์ ๋ผ๊ณ ์ง์ ํ์ต๋๋ค.
์ค์น๊ณผ ์ ์์ ๋น๊ทน: ์๋๊ทธ์ ๊ทธ๋ฆฌ๋น ๊ต์์ ๊ณ ๋ฐฑ
๋งคํ ์๋๊ทธ์๊ฒ๋ ๊ฐ์ธ์ ์ผ๋ก๋ ๊ณ ํต์ค๋ฌ์ด ๊ฒฝํ์ด ์์์ต๋๋ค. ํ๋ถ ์์ ๊ทธ์ ๋ฉํ ์๋ ์ค์ค๋ง ๊ทธ๋ฆฌ๋น(Othman Ghribi) ๊ต์๋ ์๋๊ทธ์๊ฒ ๋ง์ ๊ฒ์ ๊ฐ๋ฅด์ณ์ฃผ์๊ณ , ์ข์ ์น๊ตฌ์์ต๋๋ค. ์๋๊ทธ๋ ๊ทธ๋ฆฌ๋น ๊ต์์ ํจ๊ป ๊ณ ์ฝ๋ ์คํ ๋กค ์์ด๋ฅผ ์ญ์ทจํ ํ ๋ผ์ ๋์์ ํ๋ผํฌ ์ถ์ ์ ์ฐ๊ตฌํ๋ ๋ ผ๋ฌธ์ ๋ฐํํ๊ธฐ๋ ํ์ต๋๋ค.
์นด์ฌ๋ฐ ์ฌ์ด์ธ์ค ์ฌ๊ฑด ์ดํ, ์๋๊ทธ๋ ์์ ์ ๋ชจ๋ ๊ณผ๊ฑฐ ์ฐ๊ตฌ๊ฐ ๋ถ์๋ ๊ฒ์ด๋ผ๊ณ ์์ํ์ง๋ง, ์์ ์ ์ฐ๊ตฌ์์ ๋ฌธ์ ๊ฐ ๋ฐ๊ฒฌ๋ ์ค์ ๋ชฐ๋์ต๋๋ค. ๊ทธ๋ฌ๋ ๋๊ตฐ๊ฐ ๊ทธ์๊ฒ โ๋น์ ์ด ์ฒ์ ๋ฐํํ ๋ ผ๋ฌธ์ ์ด๋ฏธ์ง๋ค์ ๋ณธ ์ ์ด ์๋๋?โ๊ณ ๋ฌผ์๊ณ , ์๋๊ทธ๋ ํ๋ฏธ๊ฒฝ ์ด๋ฏธ์ง์์ ๋ด๋ฐ๋ค์ด ๋๋ฌด ์ ์ฌํ์ฌ ๋ณต์ฌ-๋ถ์ฌ๋ฃ๊ธฐ๋ ๊ฒ์ฒ๋ผ ๋ณด์ด๋ ์กฐ์ ํ์ ์ ๋ฐ๊ฒฌํ์ต๋๋ค.
์๋๊ทธ๋ ๊ทธ๋ฆฌ๋น ๊ต์์๊ฒ ์ด ์ฌ์ค์ ๋ฌผ์๊ณ , ๊ทธ๋ฆฌ๋น ๊ต์๋ ์์ ์ด ๋ ผ๋ฌธ์ ๊ทธ๋ฆผ์ ๋ด๋นํ์ผ๋ฉฐ, โ๊ธฐ์ด ๊ณผํ์ ์ณ๋ค๊ณ ์๊ฐํ๊ณ , ๋ ๋ช ํํ๊ฒ ๋ณด์ด๋ ค๊ณ ํ์ ๋ฟ ์กฐ์์ด๋ผ๊ธฐ๋ณด๋ค๋ ๊ณผ์ฅ์ ๊ฐ๊น์ ๋คโ๊ณ ์ธ์ ํ์ต๋๋ค. ๊ทธ๋ ๋ํ โ๋ชจ๋๊ฐ ๊ทธ๋ ๊ฒ ํ๋ค๊ณ ์๊ฐํ๋คโ๊ณ ๋งํ์ง๋ง, ์๋๊ทธ๋ ์ด๋ ์ฌ์ค์ด ์๋๋ฉฐ ๊ณผํ์ ๊ธฐ์ค์ ํผ์ํ๋ ๊ฒ์ด๋ผ๊ณ ๊ฐ์กฐํ์ต๋๋ค. ๊ทธ๋ฆฌ๋น ๊ต์๋ ์์น์ฌ ๋๋ฌธ์ ์ฐ๊ตฌ์ค์ ํ์ํ๊ณ ๋ชจ๋ ๊ธฐ๋ก์ ํ๊ธฐํ๋ค๊ณ ์๋๊ทธ์๊ฒ ํธ์ด๋์์ต๋๋ค.
์๋๊ทธ๋ ๊ทธ๋ฆฌ๋น ๊ต์์์ ํตํ๊ฐ ๋งค์ฐ ํ๋ค์๋ค๊ณ ํ๊ณ ํฉ๋๋ค. ๊ทธ๋ฆฌ๋น ๊ต์๋ ์๋๊ทธ๋ฅผ ๋ฌธ์ ๋ก๋ถํฐ ๋ณดํธํ๋ ค ํ์ง๋ง, ๊ฒฐ๊ตญ ๋์ค ์์์ ์์ ์ ํ์๋ฅผ ์ธ์ ํ์ง ์์์ต๋๋ค. ์๋๊ทธ๋ ๊ทธ๋ฆฌ๋น ๊ต์๊ฐ ํ๋ฅญํ ์ค์น์ด์์ง๋ง, ์์ ์ ๊ฒฝํ์ ํตํด ๋ค๋ฅธ ๊ณผํ์๋ค์ด ๊ณผํ์ ์์น์ ์งํค๋ ๊ฒ์ ์ค์์ฑ์ ๊ฐ๋ฅด์น ์ ์์์ ๊ฒ์ด๋ผ๋ฉฐ ์ํ๊น์์ ํํ์ต๋๋ค.
2006๋ ๊ทธ๋ฆฌ๋น, ์๋๊ทธ ๋ฐ ๋ค๋ฅธ ๊ณต๋ ์ ์๋ค์ด ๋ฐํํ๋ ํ ๋ผ ๋ ผ๋ฌธ์ 2023๋ ์ ์ฒ ํ๋์๊ณ , ์ดํ ๊ทธ๋ฆฌ๋น ๊ต์์ ๋ค๋ฅธ ๋ ผ๋ฌธ๋ค๋ ์ฒ ํ๋์์ต๋๋ค. ๊ทธ๋ฆฌ๋น ๊ต์๋ โ์ฐ๊ตฌ๋ ํ์ํฌ์ด๋ฉฐ, ๋๋ ๋ด ์ฐ๊ตฌ์ ๋ฐ์ดํฐ, ์ ์ด๋ ๋๋ถ๋ถ์ ๋ฐ์ดํฐ๋ฅผ ์ง์งํ๋ค. ์กฐ์์ ์์์ง๋ง, ์ฝ๊ฐ์ ๋ค๋ฌ๊ธฐ๋ ์์์ ์ ์๋ค. ๊ทธ๋ฌ๋ ๋๋ ์ฐ๊ตฌ์ค ์ฑ ์์๋ก์ ์ ์ ์ธ ์ฑ ์์ ์ง๋คโ๊ณ ์ด๋ฉ์ผ์ ํตํด ๋ฐํ์ต๋๋ค. ๊ทธ๋ ๋ํ ์๋๊ทธ์ ๋ ผ๋ฌธ์ ์๋ ์ด๋ฏธ์ง๋ค์ ์๋๊ทธ๊ฐ ์์ ์๊ฒ ๋๊ฒจ์ค ๊ฒ์ด๋ฉฐ, ์๋๊ทธ๊ฐ ์กฐ์๋์๋ค๊ณ ๋ฏฟ์๋ค๋ฉด ์๋ฌธ์ ์ ๊ธฐํ ์๊ฐ์ด ๋ง์๋ค๊ณ ์ฃผ์ฅํ์ต๋๋ค.
๊ณผํ์ ๋ํ ์ ๋ขฐ, ๋ค์ ์์์ฌ๋ฆด ์ ์์๊น?
์ฐฐ์ค ํ๋ฌ๋ ์ด๋ฌํ ํญ๋ก๊ฐ ๊ณผํ์ ๋ํ ๋์ค์ ๋ถ์ ์ ์ฌํ์ํฌ ์ ์๋ค๋ ์ฐ๋ ค์ ๋ํด ๊น์ด ๊ณต๊ฐํ๋ค๊ณ ๋งํฉ๋๋ค. ๊ทธ๋ฌ๋ ๊ทธ๋ โ์ ๋ขฐ๋ ์ป์ด์ผ ํ๋ ๊ฒโ์ด๋ผ๋ฉฐ, ๊ณผํ๊ณ ๋ด๋ถ์ ์ค๋งํจ๊ณผ ์์ผํจ, ๊ทธ๋ฆฌ๊ณ ๋์ค์ ํฅํ ์ค๋งํ ํ๋์ ๋ํ ๋นํ์ด ํ์ํ๋ค๊ณ ๊ฐ์กฐํฉ๋๋ค. ์๋ชป๋ ํ์๋ฅผ ๋ฐํ๋ ๊ฒ์ ๊ณผํ์ ๋ฌด๊ฒฐ์ฑ์ ํ๋ณตํ๊ณ ๊ถ๊ทน์ ์ผ๋ก ๊ณผํ์ ๋ํ ์ ๋ขฐ๋ฅผ ๋ค์ ์๋ ๋ฐ ํ์์ ์ด๋ผ๋ ๊ฒ์ ๋๋ค.
์์ธ ํ์ด๋จธ๋ณ ์น๋ฃ์ ๊ฐ๋ฐ์ ์ฌ์ ํ ์ธ๋ฅ๊ฐ ์ง๋ฉดํ ๊ฐ์ฅ ํฐ ๊ณผ์ ์ค ํ๋์ ๋๋ค. ๊ทธ๋ฌ๋ ์ด๋ฌํ ๋ด๋ถ ๊ณ ๋ฐ๊ณผ ํ์ฌ ๋ณด๋๋ ์์ธ ํ์ด๋จธ ์ฐ๊ตฌ๊ฐ ๋์๊ฐ์ผ ํ ๋ฐฉํฅ์ ๋ํ ์ค์ํ ์ง๋ฌธ์ ๋์ง๊ณ ์์ต๋๋ค. ์๋ชป๋ ๊ฐ์ค์ ๋ํ ๋งน๋ชฉ์ ์ธ ์ถ์ข ๊ณผ ๊ณผํ์ ๋ถ์ ํ์๊ฐ ๋ง์ฐํ ํ๊ฒฝ์์๋ ์ง์ ํ ๋ฐ์ ์ด ๋ถ๊ฐ๋ฅํ๋ค๋ ๊ฒ์ ๋ณด์ฌ์ฃผ๊ธฐ ๋๋ฌธ์ ๋๋ค. ์ด์ ์์ธ ํ์ด๋จธ ์น๋ฃ๋ ์ด๋๋ก ํฅํด์ผ ํ ๊น์? ๊ณผํ์ ๋ฌด๊ฒฐ์ฑ์ ํ๋ณตํ๊ณ , ํธํฅ๋์ง ์์ ์ฐ๊ตฌ๋ฅผ ์ง์ํ๋ฉฐ, ๋ค์ํ ๊ฐ์ค์ ํ๊ตฌํ๋ ์ด๋ฆฐ ์์ธ๊ฐ ๊ทธ ์ด๋ ๋๋ณด๋ค ํ์ํ ์์ ์ ๋๋ค.
โScaling Global Organizations in the Age of AI with ServiceNow CEO Bill McDermottโ โ No Priors: AI, Machine Learning, Tech, & Startups ๊ธฐ๋ฐ ๊ธฐ์ฌ ์๋ณธ ์์ ๋ณด๊ธฐ
AI ์๋, ๊ธฐ์ ์ ์์กด ๋ฐฉ์ ์: ์๋น์ค๋์ฐ CEO ๋น ๋งฅ๋๋ฉ, โ์ธ๊ฐ๊ณผ AI์ ์๋์งโ๋ฅผ ๋งํ๋ค
AI ๊ธฐ์ ์ ๋ฌผ๊ฒฐ์ด ๋ชจ๋ ์ฐ์ ๊ณผ ๋น์ฆ๋์ค, ๊ทธ๋ฆฌ๊ณ ๊ฐ์ธ์ ์ถ์ ๋คํ๋ค๊ณ ์์ต๋๋ค. ๊ธ๋ณํ๋ ํ๊ฒฝ ์์์ ๊ธฐ์ ๋ค์ ์์กด๊ณผ ์ฑ์ฅ์ ์ํ ์๋ก์ด ์ ๋ต์ ๋ชจ์ํด์ผ ํ๋ ๊ธฐ๋ก์ ์ ์์ต๋๋ค. ์ด๋ฌํ ๊ฒฉ๋์ ์๊ธฐ์ ์ํฐํ๋ผ์ด์ฆ ๊ธฐ์ ์ ๊ณ์ โ๋ก์คํโ๋ก ๋ถ๋ฆฌ๋ ์๋น์ค๋์ฐ(ServiceNow)์ CEO ๋น ๋งฅ๋๋ฉ(Bill McDermott)์ AI ์๋์ ๋ฆฌ๋์ญ, ๊ธฐ์ ๊ณ ๊ฐ์ด ์ง์ ์ผ๋ก ์ํ๋ ๊ฒ, ๊ทธ๋ฆฌ๊ณ โSAS ์ข ๋ง๋ก (SAS apocalypse)โ์ ๋์ด์ ํ๋ซํผ์ ๊ฐ์น์ ๋ํ ๊น์ด ์๋ ํต์ฐฐ์ ์ ์ํฉ๋๋ค.
๋งฅ๋๋ฉ CEO๋ ์ต๊ทผ โNo Priorsโ ํ์บ์คํธ์ ์ถ์ฐํ์ฌ AI ์๋์ ๊ธ๋ก๋ฒ ๊ธฐ์ ์ ํ์ฅํ๋ ๋ฐฉ๋ฒ์ ๋ํ ์์ ์ ๋น์ ๊ณผ ์๋น์ค๋์ฐ์ ์ ๋ต์ ๊ณต์ ํ์ต๋๋ค. ๊ทธ๋ AI๋ฅผ ํตํด ์ธ๊ฐ์ ์ผ๋ง์ ์ฆ์งํ๊ณ , ์ธ๊ฐ์ ์ธ ์ฐ๊ฒฐ์ ์ค์์ฑ์ ๊ฐ์กฐํ๋ฉฐ, ์ํฌํ๋ก(workflow) ํ๋ซํผ์ด ๊ฐ์ง ๋ ๋ณด์ ์ธ ๊ฐ์น๋ฅผ ์ญ์คํ์ต๋๋ค.
16์ธ ๋ธ๋ฆฌ ์๋ ์ ๋ฆฌ๋์ญ ์ฒ ํ: ๊ณ ๊ฐ ์ค์ฌ๊ณผ ์ธ๊ฐ์ ์ธ ์ฐ๊ฒฐ
๋น ๋งฅ๋๋ฉ CEO์ ๋ฆฌ๋์ญ ์ฒ ํ์ ๊ทธ์ ๋ ํนํ ์ฑ์ฅ ๋ฐฐ๊ฒฝ์์ ๋น๋กฏ๋ฉ๋๋ค. ๊ทธ๋ 16์ธ์ ๋ธ๋ฆฌ(deli, ์๋์์น ๊ฐ๊ฒ)๋ฅผ ์ธ์ํ ๊ฒฝํ์ ํตํด ํ์์ ๊ตํ์ ์ป์๋ค๊ณ ํ๊ณ ํฉ๋๋ค. ๋น์ ์ฌ๋ฌ ์๋ฅด๋ฐ์ดํธ๋ฅผ ํ๋ ๊ทธ๋ ๋ชจ๋ ์ผ์ ๋ธ๋ฆฌ ํ๋๋ก ํตํฉํ๋ฉฐ โ๊ธฐํ(shot)โ๋ฅผ ์ก์๊ณ , โ๊ฒฐ๊ตญ ๊ณ ๊ฐ๋ง์ด ๋น์ ์ ์นํจ๋ฅผ ๊ฒฐ์ ํ๋คโ๋ ๋จ์ํ์ง๋ง ๊ฐ๋ ฅํ ์ง๋ฆฌ๋ฅผ ๊นจ๋ฌ์์ต๋๋ค.
๊ทธ๋ ๋ธ๋ฆฌ์ ์ฃผ์ ๊ณ ๊ฐ์ธต์ธ ๋ธ๋ฃจ์นผ๋ผ ๋ ธ๋์, ๋ ธ์ธ, ๊ทธ๋ฆฌ๊ณ ๋น๋์ค ๊ฒ์์ ์ข์ํ๋ ์์ด๋ค์ ๋์ฆ๋ฅผ ์ ํํ ํ์ ํ๊ณ ๋ง์ถคํ ์๋น์ค๋ฅผ ์ ๊ณตํ์ต๋๋ค. ํนํ, 7-Eleven์ ์ง๋ ์์ ๋ค์ ๊ฐ๊ฒ๋ก ์ค๊ฒ ํ๊ธฐ ์ํด ๋น๋์ค ๊ฒ์์ ์ค์นํ๊ณ , ๊ณ ๊ฐ๋ค์ด โ์ข์ ์์์ ๋จน๊ณ , ์กด์๊ณผ ์กด์ค์ ๋ฐ์ผ๋ฉฐ, ๋น๋์ค ๊ฒ์์ ํ๊ณ ์ถ์ ๋ ์ฐ๋ฆฌ ๊ฐ๊ฒ์ ์จ๋คโ๊ณ ๋งํ๋ ์ผํ๋ ๊ทธ๊ฐ ์ผ๋ง๋ ๊ณ ๊ฐ ์ค์ฌ์ ์ด์๋์ง๋ฅผ ๋ณด์ฌ์ค๋๋ค. ๋งฅ๋๋ฉ์ ์ด ๊ฒฝํ์ ํตํด ํ๋ฃจ 500๋ช ์ ๊ณ ๊ฐ์ ์๋ํ๋ฉฐ ์ป์ ๊ฐ์ฑ ์ง๋ฅ(EQ)์ ๊ฐ์น๋ฅผ ๊ฐ์กฐํ๋ฉฐ, ์ค๋๋ ์ธ์์ ๋ ๋ง์ EQ๊ฐ ํ์ํ๋ค๊ณ ์ญ์คํฉ๋๋ค.
๊ทธ์ ๋ฆฌ๋์ญ์ ์ ๋ก์ค(Xerox)์ ์ ์ฌํ ๋์ ์ผํ์์๋ ๋น์ ๋ฐํฉ๋๋ค. 21์ธ์ ์ ์ ๋์ด์ ๋ฉด์ ์์ โ์๋ฒ์ง๊ป ์ฌ์์ฆ์ ๊ฐ์ง๊ณ ๋์์ค๊ฒ ๋ค๊ณ ์ฝ์ํ๋คโ๋ฉฐ ๋น๋ํ๊ฒ ์์ ์ ์ดํํ๊ณ , ๊ฒฐ๊ตญ ์ฑ์ฉ๋ฉ๋๋ค. 25๋ ํ, ๊ทธ๋ฅผ ์ฑ์ฉํ ์์ฌ๋ โ์ ๋ก์ค์์ ์ ์ฑ ์ ์ด๊ธด ์ ์ผํ ์๊ฐ์ด์๋คโ๊ณ ํ๊ณ ํ๋ฉฐ ๊ทธ์ ์ธ๊ฐ์ ์ธ ๋งค๋ ฅ๊ณผ ์ ์ฌ๋ ฅ์ ๋์ด ํ๊ฐํ์ต๋๋ค. ๋งฅ๋๋ฉ์ ์ด ๊ฒฝํ์ ํตํด โ์ฌ๋๋ค์๊ฒ ์์ฒญ๋ ๊ด์ฌ๊ณผ ์ธ์ฌํ ์ฃผ์๋ฅผ ๊ธฐ์ธ์ฌ์ผ ํ๋คโ๋ฉฐ, AI ์๋์๋ ์ธ๊ฐ์ ์ธ ์ฐ๊ฒฐ๊ณผ ๊ฐ๊ฐ์ธ์ ๊ณ ์ ํ ๋ฅ๋ ฅ์ ๊ธฐ์ตํ๋ ๊ฒ์ด ์ค์ํ๋ค๊ณ ๋งํฉ๋๋ค. โAI๋ ์ฌ๋๋ค์ ์ฌ๊ธฐ๊ณ , ์ธ๊ฐ์ ์ผ๋ง์ ๋ ํฌ๊ฒ ๋ง๋ค์ด์ผ์ง, ์ฐ๋ฆฌ์๊ฒ์ ๊ทธ๊ฒ์ ๋นผ์์์๋ ์ ๋๋คโ๋ ๊ทธ์ ์ฒ ํ์ AI ์๋ ๋ฆฌ๋์ญ์ ํต์ฌ ๋ฉ์์ง๋ก ๋ค๊ฐ์ต๋๋ค.
โSAS ์ข ๋ง๋ก โ์ ๋์ด์: AI ์๋ ์ํฌํ๋ก ํ๋ซํผ์ ์ง์ ํ ๊ฐ์น
์ต๊ทผ ๊ธฐ์ ์ ๊ณ์์๋ ํด๋ผ์ฐ๋ ๊ธฐ๋ฐ ์ํํธ์จ์ด ์๋น์ค(SaaS) ๊ธฐ์ ๋ค์ด AI ๋ชจ๋ธ์ ์ํด ๋์ฒด๋ ๊ฒ์ด๋ผ๋ โSAS ์ข ๋ง๋ก โ์ด ํ์๋๊ณค ํฉ๋๋ค. ํ์ง๋ง ๋งฅ๋๋ฉ CEO๋ ์ด๋ฌํ ์ฃผ์ฅ์ ์ ๋ฉด์ผ๋ก ๋ฐ๋ฐํ๋ฉฐ, ์ํฐํ๋ผ์ด์ฆ ์ํฌํ๋ก ํ๋ซํผ์ ๋ ๋ณด์ ์ธ ๊ฐ์น๋ฅผ ๊ฐ์กฐํฉ๋๋ค.
๊ทธ๋ ๊ฑฐ๋ ์ธ์ด ๋ชจ๋ธ(LLM)๋ง์ผ๋ก ๊ธฐ์ ์ ๋ณต์กํ ํ๋ซํผ์ ๋ณต์ ํ๋ ค๋ ์๋๊ฐ ์ผ๋ง๋ ๋นํจ์จ์ ์ด๊ณ ๋น์ฉ์ด ๋ง์ด ๋๋์ง ๊ตฌ์ฒด์ ์ธ ์๋ฅผ ๋ค์ด ์ค๋ช ํฉ๋๋ค. โ๊ธฐ์กด ์ํฐํ๋ผ์ด์ฆ ํ๋ซํผ์ LLM์ผ๋ก ๋ณต์ ํ๋ ค๋ฉด ๋น์ฉ์ด 10๋ฐฐ ๋ ๋ง์ด ๋ ๋คโ๋ ๊ฒ์ด ๊ทธ์ ์ฃผ์ฅ์ ๋๋ค. ์ฌ๊ธฐ์๋ ํ๋ซํผ ๊ต์ฒด ๋น์ฉ, ํด๋น ์์ ์ ์ํํ ์ธ์ ์๋ณธ ๋น์ฉ, GPU ๊ณต์ฅ ๋ฐ ํ ํฐ ๋น์ฉ ๋ฑ์ด ๋ชจ๋ ํฌํจ๋ฉ๋๋ค.
๋ ์ค์ํ ๋ฌธ์ ๋ โ์ ํ์ฑโ๊ณผ โ์ฑ ์โ์ ๋๋ค. ๋งฅ๋๋ฉ์ โ์ฌ์ ์ ์ด์ํ๋ ์ฌ๋๋ค์ ์ฌ๋์ด ์ค์๋ฅผ ํ๋ค๋ ๊ฒ์ ์ดํดํ์ง๋ง, ์ํํธ์จ์ด๊ฐ ์ค์๋ฅผ ํ๋ ๊ฒ์ ์ ๋ ์ฉ์ํ์ง ์๋๋คโ๊ณ ์ง์ ํฉ๋๋ค. LLM์ ํ๊ฐ(hallucination) ํ์์ด ๋ฐ์ํ ์ ์์ผ๋ฉฐ, ๊ฒฐ์ ๋ก ์ (deterministic)์ด์ง ์์ต๋๋ค. ์ฆ, ๋๊ฐ์ ์ง๋ฌธ์๋ ๋งค๋ฒ ๋์ผํ๊ณ ์ ํํ ๋ต๋ณ์ ๋ณด์ฅํ์ง ๋ชปํ๋ฉฐ, ๊ธฐ์ ๋ด๋ถ์ ์ถ์ ๋ ๋ฐ์ดํฐ์ ๋งฅ๋ฝ์ ์๋ฒฝํ๊ฒ ์ดํดํ์ง ๋ชปํฉ๋๋ค. ๋ฐ๋ฉด ์๋น์ค๋์ฐ์ ๊ฐ์ ์ํฌํ๋ก ํ๋ซํผ์ ์์ญ ๋ ๊ฐ ์ถ์ ๋ ๊ณ ๊ฐ ๋ฐ์ดํฐ์ ๋น์ฆ๋์ค ๋งฅ๋ฝ์ ๊ธฐ๋ฐ์ผ๋ก ์์ธก ๊ฐ๋ฅํ๊ณ ์ฑ ์๊ฐ ์๋ ๊ฒฐ๊ณผ๋ฅผ ์ ๊ณตํฉ๋๋ค.
๊ทธ๋ โAI๋ ์๊ฐํ๊ณ , ์ํฌํ๋ก๋ ํ๋ํ๋ค(AI thinks, workflow acts)โ๋ ๋ช ์ ๋ฅผ ์ ์ํ๋ฉฐ AI์ ์ํฌํ๋ก ํ๋ซํผ์ ์ํธ ๋ณด์์ ์ธ ๊ด๊ณ๋ฅผ ์ค๋ช ํฉ๋๋ค. ์๋ฅผ ๋ค์ด, ๊ธฐ์ ์ ๋ณด์ ๋ฌธ์ ๋ฅผ ํด๊ฒฐํ ๋ LLM์ โ๋จ๊ณ 1, 2, 3์ ๊ณ ๋ คํ๋ผโ๊ณ ์กฐ์ธํ ์ ์์ง๋ง, ์ค์ ์ฌ๋ก๋ฅผ ์ข ๊ฒฐํ๊ณ ๋ถ์ ๊ฐ์ ๋ณต์กํ ๋ฐ์ดํฐ์ ์ํฌํ๋ก๋ฅผ ์กฐ์ ํ๋ฉฐ ๋ฌธ์ ๋ฅผ ํด๊ฒฐํ๋ ๊ฒ์ ์๋น์ค๋์ฐ์ ๊ฐ์ ์ํฌํ๋ก ํ๋ซํผ์ ์ญํ ์ด๋ผ๋ ๊ฒ์ ๋๋ค.
์๋น์ค๋์ฐ๋ ๋ชจ๋ ํ์ดํผ์ค์ผ์ผ๋ฌ(hyperscalers), ์ธ์ด ๋ชจ๋ธ, ๊ทธ๋ฆฌ๊ณ ์์คํ ์ค๋ธ ๋ ์ฝ๋(systems of record)์ ํตํฉ๋๋ โAI ์ปจํธ๋กค ํ์(AI control tower)โ๋ก์ ๊ธฐ์ ์ ๋น์ฆ๋์ค ์ฌ์ฐฝ์กฐ๋ฅผ ๋๋๋ค๋ ๋น์ ์ ๊ฐ์ง๊ณ ์์ต๋๋ค. ๋ํ, ์ฌ์ด๋ฒ ๋ฒ์ฃ๊ฐ ์ธ๊ณ 3์ ๊ฒฝ์ ๊ท๋ชจ์ ์ก๋ฐํ๋ 1์กฐ ๋ฌ๋ฌ ๊ท๋ชจ์ ๋ฌธ์ ๋ก ์ฑ์ฅํจ์ ๋ฐ๋ผ, ๋ณด์(security) ๋ถ์ผ๋ก์ ํ์ฅ(Armis ์ธ์)๊ณผ ์ด์ ๊ธฐ์ (OT, operational technology) ๊ด๋ฆฌ ์ญ๋ ๊ฐํ๋ ๊ธฐ์ ์๊ฒ ์ ์ฒด์ ์ธ IT ๋ฐ OT ํ๊ฒฝ์ ๋ํ ๊ฐ์์ฑ๊ณผ ํต์ ๋ฅผ ์ ๊ณตํฉ๋๋ค. ๋งฅ๋๋ฉ์ โ์ฌ๋, ์ฅ์, ์ฌ๋ฌผ๋ฟ๋ง ์๋๋ผ ๊ฑด๋ฌผ๊ณผ ๊ทธ ์์ ์๋ ๋ชจ๋ ๊ฒ์ ํฌํจํ๋ ์ค์ํ ์ธํ๋ผ๋ฅผ ํ๋์ ์๋ํ๋ AI ํ๋ซํผ์์ ๊ด๋ฆฌํ๋ ๊ฒ์ ๋งค์ฐ ๊ฐ๋ ฅํ ๋น์ โ์ด๋ผ๊ณ ๊ฐ์กฐํฉ๋๋ค.
ํ๋ซํผ ๋น์ฆ๋์ค์ ์ ์์ ์๋น์ค๋์ฐ์ ์ฐจ๋ณ์
๋งฅ๋๋ฉ CEO๋ SAP์ ์๋น์ค๋์ฐ๋ผ๋ ๋ ๊ฐ์ ๋ช ํํ ํ๋ซํผ ๋น์ฆ๋์ค๋ฅผ ์ด๋์ด์จ ๊ฒฝํ์ ๋ฐํ์ผ๋ก โํ๋ซํผ ๋น์ฆ๋์คโ์ ์ ์๋ฅผ ๋ช ํํ ํฉ๋๋ค. ๊ทธ์๊ฒ ํ๋ซํผ์ ๋จ์ํ ํ๋์ ๊ธฐ๋ฅ์ ์ ๊ณตํ๋ ์์คํ ์ด ์๋๋ผ, โ๋ฏธ์ ํฌ๋ฆฌํฐ์ปฌ(mission-critical)ํ๋ฉฐ, ์ ํ ๋น์ฉ(switching costs)์ด ๋งค์ฐ ๋๊ณ , ๋ค๋ฅธ ๋ชจ๋ ์์คํ ์ค๋ธ ๋ ์ฝ๋๋ฅผ ํตํฉํ๋ ํ๋ธโ๋ฅผ ์๋ฏธํฉ๋๋ค.
์๋น์ค๋์ฐ๋ IT ์๋น์ค ๋ฐ์คํฌ๋ก ์์ํ์ง๋ง, ์ด์ ๋ ๊ธฐ์ ์ ๋ชจ๋ ๋ถ์(IT, ์ธ์ฌ, ๊ณ ๊ฐ ๊ด๊ณ ๊ด๋ฆฌ, ๋ณด์ ๋ฑ)๋ฅผ ์์ฐ๋ฅด๋ ์ ์ฌ์ ์ํฌํ๋ก๋ฅผ ์ ๊ณตํฉ๋๋ค. ์๋น์ค๋์ฐ์ ์ธ๋ฉ๋ชจ๋ฆฌ ์คํ์์ค ๋ฐ์ดํฐ ํ๋ซํผ์ ๋ฐฉ๋ํ ๋ฐ์ดํฐ๋ฅผ ๋น ๋ฅด๊ฒ ์ฒ๋ฆฌํ๊ณ , ๋ค๋ฅธ ์์คํ ์ ์๋ ๋ฐ์ดํฐ๋ฅผ โ์ ๋ก ์นดํผ(zero-copy)โ ๋ฐฉ์์ผ๋ก ํ์ฉํ์ฌ ์ค์๊ฐ์ผ๋ก ๋น์ฆ๋์ค ํ๋ก์ธ์ค๋ฅผ ๋ณ๊ฒฝํ๊ณ ์คํํ ์ ์๊ฒ ํฉ๋๋ค.
ํนํ AI ์๋์ ์๋น์ค๋์ฐ์ ๊ฐ์ ์ ๋์ฑ ๋ถ๊ฐ๋ฉ๋๋ค. โ์๋ก์ด ๋น์ฆ๋์ค ํ๋ก์ธ์ค๋ฅผ AI ๋๊ตฌ๋ฅผ ์ฌ์ฉํ์ฌ ์ฆ์ ๋ณ๊ฒฝํ ์ ์๋คโ๋ ์ ์ ๊ธฐ์ ์ด ๋ณํ์ ๋น ๋ฅด๊ฒ ๋์ํ๊ณ ํ์ ํ ์ ์๋ ๊ธฐ๋ฐ์ ์ ๊ณตํฉ๋๋ค. ๋งฅ๋๋ฉ์ โ๊ธฐ์ ๊ณ ๊ฐ๋ค์ ๊ฐ์น๋ฅผ ๋งค์ฐ ๋น ๋ฅด๊ฒ ์ป๊ธฐ๋ฅผ ์ํ๋คโ๋ฉฐ, ์๋น์ค๋์ฐ๊ฐ ์ฃผ์ ๊ณ ๊ฐ๋ค์๊ฒ 30์ผ ์ด๋ด์ ์์คํ ์ ๊ตฌ์ถ(go live)ํ ์ ์๊ฒ ํ๋ ์๋๋ AI ์๋์ ํต์ฌ ๊ฒฝ์๋ ฅ์ด๋ผ๊ณ ์ค๋ช ํฉ๋๋ค. ์ด๋ ๊ณผ๊ฑฐ ์๊ฐ์ ๋๋ ์๋ ์ด ๊ฑธ๋ฆฌ๋ ๋ณต์กํ ํ๋ก์ ํธ๋ฅผ ํจ์ฌ ๋น ๋ฅด๊ณ ํจ์จ์ ์ผ๋ก ์งํํ ์ ์๊ฒ ํ์ฌ ๊ณ ๊ฐ์ ํฌ์ ์์ต๋ฅ (ROI)์ ๊ทน๋ํํฉ๋๋ค.
AI๊ฐ ๋ฐ๊พธ๋ ๊ธฐ์ ์ ๋ฏธ๋: ์์ฐ์ฑ๊ณผ ์ธ๋ ฅ ๊ตฌ์กฐ์ ๋ณํ
AI๋ ๊ธฐ์ ์ ์์ฐ์ฑ์ ๊ทน๋ํํ๊ณ ์ธ๋ ฅ ๊ตฌ์กฐ์๋ ๊ทผ๋ณธ์ ์ธ ๋ณํ๋ฅผ ๊ฐ์ ธ์ฌ ๊ฒ์ ๋๋ค. ๋งฅ๋๋ฉ CEO๋ ํฅํ 5๋ ๋ด ์๋น์ค๋์ฐ์ ์ง์ ๊ตฌ์ฑ์ด ๋ฌ๋ผ์ง ๊ฒ์ด๋ผ๊ณ ์์ธกํฉ๋๋ค. โ์์ฆ ์ธ๋ ฅ(net new added headcount)์ ๊ทน์ ์ผ๋ก ์ค์ด๋ค ๊ฒโ์ด๋ฉฐ, ์ด๋ โ์์ด์ ํธ(agents)โ๊ฐ ์์ฒญ๋ ์ ๋ฌด๋์ ๋ด๋นํ๊ฒ ๋๊ธฐ ๋๋ฌธ์ ๋๋ค. ์๋ฅผ ๋ค์ด, ์๋น์ค๋์ฐ ๊ณ ๊ฐ ์๋น์ค ์ฌ๋ก์ 90%๋ ์ด๋ฏธ ์์ด์ ํธ๊ฐ ์ฒ๋ฆฌํ๊ณ ์์ผ๋ฉฐ, ์ด๋ ์ธ๊ฐ ์ง์์ด โ์ ์ ์ ์ธ ์ธ๋ถ ์์ โ ๋์ โ๋นํ์ ์ฌ๊ณ ์ ํ๋จโ์ด ํ์ํ ์ ๋ฌด์ ์ง์คํ ์ ์๊ฒ ํฉ๋๋ค.
์ด๋ฌํ ๋ณํ๋ ์ง์๋ค์๊ฒ๋ ์๋ก์ด ๊ธฐํ์ด์ ๋์ ์ผ๋ก ๋ค๊ฐ์ต๋๋ค. ์์ด์ ํธ๊ฐ ์ฝ๊ฒ ๋ณต์ ํ ์ ์๋ ์ฐจ๋ณํ๋ ๊ธฐ์ ์ ๊ฐ์ถ๋ ๊ฒ์ด ์ค์ํด์ง ๊ฒ์ ๋๋ค. ๋งฅ๋๋ฉ์ ํฅํ ๋ช ๋ ์์ 22์ต ๊ฐ์ ์์ด์ ํธ๊ฐ ๋ ธ๋ ์์ฅ์ ์ง์ ํ ๊ฒ์ด๋ฉฐ, ์ด๋ ์ฌ๋๋ณด๋ค ์์ด์ ํธ๊ฐ ๋ ๋ง์์ง ์ ์์์ ์๋ฏธํ๋ค๊ณ ๊ฒฝ๊ณ ํฉ๋๋ค. ๊ธฐ์ ๋ค์ ์ธ๋ ฅ์ ๋๋ฆด ๋ โ๋ช ํํ ROI(ํฌ์ ์์ต๋ฅ )โ๋ฅผ ์๊ตฌํ๊ฒ ๋ ๊ฒ์ ๋๋ค. ํนํ ์ฝ๋ก๋19 ํฌ๋ฐ๋ฏน ๊ธฐ๊ฐ ๋์ ์๊ฒฉ ์ฑ์ฉ์ ํตํด ๊ธฐ์ ๋ค์ด ๋ชธ์ง์ ๋ถ๋ ธ๋ ์ํฉ์ ๋๋์๋ณด๋ฉฐ, AI ์๋์๋ ๋์ฑ ๋ ๋ ตํ๊ณ ์ค๋งํธํ ์กฐ์ง์ผ๋ก์ ๋ณํ๋ฅผ ์ถ๊ตฌํ๊ฒ ๋ ๊ฒ์ด๋ผ๊ณ ์์ธกํฉ๋๋ค.
AI ๋์ ์ ํ์ฃผ์์ ์๋น์ค๋์ฐ์ ๋น์
ํ์ฌ AI ๋์ ์ ํ์ฃผ์๋ ์ฐ์ ๊ณผ ์ง์ญ์ ๋ฐ๋ผ ๋งค์ฐ ๋ค์ํฉ๋๋ค. ๋งฅ๋๋ฉ์ ๋ธ๋ผ์ง์ ๊ฒฝ์ฐ 11%์ ๊ธฐ์ ๋ง์ด AI ์คํ ๋จ๊ณ๋ฅผ ๋์ด์ฐ๋ค๊ณ ์ธ๊ธํ๋ฉฐ, ๋ง์ ๊ธฐ์ ์ด ์์ง ์ด๊ธฐ ๋จ๊ณ์ ๋จธ๋ฌผ๋ฌ ์์์ ๋ณด์ฌ์ค๋๋ค. ๊ณต๊ณต ๋ถ๋ฌธ์ ๋ณต์ก์ฑ ํตํฉ๊ณผ ์ฌ๊ธฐ ๋ฐฉ์ง์ ์ฃผ๋ ฅํ๊ณ , ํฌ์ค์ผ์ด๋ ํ์ ๊ฒฝํ ํ๋ํ์ ์ง์คํ๋ ๋ฐ๋ฉด, ๊ธ์ต ์๋น์ค ๊ธฐ์ ๋ค์ AI๋ฅผ ํตํด ์ธ๋ ฅ ๊ด๋ฆฌ๋ฅผ ์ต์ ํํ๊ณ โ๋ ์ ์ ๊ฒ์ผ๋ก ๋ ๋ง์ ๊ฒ์ ํด๋ด๋โ ๋ฐฉํฅ์ผ๋ก ๋น์ฆ๋์ค ๋ชจ๋ธ์ ๊ทผ๋ณธ์ ์ผ๋ก ์ฌ๊ณ ํ๊ณ ์์ต๋๋ค.
๋ฏธ๊ตญ์ AI ์๋น ๋ฐ ์ฑํ์์ ์ ๋๋ฅผ ๋ฌ๋ฆฌ๊ณ ์์ผ๋ฉฐ, โ์ต์ด ์ดํ์(first movers)โ๋ค์ ์ด๋ฏธ AI๋ฅผ ํตํด ๋น์ฆ๋์ค ๋ชจ๋ธ์ ๋ณํ์ํค๊ณ ์ธ๋ ฅ ์ต์ ํ๋ฅผ ์ถ์งํ๋ ๋ฐ ์ ๊ทน์ ์ผ๋ก ๋์๊ณ ์์ต๋๋ค. ๋งฅ๋๋ฉ์ ์ด๋ฌํ ๋ณํ๊ฐ ๊ธฐ์ ์ ๊ถ๊ทน์ ์ธ ์์ต์ฑ, ์ง์๋น ๋งค์ถ, ๊ทธ๋ฆฌ๊ณ ํ์ฅ๋๋ ๋ง์ง์ผ๋ก ์ด์ด์ง ๊ฒ์ด๋ผ๋ ๋์ ๊ธฐ๋๋ฅผ ๊ฐ์ง๊ณ ์์ต๋๋ค.
์๋น์ค๋์ฐ์ ๋น์ ์ ๊ณ ๊ฐ ๊ด๊ณ์ ๋ชจ๋ ๋ฉด์์ โ๋ ๋ง์ด ์๊ณ , ๋ ๋ง์ด ์ ๊ฒฝ ์ฐ๊ณ , ๋ ๋ง์ด ํด๋ผ ์ ์๋โ ๊ธฐ์ ์ด ๋๋ ๊ฒ์ ๋๋ค. AI๋ฅผ ํตํด ๊ธฐ์ ์ด์ ๋ฐฉ์์ ๋ํ ๋๋ผ์ด ์ค์๊ฐ ํต์ฐฐ๋ ฅ๊ณผ ๊ฐ์์ฑ์ ์ ๊ณตํ๊ณ , ๊ณ ๊ฐ์ด ๋ ๋์ ๋ฐฉ์์ผ๋ก ๋น์ฆ๋์ค๋ฅผ ์ด์ํ ์ ์๋๋ก ๋๋ ๊ฒ์ด ๊ทธ๋ค์ ๋ชฉํ์ ๋๋ค.
๋น ๋งฅ๋๋ฉ CEO์ ํต์ฐฐ์ AI ์๋๋ฅผ ๋จ์ํ ๊ธฐ์ ์ ๋ณํ๋ก๋ง ๋ณผ ๊ฒ์ด ์๋๋ผ, ์ธ๊ฐ์ ์ ์ฌ๋ ฅ์ ๊ทน๋ํํ๊ณ ๊ธฐ์ ์ ๋ณธ์ง์ ์ธ ๊ฐ์น๋ฅผ ์ฌ์ ์ํ๋ ๊ธฐํ๋ก ์ผ์์ผ ํ๋ค๋ ๊ฐ๋ ฅํ ๋ฉ์์ง๋ฅผ ์ ๋ฌํฉ๋๋ค. โSAS ์ข ๋ง๋ก โ๊ณผ ๊ฐ์ ๋ง์ฐํ ๋๋ ค์์ ๋์ด, ์ํฌํ๋ก ํ๋ซํผ์ ๊ฒฌ๊ณ ํจ๊ณผ AI์ ํ์ ์ ์ธ ์ ์ฌ๋ ฅ์ ๊ฒฐํฉํ์ฌ ์ธ๊ฐ ์ค์ฌ์ ๋ฏธ๋๋ฅผ ๋ง๋ค์ด๊ฐ๋ ์๋น์ค๋์ฐ์ ํ๋ณด๊ฐ ์ฃผ๋ชฉ๋ฉ๋๋ค. ๊ฒฐ๊ตญ AI๋ ์ธ๊ฐ์ ์ผ๋ง์ ์ฆ์งํ๊ณ , ์ธ๊ฐ์ ์ธ ์ฐ๊ฒฐ์ ๋์ฑ ์ค์ํ๊ฒ ๋ง๋๋ ๋๊ตฌ๋ก ํ์ฉ๋ ๋ ์ง์ ํ ๊ฐ์น๋ฅผ ๋ฐํํ ๊ฒ์ ๋๋ค.
โWhat to Know About the Israel-Lebanon Cease-Fire, and R.F.K. Jr.โs Shifting Tone on Vaccinesโ โ New York Times Podcasts ๊ธฐ๋ฐ ๊ธฐ์ฌ ์๋ณธ ์์ ๋ณด๊ธฐ
์ค๋์ ๋ถ์ํ ํํ, ๋ฐฑ์ ์ ์ฅ ์ ํ, ๊ทธ๋ฆฌ๊ณ AI ์๋์ ๊ทธ๋ฆผ์: ๋ด์ํ์์ค ํค๋๋ผ์ธ ๋ถ์
์ค๋ ๋ด์ํ์์ค ํค๋๋ผ์ธ์ ๊ตญ์ ์ ์ธ์ ๋ถ์ํ ํํ๋ถํฐ ๊ตญ๋ด ๋ณด๊ฑด ์ ์ฑ ์ ๋ฏธ๋ฌํ ๋ณํ, ๊ทธ๋ฆฌ๊ณ ๊ธ๋ณํ๋ ๊ธฐ์ ์๋์ ์๋ก์ด ๋ ผ์๊น์ง, ๋ค์ฑ๋ก์ด ์์๋ค์ ์ ํฉ๋๋ค. ์ด์ค๋ผ์-๋ ๋ฐ๋ ผ ํด์ ์ ๋ณต์กํ ํ์ค, ๋ฏธ๊ตญ-์ด๋ ๊ฐ์ ๊ณ ์กฐ๋๋ ๊ธด์ฅ, ๋ก๋ฒํธ F. ์ผ๋ค๋ ์ฃผ๋์ด(R.F.K. Jr.)์ ๋ฐฑ์ ๊ด๋ จ ๋ฐ์ธ ๋ณํ, ๊ทธ๋ฆฌ๊ณ ์ธ๊ณต์ง๋ฅ(AI) ๋ฐ์ดํฐ์ผํฐ ๊ฑด์ค์ ๋๋ฌ์ผ ๋ ผ๋๊น์ง, ์ฃผ์ ์ด์๋ค์ ์ฌ์ธต์ ์ผ๋ก ๋ค์ฌ๋ค๋ด ๋๋ค.
์ด์ค๋ผ์-๋ ๋ฐ๋ ผ, ๋ถ์ํ 10์ผ ํด์ ๋์
๋ ๋ฐ๋ ผ์์๋ ์ง๋๋ฐค ๋ฆ๊ฒ ์ด์ค๋ผ์๊ตฐ๊ณผ ์ด๋์ ์ง์์ ๋ฐ๋ ๋ฌด์ฅ ๋จ์ฒด ํค์ฆ๋ณผ๋ผ(Hezbollah) ๊ฐ์ 10์ผ๊ฐ ํด์ ์ด ๋ฐํจ๋์์ต๋๋ค. ๋ด์ํ์์ค ๊ธฐ์ ํฌ๋ฆฌ์คํฐ๋ ๊ณจ๋๋ด(Christina Goldbomb)์ ๋ฒ ์ด๋ฃจํธ์์ ๋จ๋ถ ๋ ๋ฐ๋ ผ์ ์๋ ์ฃผ์ ๊ณ ์๋๋ก์์ ํ์ง ์ํฉ์ ์ ํ์ต๋๋ค. ํด์ ๋ฐํ ์ดํ ๋จ๋ถ ์ง์ญ์์ ํผ๋ ์๋ ์์ฒ ๊ฐ๊ตฌ๊ฐ ์ง์ผ๋ก ๋์๊ฐ๋ ค ํ๋ฉด์ ๊ทน์ฌํ ๊ตํต ์ฒด์ฆ์ด ๋ฐ์ํ์ต๋๋ค. ์ด์ค๋ผ์๊ตฐ์ ํญ๊ฒฉ์ผ๋ก ์ฃผ์ ๋ค๋ฆฌ ๋ ๊ฐ๊ฐ ํ๊ดด๋ ํ์, ํ์ฌ๋ ์์๋ก ๋ง๋ค์ด์ง ๋นํฌ์ฅ ๋๋ก ํ๋๋ง ์ด์ฉํ ์ ์์ด ์ฐจ๋๋ค์ด ํ ๋์ฉ ๊ฑด๋๊ณ ์์ต๋๋ค. ์ด๋ก ์ธํด ๋ง์ ์ด๋ค์ด ๋ฐค๋ฆ๊ฒ๊น์ง ๋๋ก์ ๋จธ๋ฌผ๋ฌ์ผ ํ ์ํฉ์ ๋๋ค.
ํ์ง ์ฃผ๋ฏผ๋ค์ ํด์ ์ด ๊ณผ์ฐ ์ง์๋ ์ง, ํน์ ์์ผ๋ก ์ด๋ค ์ผ์ด ๋ฒ์ด์ง์ง ํ์ ํ์ง ๋ชปํ๋ ๋ถ์ํ ๊ธฐ์์ ๋ด๋น์น๊ณ ์์ต๋๋ค. ๊ทธ๋ฌ๋ ์ง๋ ํ ๋ฌ ๋ฐ ๋์์ ์ ์ ์ดํ ๋จ์ ๊ฒ์ด ๋ฌด์์ธ์ง ํ์ธํ๊ธฐ ์ํด ๊ณ ํฅ์ผ๋ก ๋์๊ฐ๊ณ ์ ํ๋ ๊ฐํ ์ด๋ง์ ๋ณด์ด๊ณ ์์ต๋๋ค. ๋ ๋ฐ๋ ผ ๋น๊ตญ์ ๋ฐ๋ฅด๋ฉด 3์ ์ดํ ์ด์ค๋ผ์์ ํค์ฆ๋ณผ๋ผ์ ๋ํ ๊ตฐ์ฌ ์์ ์ผ๋ก 2,100๋ช ์ด์์ด ์ฌ๋งํ๊ณ 100๋ง ๋ช ์ด์์ด ํผ๋๊ธธ์ ์ฌ๋์ต๋๋ค. ์ด๋ฒ ํด์ ์ ๋ฏธ๊ตญ์ด ์ด๋๊ณผ์ ํด์ ์ ์ถ์งํ๋ ๋ฐ ์ํ์ด ๋ ์ ์์๋ ์ํฉ์์, ๋ฏธ๊ตญ์ด ๊ฐ๋ ฅํ ์ถ์งํ ๊ฒฐ๊ณผ์ ๋๋ค.
์ด๋ฒ ํฉ์๋ ๊ธฐ์ ์ ์ผ๋ก ์ด์ค๋ผ์๊ณผ ๋ ๋ฐ๋ ผ ์ ๋ถ ๊ฐ์ ์ด๋ฃจ์ด์ก์ง๋ง, ๋ ๋ฐ๋ ผ ์ ๋ถ๋ ํค์ฆ๋ณผ๋ผ๋ฅผ ์ง์ ํต์ ํ์ง ๋ชปํ๋ค๋ ์ ์ด ๋ณ์์ ๋๋ค. ํค์ฆ๋ณผ๋ผ๋ ์ด์ ์ ๋ ๋ฐ๋ ผ ์ ๋ถ๊ฐ ํ์ํ ์ผ๋ถ ํฉ์๋ฅผ ์ค์ํ์ง๋ง, ์ด๋ฒ ํด์ ์์ฉ ์ฌ๋ถ์ ๋ํด์๋ โ์ํฉ ์ ๊ฐ์ ๋ฐ๋ผ ํ๋ํ ๊ฒโ์ด๋ผ๊ณ ๋ง ๋ฐํ ๋ถํ์ค์ฑ์ ๋จ๊ฒผ์ต๋๋ค. ์ด์ค๋ผ์ ์ธก์์๋ ๋ฒ ๋๋ฏผ ๋คํ๋ํ(Benjamin Netanyahu) ์ด๋ฆฌ๊ฐ ํด์ ๊ธฐ๊ฐ์๋ ์ด์ค๋ผ์๊ตฐ์ด ๋จ๋ถ ๋ ๋ฐ๋ ผ์ ์ฃผ๋ํ ๊ฒ์ด๋ฉฐ, ์ ๋ นํ ์ง์ญ์ โํ๋๋ ์๋ณด ์ง๋โ๋ก ์นญํ๋ฉฐ โ์ฐ๋ฆฌ๋ ๋ ๋์ง ์์ ๊ฒโ์ด๋ผ๊ณ ๋ชป ๋ฐ์ ๊ธด์ฅ๊ฐ์ ๋ํ์ต๋๋ค.
๋ฏธ๊ตญ-์ด๋ ์ ์: ๊ณ ์กฐ๋๋ ์๋ฐ๊ณผ ๊ฒฝ์ ์ ํ์ฅ
์ด๋๊ณผ์ ์ ์์ ๋๋ฌ์ผ ์ธ ๊ฐ์ง ์ฃผ์ ์์๋ ์ ํด์ก์ต๋๋ค. ์ฒซ์งธ, ํผํธ ํฅ์คํ์ค(Pete Hegsath) ๊ตญ๋ฐฉ์ฅ๊ด์ ์ด๋์ ์ ์ ๊ถ์ด ์๋ชป๋ ์ ํ์ ํ๊ณ ํฉ์์ ๋์ํ์ง ์์ ๊ฒฝ์ฐ, ์ ํฌ ์์ ์ ์ฌ๊ฐํ ์ค๋น๊ฐ ๋์ด ์๋ค๊ณ ์ํํ์ต๋๋ค. ๊ทธ๋ ์ด๋์ ๋ฐ์ ์๋ฅผ ํ์ ์ผ๋ก ์ผ์ ์ ์๋ค๊ณ ๊ฒฝ๊ณ ํ๋๋ฐ, ๊ตญ์ ๋ฒ์ ์๋์ง ๊ธฐ๋ฐ ์์ค์ ๋ํ ์๋์ ์ธ ๊ณต๊ฒฉ์ ์ ์ ๋ฒ์ฃ์ ํด๋นํ ์ ์์ต๋๋ค. ๊ตญ๋ฐฉ๋ถ๋ โ๊ทธ๋ ๊ฒ ํ๊ณ ์ถ์ง๋ ์์ง๋ง, ๋ํต๋ น์ ๋ช ๋ น์ ๋ฐ๋ผ ์ฆ์ ์คํํ ์ค๋น๊ฐ ๋์ด ์๋คโ๊ณ ๋ฐํ์ต๋๋ค. ์ง๋ ์ฃผ๋ง ๋ฏธ๊ตญ๊ณผ ์ด๋ ๊ฐ์ ํ๋ด์ด ๊ฒฐ๋ ฌ๋ ์ดํ, ํธ๋ผํ ํ์ ๋ถ๋ ์ด๋์ ๋ํ ์๋ฐ ์์๋ฅผ ๋์ด๊ณ ์์ผ๋ฉฐ, ์ด๋์ ์ค๊ฐ๋ ๋ชจ๋ ์ ๋ฐ์ ๋ํ ํด์ ๋ด์ ์กฐ์น๊น์ง ํฌํจํ๊ณ ์์ต๋๋ค.
๋์งธ, ๋ฐฑ์ ๊ด ์์ฐ๊ตญ์ฅ์ ์ด์ ์ํ ์ฒญ๋ฌธํ์์ ์ ์ ๋น์ฉ์ ๋ํ ์ง๋ฌธ์ ํํผํ์ต๋๋ค. ๊ทธ๋ ๋น์ฉ์ด ๋งค์ผ ๋ณ๋ํ๊ธฐ ๋๋ฌธ์ ์ ํํ ์์น๋ฅผ ์ ๊ณตํ ์ ์๋ค๊ณ ๋งํ์ผ๋ฉฐ, ์ฌ์ง์ด ๋๋ต์ ์ธ ๋ฒ์์กฐ์ฐจ ๋ฐํ๊ธฐ๋ฅผ ๊ฑฐ๋ถํ์ต๋๋ค. ์ด๋ฌ ์ด ๋ถ์๊ฐ๋ค์ ๋ฏธ๊ตญ์ด ์ด๋๊ณผ์ ๊ตฐ์ฌ ์์ ์ ํ๋ฃจ ์ฝ 10์ต ๋ฌ๋ฌ๋ฅผ ์ง์ถํ๊ณ ์๋ค๊ณ ์ถ์ ํ์ต๋๋ค.
์ ์งธ, ์ด์ ๋ผ์ค๋ฒ ์ด๊ฑฐ์ค ์ฐ์ค์์ ํธ๋ผํ ๋ํต๋ น์ ์ ์์ผ๋ก ์ธํ ๊ธ๋ก๋ฒ ์์ ๊ณต๊ธ ์ฐจ์ง๋ก ์ ๊ฐ๊ฐ ์์นํ๋ ๊ฒ์ ๋์๋กญ์ง ์๊ฒ ์ฌ๊ฒผ์ต๋๋ค. AAA์ ๋ฐ๋ฅด๋ฉด ํ์ฌ ๋ฏธ๊ตญ ์ด์ ์๋ค์ ํ๊ท ํ๋ฐ์ ๊ฐ๊ฒฉ์ ๊ฐค๋ฐ๋น 4๋ฌ๋ฌ๋ฅผ ๋์ด์ฐ๊ณ , ํญ๊ณต์ ๊ฐ๊ฒฉ ์์น์ผ๋ก ํญ๊ณต๊ถ๋ ๋น์ธ์ก์ต๋๋ค. ํธ๋ผํ ๋ํต๋ น์ ์ด๋ฌํ ์ํฅ์ด ๋ฏธ๊ตญ์ธ๋ค์๊ฒ ๋ฏธ์น๋ ์ํฅ์ ์ถ์ํ๋ฉด์๋, ์ ๋ฐ์ ์ธ ๋ฏธ๊ตญ ๊ฒฝ์ ์ ๋ํ ์์ ๊ฐ์ ํ์ถํ์ต๋๋ค. ๊ทธ๋ ์ฃผ์ ์์ฅ์ ์ค์ ์ ๊ฐ์กฐํ๊ณ ๋ง์ ๋ฏธ๊ตญ์ธ๋ค์ด ๋ฐ๊ณ ์๋ ์ธ๊ธ ํ๊ธ ์ฆ๊ฐ๋ฅผ ์ธ๊ธํ๋ฉฐ ์ฒญ์ค์๊ฒ โ์ฒ๋ง์์(Youโre welcome)โ๋ผ๊ณ ๋งํ์ต๋๋ค.
R.F.K. Jr.์ ๋ฐฑ์ ์ ์ฅ ์ ํ์ ํ์ฐ๋๋ ํ์ญ
์์ฑํด์์๋ ํธ๋ผํ ํ์ ๋ถ๊ฐ ๊ทธ๋์ ๋ณด๊ฑด๋ถ ์ฅ๊ด ๋ก๋ฒํธ F. ์ผ๋ค๋ ์ฃผ๋์ด(Robert F. Kennedy Jr.)๊ฐ ์ฃผ๋ํด์จ ์ผ๋ถ ๊ฐ๊ฒฝํ ๋ฐฑ์ ํ์๋ก ์์ ๋ฌผ๋ฌ์๋ ๋ชจ์ต์ ๋ณด์ด๊ณ ์์ต๋๋ค. ์ด์ ์์๋ค๊ณผ์ ์ฒญ๋ฌธํ์์ ์ผ๋ค๋ ์ฅ๊ด์ ํ์ญ, ๋ณผ๊ฑฐ๋ฆฌ, ํ์ง(MMR) ๋ฐฑ์ ์ด ์์ ํ๊ณ ํจ๊ณผ์ ์ด๋ฉฐ, ํ์ญ์ ๊ฑธ๋ฆฌ๋ ๊ฒ๋ณด๋ค ๋ฐฑ์ ์ ๋ง๋ ๊ฒ์ด ํจ์ฌ ์์ ํ๋ค๋ ๋ฐ ๋์ํ์ต๋๋ค. ๊ทธ๋ ๋ํ ์ฌํด ์ด ํ ์ฌ์ค์์ ๋ฐ์ํ ํ์ญ ๋ฐ๋ณ์ผ๋ก ์ฌ๋งํ ๋ ๋ช ์ ์ด๋ฆฐ์ด๊ฐ ๋ฐฑ์ ์ ๋ง์๋ค๋ฉด ์ด ์ ์์์ ๊ฒ์ด๋ผ๊ณ ์ธ์ ํ์ต๋๋ค.
์ผ๋ค๋ ์ฅ๊ด์ ๋ฐ์ธ์ ๋ฐฑ์ ์ ์ ์ ์ผ๋ก ์์ฉํ๋ ์์ค์ ์๋์์ง๋ง, ์๋ ๊ฐ ๋ฐฑ์ ์ด ์ํํ๋ค๋ ๊ทผ๊ฑฐ ์๋ ์ฃผ์ฅ์ ํผ์ณ์๋ ๊ทธ์ ๊ณผ๊ฑฐ ๋ฐ์ธ๊ณผ๋ ํ์ฐํ ๋์กฐ๋๋ ๊ฒ์ด์์ต๋๋ค. ์ต๊ทผ ๋ฐฑ์ ๊ด์ ๊ทธ์๊ฒ ์ค๊ฐ์ ๊ฑฐ๋ฅผ ์๋๊ณ ๋ฐฑ์ ๋ฌธ์ ์ ์ง์คํ์ง ๋ง๊ณ ๊ฑด๊ฐํ ์์ต๊ด๊ณผ ๊ฐ์ด ์ ์น์ ์ผ๋ก ๋ ๋ฏผ๊ฐํ ๋ฌธ์ ์ ๋ํด ์ด์ผ๊ธฐํ ๊ฒ์ ์ด๊ตฌํ ๋ฐ ์์ต๋๋ค.
๋ํ ์ด์ ํธ๋ผํ ๋ํต๋ น์ ์๋ก์ด ์ง๋ณํต์ ์๋ฐฉ์ผํฐ(CDC) ๊ตญ์ฅ์ผ๋ก ์๋ฆฌ์นด ์์์ธ (Erica Schwarz) ๋ฐ์ฌ๋ฅผ ์ง๋ช ํ์ต๋๋ค. ์์์ธ ๋ฐ์ฌ๋ ์ ๋ฏธ๊ตญ ๊ณต์ค๋ณด๊ฑด์๋น์ค ๋ถ๊ตญ์ฅ์ ์ง๋์ผ๋ฉฐ, ๋ฐฑ์ ์ ๊ณต๊ฐ์ ์ผ๋ก ์ง์งํด์จ ์ธ๋ฌผ์ ๋๋ค. ํ์ ๋ถ์ ๋ฐฑ์ ๊ด๋ จ ์ ์ฅ ๋ณํ๋ ์ ๊ตญ์ ์ผ๋ก ํ์ญ ๋ฐ๋ณ์ด ๊ณ์ ํ์ฐ๋๋ ๊ฐ์ด๋ฐ ์ด๋ฃจ์ด์ก์ต๋๋ค. ์ต๊ทผ์๋ ์ ํ์ฃผ๊ฐ ์๋ก์ด ํ์ฐ์ ์ค์ฌ์ง๊ฐ ๋์ด, ๋ณด๊ฑด ๋น๊ตญ์ ๊ฑฐ์ 600๊ฑด์ ์ฌ๋ก๋ฅผ ๋ณด๊ณ ํ์ผ๋ฉฐ, ๋๋ถ๋ถ์ด ์ด๋ฆฐ์ด์๊ณ ์ฝ 50๋ช ์ด ์ ์ํ์ต๋๋ค.
์ ํ์ฃผ์ ํ์ฐ์ ์ฃผ๋ชฉํ ๋งํฉ๋๋ค. ํ ์ฌ์ค๋ ์ฌ์ฐ์ค์บ๋กค๋ผ์ด๋์ ๊ฐ์ด ๊ณ ๋ฆฝ๋ ์ข ๊ต ๊ณต๋์ฒด์ ๊ด๋ จ์ด ์์๋ ๋ค๋ฅธ ์ต๊ทผ ํ์ญ ๋ฐ๋ณ ์ฌ๋ก์ ๋ฌ๋ฆฌ, ์ ํ์ ์ฌ๋ก๋ ๊ด๋ฒ์ํ ์ผ๋ฐ ์ธ๊ตฌ์์ ๋ํ๋๊ณ ์๊ธฐ ๋๋ฌธ์ ๋๋ค. ์ํธ๋ ์ดํฌ์ํฐ์ ํ ๊ฐ์ผ๋ณ ์ ๋ฌธ์๋ ๋ด์ํ์์ค์์ ์ธํฐ๋ทฐ์์ ์์ ์ด ์น๋ฃํ๋ ์ํ ์ด๋ฆฐ์ด๋ค์ ๋ง์ ๊ฐ์กฑ๋ค์ด ๋ฐ(ๅ)๋ฐฑ์ ์ฃผ์์๋ ์๋์ง๋ง, ๊ทธ๊ฐ โ์คํด์ ์๋ชป๋ ์ ๋ณด์ ํ์ฐ(epidemic of misinformation)โ์ด๋ผ๊ณ ๋ถ๋ฅด๋ ํ์ ์์์ ๊ฐ๋ฅํ ๋ถ์์ฉ์ ๋ํ ์ฐ๋ ค ๋๋ฌธ์ ์ ์ข ์ ๋ฏธ๋ค๋ค๊ณ ์ ํ์ต๋๋ค.
AI ์๋์ ๊ทธ๋ฆผ์: ๋ฐ์ดํฐ์ผํฐ ๊ฑด์ค ๋ ผ๋
์ด๋ฒ ๋ฌ ๋ฉ์ธ์ฃผ๋ ๋ฏธ๊ตญ์์ ์ฒ์์ผ๋ก ์๋ก์ด ๋ฐ์ดํฐ์ผํฐ ๊ฑด์ค์ ๊ธ์งํ๋ ์ฃผ๊ฐ ๋ ์ ์์ต๋๋ค. ์ธ๊ณต์ง๋ฅ(AI)์ ํ์ํ ์ปดํจํ ํ์๋ฅผ ํ๋ณดํ๊ธฐ ์ํด ๊ธฐ์ ๊ธฐ์ ๋ค์ด ์๋๋ฅด๋ฉด์, ๋ฏธ๊ตญ ์ ์ญ์ ๋คํ, ๋์ง, ์ฒ์ ๋ฐ์ดํฐ์ผํฐ๊ฐ ๊ธ์ฆํ๊ณ ์์ต๋๋ค. ์ด๋ฌํ ๊ฑด์ค ๋ถ์ ๋ง๋ํ ์์ ๋ฌผ๊ณผ ์ ๊ธฐ ์ฌ์ฉ๋, ๊ทธ๋ฆฌ๊ณ ๊ทธ๋ก ์ธํด ์ธ๊ทผ ์ฃผ๋ฏผ๋ค์ ๊ณต๊ณต์๊ธ์ด ์์นํ ๊ฐ๋ฅ์ฑ์ ๋ํ ์ฐ๋ ค๋ฅผ ๋ณ๊ณ ์์ต๋๋ค.
๋ฉ์ธ์ฃผ ์์๋ค์ ๋๊ท๋ชจ ์ ๊ท ์ปดํจํ ์์ค ๊ฑด์ค์ ๋ํด 18๊ฐ์๊ฐ์ ์ ์ ์กฐ์น(moratorium)๋ฅผ ์น์ธํ์ผ๋ฉฐ, ์ด ๋ฒ์์ ์ด์ ์ฃผ์ง์ฌ์ ์์ ๋ฌ๋ ค ์์ต๋๋ค. ๋ฉ์ธ์ฃผ์ ์ ์ฌํ ๊ท์ ๋ฅผ ์ ๊ทน์ ์ผ๋ก ๊ณ ๋ คํ๊ณ ์๋ ๋ค๋ฅธ ์ฃผ๋ค์ ๋ฐ์ดํฐ์ผํฐ ๊ฑด์ค ๋ฐ๋ ๋ ธ๋ ฅ์ ํธ๋ผํ ํ์ ๋ถ์์ ์ถฉ๋์ ์ผ๊ธฐํ ์ ์์ต๋๋ค. ํธ๋ผํ ๋ํต๋ น์ AI ์ฐ์ ์ ์ฑ์ฅ์ ์ ํํ๋ ์ฃผ๋ค์ ๊ณ ์ํ๊ณ ์ฐ๋ฐฉ ์๊ธ ์ง์์ ๋ณด๋ฅํ๊ฒ ๋ค๊ณ ์ํํ ๋ฐ ์์ต๋๋ค.
๋ํ, ์ ๊ท ๊ฑด์ค์ ๋ง์ผ๋ ค๋ ๋ ธ๋ ฅ์ ์ง์ญ ์ฐจ์์์๋ ๋ ผ๋์ด ๋ ์ ์์ต๋๋ค. ์๋ฅผ ๋ค์ด ๋ฉ์ธ์ฃผ์ ํ ๋ฐ์ดํฐ์ผํฐ ํ๋ก์ ํธ๋ ์ค๋๋ ๋ฒ๋ ค์ง ๋ฐฉ์๊ฐ์ ํ์ฑํํ๊ณ ๋ฏธ๊ตญ ๊ฒฝ์ ์ ํต์ฌ ๋ถ๋ถ์ด ๋ ์ฐ์ ์์ ์๋ก์ด ์ผ์๋ฆฌ๋ฅผ ์ฐฝ์ถํ๋ ๋ฐฉ๋ฒ์ผ๋ก ์ ์๋์์ต๋๋ค. ์ด ํ๋ก์ ํธ์ ๊ฐ๋ฐ์ ์ค ํ ๋ช ์ ๋ด์ํ์์ค์ โ์ฌ๋๋ค์ AI๋ฅผ ๋๋ ค์ํฉ๋๋ค. ์ ๋ AI๋ฅผ ๋๋ ค์ํฉ๋๋ค. ํ์ง๋ง ์์๋์? AI๋ ์ค๊ณ ์์ต๋๋คโ๋ผ๊ณ ๋งํ์ต๋๋ค.
๋ฐํด ๋ฌ๋ฆฐ ์ ๋ฐ โํ๋ฆฌ์ค(Heelys)โ์ ๋ฐ๋ช ๊ฐ, ๋ก์ ์ ๋ค์ค ๋ณ์ธ
2000๋ ๋ ์ด๋ฐ ์ ํ์ ์ธ ์ธ๊ธฐ๋ฅผ ๋์๋ ๋ฐํด ๋ฌ๋ฆฐ ์ ๋ฐ โํ๋ฆฌ์ค(Heelys)โ์ ๋ฐ๋ช ๊ฐ ๋ก์ ์ ๋ค์ค(Roger Adams)๊ฐ 71์ธ์ ๋์ด๋ก ๋ณ์ธํ์ต๋๋ค. ๋ก์ ์ ๋ค์ค๋ ์์ฑํด์ฃผ ํ์ฝ๋ง์์ ๋กค๋ฌ์ค์ผ์ดํธ์ฅ์ ์์ ํ๋ ๋ถ๋ชจ๋ ๋ฐ์์ ์๋ผ 9๊ฐ์ ๋ง์ ๋ฐํด ์์์ ๊ท ํ์ ์ก์ ์ ๋๋ก ์ค์ผ์ดํธ์ ๋ฅ์ํ์ต๋๋ค. ํ์ง๋ง ๊ทธ์ ์ ๋ฌธ์ฑ์ ํ์ฐธ ๋ค์ธ 40๋์ ์ด๋ฅด๋ฌ์์ผ ๋ฐํ๋์์ต๋๋ค. ์ดํผ์ ๊ฒช๊ณ ์ฌ๋ฆฌํ์๋ผ๋ ์ง์ ์ ํ์๋ฅผ ๋๋ผ๋ฉฐ ์ธ์์ ๋ง๋ค๋ฅธ ๊ณจ๋ชฉ์ ๋ค๋ค๋ฅธ ๊ทธ๋, ํ์ํ ๋ ๋ฐํด๋ฅผ ๊บผ๋ด ์ ์ ์ ์๋ ์ ๋ฐ์ ๋ํ ๋น์ ์ ๊ฐ๊ฒ ๋ฉ๋๋ค. ์น๊ตฌ์ ์ฐจ๊ณ ์์ ์ด๋ํ ๋ค๊ฟ์น๋ฅผ ์๋ผ๋ด๊ณ ๋ฐํด๋ฅผ ๋ผ์ ์์ ํ์ ๋ง๋ค์๊ณ , ๊ทธ๋ ๊ฒ ํ๋ฆฌ์ค๊ฐ ํ์ํ์ต๋๋ค.
2008๋ ๊น์ง ํ๋ฆฌ์ค๋ ์ฐ๊ฐ 700๋ง ์ผค๋ ์ด์ ํ๋งค๋๋ฉฐ ํญ๋ฐ์ ์ธ ์ฑ๊ณต์ ๊ฑฐ๋์์ต๋๋ค. ๋น์ ์ผํ๋ชฐ์์๋ ๋ฐํด ์๋ฆฌ๊ฐ ๋์ด์ง ์์๊ณ , ์์ด๋ค์ด ํ๋ฒํ๊ฒ ๊ฑท๋ค๊ฐ๋ ๊ฐ์๊ธฐ ๋ณ์ ๋ก๋ด์ฒ๋ผ ๋ฐ๋ฐ์์ ๋ฐํด๋ฅผ ๊บผ๋ด ๋ฏธ๋๋ฌ์ ธ ๋๊ฐ๋ ๋ชจ์ต์ ํํ ํ๊ฒฝ์ด์์ต๋๋ค. ๋ด์ํ์์ค ๋ถ๊ณ ๋ด๋น ๊ธฐ์ ์ ์ฌ ์ฐ(Ash Woo)๋ ์ฑ์ธ์ด ๋์ด ๋ค์ ํ๋ฆฌ์ค๋ฅผ ์ ์ด๋ณธ ๊ฒฝํ์ ์ ํ๋ฉฐ โ์กฐ์ฌ์ฑ์ด ํต์ฌโ์ด๋ผ๊ณ ๊ฐ์กฐํ์ต๋๋ค.
ํ๋ฆฌ์ค๋ ์ ๋ค์ค์๊ฒ ์๋ฐฑ๋ง ๋ฌ๋ฌ๋ฅผ ๋ฒ์ด๋ค ์ค ๋์ฑ๊ณต์์ด์์ง๋ง, ์๊ธ์ค ์์ฌ๋ค์๊ฒ๋ ๋ฌ๊ฐ์ง ์์์ต๋๋ค. ์ ๋ฐ์ด ๊ฐ์๊ธฐ ๋กค๋ฌ์ค์ผ์ดํธ๋ก ๋ณํ๋ฉด์ ๋ฐ์ํ ์ ์๋ ๋จธ๋ฆฌ ๋ฐ ์ฒ์ถ ๋ถ์ ์ํ ๋๋ฌธ์ ํ๋ โ์ต์ ์ ์ฅ๋๊ฐโ ๋ชฉ๋ก์ ์ค๋ฅด๊ธฐ๋ ํ์ต๋๋ค. ์ดํ์ด ๋ถ๋ ๋ช ๋ ํ, ์ ๋ค์ค๋ ํ๋ฆฌ์ค ํ์ฌ๋ฅผ ๋ ๋ ๋์ค์ ์์ ์์ ๋ฒ์ด๋ฌ์ง๋ง, ๋ค๋ฅธ ์จ์ด๋ฌ๋ธ ๋กค๋ฌ ์ฅ์น์ ๋ํ ํนํ ์ถ์์ ๊ณ์ํ์ต๋๋ค.
์ค๋ ๋ด์ํ์์ค ํค๋๋ผ์ธ์ ๋ณต์กํ ์ธ๊ณ์ ๊ธ๋ณํ๋ ์ฌํ์ ๋จ๋ฉด์ ๋ณด์ฌ์ค๋๋ค. ๋ถ์์ ํ ๊ตญ์ ์ ์ธ ์ ํฌ๋ง๊ณผ ์ข์ , ๋ณด๊ฑด ์ ์ฑ ์ ๋ณํ๊ฐ ๊ฐ์ ธ์ฌ ๋ฏธ๋, ๊ทธ๋ฆฌ๊ณ ๊ธฐ์ ๋ฐ์ ์ด ๋์ง๋ ์ค๋ฆฌ์ , ์ฌํ์ ์ง๋ฌธ๊น์ง. ๋ค์ํ ์ด์๋ค์ด ์ฐ๋ฆฌ์ ํ์ฌ์ ๋ฏธ๋๋ฅผ ํ์ฑํ๊ณ ์์ต๋๋ค.
โWhy the A.I. Backlash Turned Violent in Americaโ โ Hard Fork ๊ธฐ๋ฐ ๊ธฐ์ฌ ์๋ณธ ์์ ๋ณด๊ธฐ
AI๋ฅผ ํฅํ ๋ถ๋ ธ: ์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ ์๋ฆฌํธ์ ๋์ค์ ์ถฉ๋
์ต๊ทผ ๋ช ๋ฌ๊ฐ ์ธ๊ณต์ง๋ฅ(AI)์ ๋ํ ๋์ค์ ์ธ์์ด ๊ธ๊ฒฉํ ์ ํ๋๋ฉด์, ๊ทธ ๋ฐ๊ฐ์ด ์ด์ ๋ฌผ๋ฆฌ์ ์ธ ํญ๋ ฅ์ผ๋ก๊น์ง ๋ฒ์ง๋ ์ถฉ๊ฒฉ์ ์ธ ์ฌ๊ฑด๋ค์ด ๋ฐ์ํ๊ณ ์์ต๋๋ค. ๋ด์ํ์์ค(New York Times)์ ๊ธฐ์ ์นผ๋ผ๋์คํธ ์ผ๋น ๋ฃจ์ค(Kevin Roose)์ ํ๋ซํผ์ด(Platformer)์ ์ผ์ด์ ๋ดํด(Casey Newton)์ด ์งํํ๋ ํ์บ์คํธ โํ๋ ํฌํฌ(Hard Fork)โ์์๋ ์ด๋ฌํ AI ๋ฐ๋ฐ์ ์์ธ์ ์ฌ์ธต ๋ถ์ํ๊ณ , ์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ ๋ด๋ถ์ ๋ฌธํ์ ๋จ์ ๊ณผ ์ ๋ถ์ ์ญํ ๋ถ์ฌ์ ๋ํ ๋ ์นด๋ก์ด ๋นํ์ ์ ๊ธฐํ์ต๋๋ค.
AI ๋ฐ๊ฐ, ํญ๋ ฅ์ผ๋ก ๋ฒ์ง๋ค: ์ถฉ๊ฒฉ์ ์ธ ์ฌ๊ฑด๋ค
AI์ ๋ํ ๋ถ์ ์ ์ธ ์ฌ๋ก ์ ๋ ์ด์ ๋จ์ํ ์จ๋ผ์ธ ๋ ผ์์ ๋จธ๋ฌด๋ฅด์ง ์์ต๋๋ค. ์ง๋์ฃผ, ์คํAI(OpenAI)์ CEO ์ ์ฌํธ๋จผ(Sam Altman)์ ์ํ๋์์ค์ฝ ์ํ์ 20์ธ ๋จ์ฑ์ด ํ์ผ๋ณ(Molotov cocktail)์ ๋์ง๋ ์ฌ๊ฑด์ด ๋ฐ์ํ์ต๋๋ค. ์ฉ์์๋ AI ๋ฐ๋ ๊ฒฌํด๋ฅผ ๋ด์ ๋ฌธ๊ฑด๊ณผ ๋ค๋ฅธ AI ๊ฒฝ์์ง, ํฌ์์, ์ด์ฌํ ๊ตฌ์ฑ์๋ค์ ์ด๋ฆ๊ณผ ์ฃผ์ ๋ชฉ๋ก์ ์์งํ๊ณ ์์์ผ๋ฉฐ, AI๊ฐ ์ด๋ํ ์ค์กด์ ์ํ(existential risk)์ ๋ํ ์ฐ๋ ค ๋๋ฌธ์ ์ด ๊ฐ์ ํ๋์ ์ ์ง๋ฅธ ๊ฒ์ผ๋ก ์๋ ค์ก์ต๋๋ค. ๋คํํ ์ธ๋ช ํผํด๋ ์์์ง๋ง, ์ด๋ AI์ ๋ํ ๊น์ ๋ถ์ ๊ณผ ๋ถ๋ ธ๊ฐ ์ผ๋ง๋ ์ํํ ์์ค์ ์ด๋ฅด๋ ๋์ง๋ฅผ ๋ณด์ฌ์ฃผ๋ ๋จ์ ์ธ ์์ ๋๋ค.
์ด ์ฌ๊ฑด ๋ฉฐ์น ์ ์๋ ์ธ๋์ ๋์ฃผ ์ธ๋์ ๋ํด๋ฆฌ์ค์์ ๋ ์ฐ๋ ค์ค๋ฌ์ด ์ฌ๊ฑด์ด ์์์ต๋๋ค. ๋ก ๊น์จ(Ron Gibson) ์์์๊ณผ ๊ทธ์ ์๋ค์ ์๋ฒฝ์ ์ญ์ฌ ๋ฐ์ ์ด๊ฒฉ์ผ๋ก ์ ์์ ๊นผ๊ณ , ๋ฌธ ๋ฐ์๋ โ๋ฐ์ดํฐ ์ผํฐ ๋ฐ๋(No data centers)โ๋ผ๋ ์ชฝ์ง๊ฐ ๋ฐ๊ฒฌ๋์์ต๋๋ค. ๊น์จ ์์์ ์์ ์ ์ง์ญ๊ตฌ์ ๋ฐ์ดํฐ ์ผํฐ ๊ฑด์ค์ ์ง์งํ๊ณ ๋ถ์ง ์ฌ์กฐ์ (rezoning)์ ์ฐฌ์ฑํ๋ฅผ ๋์ง ๋ฐ ์์ต๋๋ค.
์ผ๋น ๋ฃจ์ค๋ ์ด๋ฌํ ์ฌ๊ฑด๋ค์ โ๋ฐ(ๅ)AI ๊ธ์งํ์ ํญ๋ ฅ์ ์ฆ๊ฐ ์ถ์ธโ์ ์ผ๋ถ๋ก ์ง๋จํ๋ฉฐ, ์ด๋ค ํํ์ ํญ๋ ฅ๋ ์ฉ๋ฉ๋ ์ ์์์ ๊ฐ์กฐํ์ต๋๋ค. ํญ๋ ฅ์ ๋๋์ ์ผ๋ก ์๋ชป๋์์ ๋ฟ ์๋๋ผ, ์ ์น์ ๋ชฉํ ๋ฌ์ฑ์๋ ์ ํ ํจ๊ณผ์ ์ด์ง ์๋ค๋ ์ง์ ์ ๋๋ค.
๋ฐ์ดํฐ ์ผํฐ: AI ์๋์ ์ต์ ์ ๊ฐ๋ฑ
AI ๊ธฐ์ ๋ค์ด ์๋น์ค๋ฅผ ์ ๊ณตํ๊ธฐ ์ํด ํ์์ ์ธ ๋ฐ์ดํฐ ์ผํฐ(data center)๋ AI ๋ถ์ ๊ฐ์ฅ ๊ฐ์์ ์ธ ์์ง์ด์ ๋์ค์ ๋ฐ๋ฐ์ด ๊ฐ์ฅ ๋จ๊ฒ๊ฒ ์ผ์ด๋๋ ์ง์ ์ ๋๋ค. ๋ฏธ๊ตญ ์ ์ญ์์ ๋ฐ์ดํฐ ์ผํฐ ๊ฑด์ค์ ๋ํ โ๋ด ๋ท๋ง๋น์๋ ์ ๋ผ(Not In My Backyard, NIMBY)โ ์์ ๊ฐ๋ ฅํ ์ ํญ์ด ํ์ฐ๋๊ณ ์์ต๋๋ค.
๋ฉ์ธ์ฃผ(Maine)๋ 2027๋ 11์๊น์ง 20๋ฉ๊ฐ์ํธ(MW) ์ด์์ ๋ฐ์ดํฐ ์ผํฐ ๊ฑด์ค์ ์ผ์์ ์ผ๋ก ๊ธ์งํ๋ ๋ฒ์(moratorium)์ ํต๊ณผ์์ผฐ์ต๋๋ค. ์์ค์ฝ์ ์ฃผ ํฌํธ ์์ฑํด(Port Washington)์ ํ ๊ต์ธ ์ง์ญ์์๋ ๋ฏธ๋ ๋ฐ์ดํฐ ์ผํฐ ๊ฑด์ค์ ์ ํํ๊ธฐ ์ํด ์ ๊ถ์ ์น์ธ์ ์๋ฌดํํ๋ ์์ด ์๋์ ์ธ ์ฐฌ์ฑ์ผ๋ก ํต๊ณผ๋์์ต๋๋ค. ์คํ์ด์ค, ๋ฏธ์ฃผ๋ฆฌ, ์ธ๋์ ์ ๋, ์กฐ์ง์, ๋ ธ์ค์บ๋กค๋ผ์ด๋ ๋ฑ ๋ค๋ฅธ ์ฃผ์์๋ ์ ์ฌํ ์์ง์์ด ํฌ์ฐฉ๋ฉ๋๋ค. ์ฐ๋ฐฉ ์ฐจ์์์๋ ๋ฒ๋ ์๋์ค(Bernie Sanders) ์์์์๊ณผ AOC(Alexandria Ocasio-Cortez) ํ์์์์ด ์ ๊ตญ์ ์ธ ๋ฐ์ดํฐ ์ผํฐ ๊ฑด์ค ๋ชจ๋ผํ ๋ฆฌ์์ ์ ์ํ๊ธฐ๋ ํ์ต๋๋ค.
์ด๋ฌํ ๋ฐ์ดํฐ ์ผํฐ ๋ฐ๋ ์ด๋์ AI ๊ฒฝ์์ง์ ๋ํ ๊ฐ๋ณ์ ์ธ ์ํ๊ณผ ํจ๊ป, AI ์ฐ์ ์ ๋ฐ์ ๋ํ ๊น์ ๋ถ์ ์ ๋ฐ์ํฉ๋๋ค. ์คํ ํผ๋ ๋ํ๊ต์ โ2026 AI ์ง์(AI Index)โ ๋ณด๊ณ ์์ ๋ฐ๋ฅด๋ฉด, ๋ฏธ๊ตญ์ธ๋ค์ AI๋ฟ๋ง ์๋๋ผ ์ ๋ถ๊ฐ AI๋ฅผ ์ฑ ์๊ฐ ์๊ฒ ๊ท์ ํ ์ ์์์ง์ ๋ํด์๋ ๋งค์ฐ ๋ฎ์ ์ ๋ขฐ๋ฅผ ๋ณด์ ๋๋ค. ์ ์ธ๊ณ ํ๊ท ์ ๋ขฐ๋๊ฐ 54%์ธ ๋ฐ๋ฉด, ๋ฏธ๊ตญ์ 31%์ ๋ถ๊ณผํ์ต๋๋ค. ํจ ๋ฆฌ์์น ์ผํฐ(Pew Research Center)์ ์ฐ๊ตฌ ์ญ์ ๋ฐ์ดํฐ ์ผํฐ๊ฐ ํ๊ฒฝ, ์ฃผํ ์๋์ง ๋น์ฉ, ์ถ์ ์ง์ ๋ฏธ์น๋ ์ํฅ์ ๋ํด ๋ถ์ ์ ์ธ ์ธ์์ด ๋ ๊ฐํจ์ ๋ณด์ฌ์ค๋๋ค.
AI ๋ฐ๊ฐ์ ์ธ ๊ฐ์ง ๋ฟ๋ฆฌ
ํ์บ์คํธ ์งํ์๋ค์ AI์ ๋ํ ๋์ค์ ๋ถ๋ง์ด ๊ธ๊ฒฉํ ๊ณ ์กฐ๋ ์ฃผ์ ์์ธ์ ์ธ ๊ฐ์ง๋ก ๋ถ์ํฉ๋๋ค.
1. ๊ณผ์ฅ๋ ์์ฌํ์ ์ญํ: ์ ์ฌํธ๋จผ์ ์์ ์ ๋ธ๋ก๊ทธ์ ๋ด์์ปค(New Yorker) ๊ธฐ์ฌ์ฒ๋ผ โ์ ๋์ ์ธ(incendiary)โ ๊ธฐ์ฌ๋ค์ด AI์ ๋ํ ๋ถ์๊ฐ์ ํค์ ์์ ์ ๋ ์ํํ๊ฒ ๋ง๋ ๋ค๊ณ ์ฃผ์ฅํ์ต๋๋ค. ๊ทธ๋ฌ๋ ์งํ์๋ค์ ์คํ๋ ค AI CEO๋ค ์ค์ค๋ก๊ฐ AI์ โ์ค์กด์ ์ํโ์ ๋ํ ์์ฌํ์ ๋ถ์ฑ์งํ๋ค๊ณ ์ง์ ํฉ๋๋ค. ์ ์ฌํธ๋จผ ์์ ๋ 2015๋ ๋ธ๋ก๊ทธ ๊ฒ์๋ฌผ์์ โ์ด์ธ์ ์ธ ๊ธฐ๊ณ ์ง๋ฅ ๊ฐ๋ฐ์ ์ธ๋ฅ์ ์ง์์ ์ธ ์กด์ฌ์ ๊ฐ์ฅ ํฐ ์ํ์ด ๋ ๊ฒโ์ด๋ผ๊ณ ์ธ๊ธํ ๋ฐ ์์ต๋๋ค.
์ด๋ฌํ ์ง๋์๋ค์ โ์ฌํ๊ด๊ณ๋ง ์๋น์ค(SNS) CEO๋ค์ด ๊ณผ๊ฑฐ ๊ธฐ์ ์ ๊ธ์ ์ ์ธก๋ฉด๋ง ๊ฐ์กฐํ๋ค ๋นํ๋ฐ๋ ๊ฒ์ ๋ณด์๊ธฐ ๋๋ฌธ์, ์์ ๋ค์ ์์งํ๊ฒ ์ํ์ ์ธ์ ํด์ผ ํ๋คโ๋ ๋๋ ๋ง์ ๋น ์ ธ ์์ต๋๋ค. ์ํ์ ์์งํ ๋งํ๋ฉด โ์ข ๋ง๋ก ์(doomers)โ๋ก ๋น๋๋ฐ๊ณ , ๋๊ด์ ์ธ ๋น์ ์ ์ ์ํ๋ฉด โ์ข๊ฒ๋ง ํฌ์ฅํ๋ค(sugarcoat)โ๋ ๋นํ์ ๋ฐ์ต๋๋ค. ์ผ์ด์ ๋ดํด์ ์ด ๋ชจ์์ ํด๊ฒฐํ ์ธ ๋ฒ์งธ ๊ธธ์ โ์ ๋ถ์ ํ๋ ฅํ์ฌ ์ ํ์ผ๋ก ์ธํ ํผํด๋ฅผ ์ํํ ๊ท์ ์์คํ ์ ๋ง๋ จํ๋ ๊ฒโ์ด๋ผ๊ณ ์ ์ํฉ๋๋ค. ๊ทธ๋ฌ๋ ์คํAI์ ๊ฒฝ์ฐ, ์คํ๋ ค AI ๊ท์ ๋ฒ์ ํต๊ณผ๋ฅผ ๋ง์ผ๋ ค ๋ก๋น ํ๋์ ํผ์ณ์์ต๋๋ค.
2. ์ผ์๋ฆฌ ์์ค ์ฐ๋ ค์ ๊ฒฝ์ ์ ๋ถ์: ๋์ค์ด AI์ ๋ํด ๊ฐ์ฅ ๊ตฌ์ฒด์ ์ด๊ณ ๋จ๊ธฐ์ ์ผ๋ก ์ฐ๋ คํ๋ ๊ฒ์ โ์ด ๊ธฐ์ ์ด ๋ด ์ผ์๋ฆฌ๋ฅผ ์์ ํ ๋์ฒดํ๊ฑฐ๋, ํ์ฌ์ ์ผ์๋ฆฌ๋ฅผ ๋์ฐํ๊ฒ ๋ง๋ค ๊ฒโ์ด๋ผ๋ ์ ์ ๋๋ค. ์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ ๋ด๋ถ์์๋ AI๊ฐ โ์์ ์๋ํ๋ ๊ณ ๊ธ ๊ณต์ฐ์ฃผ์(fully automated luxury communism)โ๋ฅผ ๊ฐ์ ธ์ ์๋ฌด๋ ์ผํ ํ์๊ฐ ์๋ ์ ํ ํผ์๋ฅผ ๊ฟ๊พธ์ง๋ง, ์ด๋ฌํ ๋น์ ์ ๋์ค์๊ฒ๋ ๋นํ์ค์ ์ผ๋ก ๋ค๋ฆฝ๋๋ค.
์ผ๋น ๋ฃจ์ค๋ โ๋ชจ๋ ์ ์น์ ๋ง์ฐฌ๊ฐ์ง๋ก, AI ์ ์น ์ญ์ ๊ถ๊ทน์ ์ผ๋ก ์ฌ๋๋ค์ด ๊ธฐ์ ์ ๋ณด๋ฉฐ โ์ด๊ฒ์ด ๋์๊ฒ, ๋ด ๊ฐ์กฑ์ ๋ถ์ํ๊ณ ํธ์ํ๊ฒ ์ํดํ ๋ฅ๋ ฅ์ ๋ฌด์์ ํ ๊ฒ์ธ๊ฐโ๋ฅผ ์๊ฐํ๋ ๊ฒ์ผ๋ก ๊ท๊ฒฐ๋๋คโ๊ณ ๋งํฉ๋๋ค. ์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ์ ๊ธฐ์ ๋๊ด์ฃผ์์๋ค์ ๊ธ๊ฒฉํ ๊ธฐ์ ๋ณํ๋ฅผ ํ์ํ์ง๋ง, ๋๋ถ๋ถ์ ์ฌ๋๋ค์ ์ถ์ ์์ ์ฑ์ ์ํฉ๋๋ค. ์ผ์๋ฆฌ๋ฅผ ๋นผ์๊ณ ๋ฏธ๋๋ฅผ ์์ธก ๋ถ๊ฐ๋ฅํ๊ฒ ๋ง๋ค ์ ์๋ ๊ธฐ์ ์ ๋ํ ๋๋ ค์๊ณผ ๋ถ๋ ธ๋ ์์ฐ์ค๋ฌ์ด ๋ฐ์์ด๋ผ๋ ๊ฒ์ ๋๋ค.
3. ์๋ฆฌํธ์ฃผ์์ ํต์ ๋ ฅ ์์ค๊ฐ: ํ์ฌ์ AI ๋ฐ์ ์ โํํฅ์(top-down)โ ํ๋ก์ ํธ์ ๋๋ค. ์ฐจ๊ณ ์์ ๋ชจ์ธ ๊ดด์ง๋ค์ด ์์ํ ๊ฒ์ด ์๋๋ผ, ์์์ ๋งค์ฐ ๋๋ํ ์ฌ๋๋ค์ด ๋ง๋ํ ์๋ณธ์ ํ๋ณดํ์ฌ ๋น ๋ฅด๊ฒ ๊ฐ๋ฐํ๊ณ ๋ฐฐํฌํ๋ ๋ฐฉ์์ ๋๋ค. ๋์ค์ โ์ด๊ฒ์ ์๊ตฌํ ์ ๋ ์์ง๋ง, ์๋ฏธ ์๋ ํต์ ๊ถ๋ ์๋คโ๊ณ ๋๋๋๋ค.
์คํAI๊ฐ ์บ๋ฆฌํฌ๋์์ AI ํฌ๋ช ์ฑ ๋ฒ์ ํต๊ณผ๋ฅผ ์ ์งํ๊ณ , AI ๊ท์ ๋ฅผ ์ง์งํ๋ ๋น์๋ฆฌ ๋จ์ฒด์ ์ํ์ฅ(subpoena)์ ๋ณด๋ด๋ ๋ฑ ๋ฏผ์ฃผ์ ์ ์ฐจ๋ฅผ ๊ฑฐ์ค๋ฅด๋ ํ๋์ ํ ์ฌ๋ก๋ ์ด๋ฌํ ์๋ฆฌํธ์ฃผ์์ ๊ฒฝํฅ์ ๋ณด์ฌ์ค๋๋ค. ์ต๊ทผ ์ผ๋ฆฌ๋ ธ์ด์ฃผ์์ ์คํAI๊ฐ ๋ชจ๋ธ ์ฌ์ฉ์ผ๋ก ์ฌ๊ฐํ ํผํด๊ฐ ๋ฐ์ํด๋ ๊ณ ์์ ์ด์ง ์์ ํ ์ฑ ์์ ๋ฉด์ ๋ฐ๋ ๋ฒ์์ ์ง์งํ ๊ฒ์ โ์ธ๋ฅ์๊ฒ ์ค์กด์ ์ํ์ด ๋ ์ ์๋ ๊ฒ์ ๋ง๋ค๋ฉด์, ๋์์ ์ฑ ์์ผ๋ก๋ถํฐ ๋ณดํธ๋ฐ๋ ๋ฒ์์ ๋ก๋นํ๋ ๊ฒโ์ด๋ผ๊ณ ์ผ์ด์ ๋ดํด์ ๋นํํฉ๋๋ค.
๊ธฐ์ ์ ๊ณ์์๋ ์์จ์ฃผํ์ฐจ(Waymo)์ฒ๋ผ ์ธ๊ฐ ์ด์ ์๋ณด๋ค โ์ ์ฆํ ์ ์์ ์ ๋๋ก ์์ ํ(demonstrably safer)โ ๊ธฐ์ ์กฐ์ฐจ๋ โํ์ ์ด์ ์ฌ๋ค์ ์ผ์๋ฆฌ๋ ์ด๋ป๊ฒ ํ ๊ฒ์ธ๊ฐโ๋ผ๋ ์ง๋ฌธ ๋๋ฌธ์ ๋์ค์ ๋ฐ๋ฐ์ ๋ถ๋ชํ๋ ํ์์ ๋น์๊ฑฐ๋ ๋ฌด์ํ๋ ๊ฒฝํฅ์ด ์์ต๋๋ค. ์ผ๋ถ ์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ ์ธ์ฌ๋ค์ โ์ด ๊ธฐ์ ์ ๋๋ฌด ์ค์ํด์ ๋์ค์๊ฒ ๋งก๊ธธ ์ ์๋คโ๊ณ ์๊ฐํ์ง๋ง, ์ด๋ ์๋ชป๋ ํ๋๋ผ๋ ์ง์ ์ ๋๋ค.
ํด๊ฒฐ์ฑ ์ ์ด๋์ ์๋๊ฐ?
๊ทธ๋ ๋ค๋ฉด AI๊ฐ ์ด๋ํ ํด์ ๊ณผ ์ค์ ๋ฌธ์ ์ ๋ํด ๋์ค์ด ์ทจํ ์ ์๋ โ์ฌ๋ฐ๋ฅธ ์ง๋ ๋(right levers)โ๋ ๋ฌด์์ผ๊น์? ์งํ์๋ค์ ๋ฐ์ดํฐ ์ผํฐ ๊ฑด์ค์ ๋ง๋ ๋๋น(NIMBY) ํ์์ด AI ๋ฐ์ ์๋์ ์๋ฏธ ์๋ ์ํฅ์ ๋ฏธ์น์ง ๋ชปํ ๊ฒ์ด๋ผ๊ณ ํ์์ ์ธ ์๊ฐ์ ๋ณด์ ๋๋ค. ๋์ , ์ค์ง์ ์ธ ์ ์ฑ ์ ์์ผ๋ก โ์ ์ฐํ ์ฌํ ์์ ๋ง(social safety nets)โ๊ณผ โ์ฌํ๋ จ ํ๋ก๊ทธ๋จ(retraining programs)โ์ ์ ์ํฉ๋๋ค. ์ ๋ฝ์ ์ผ๋ถ ๊ตญ๊ฐ์์๋ ๋ก๋ด ๋๋ฌธ์ ์ผ์๋ฆฌ๋ฅผ ์์ ๋ ธ๋์๋ค์๊ฒ ์ผ์ ๊ธฐ๊ฐ ๊ธ์ฌ๋ฅผ ์ง๊ธํ๊ณ ์๋ก์ด ๊ธฐ์ ์ ์ฌ๊ต์กํ์ฌ ๊ณ ์ฉ์ ์ ์งํ๋ ์ ๋๋ฅผ ์ด์ํ๊ณ ์์ต๋๋ค.
์คํAI ์ญ์ โ์ง๋ฅ ์๋์ ์ฐ์ ์ ์ฑ (Industrial Policy for the Intelligence Age)โ์ด๋ผ๋ ๋ฐฑ์์์ ๋ชจ๋ ์๋ฏผ์ด AI์ ๊ฒฝ์ ์ ์ด๋์ ๊ณต์ ํ๋ โ๊ณต๊ณต ๋ถ(ๅฏ) ๊ธฐ๊ธ(public wealth fund)โ ์กฐ์ฑ, ๋ ธ๋์๋ฅผ ์ํ ์์ ๋ง ๊ฐ์ , ์๋์ง ์์ฐ ๊ฐ์ํ๋ฅผ ์ํ ๋ฏผ๊ด ํํธ๋์ญ ๊ตฌ์ถ ๋ฑ์ ์ ์ํ์ต๋๋ค. ๊ทธ๋ฌ๋ ์ผ์ด์ ๋ดํด์ ์ด๋ฌํ ์ ์์ด โ์์ฒญ๋ ๋ถ์ ์ฌ๋ถ๋ฐฐโ๋ฅผ ์๊ตฌํ๋ ๊ฒ์์๋ ๋ถ๊ตฌํ๊ณ , ์คํAI์ ์ค์ ๋ก๋น ํ๋๊ณผ ์ ์น ๊ธฐ๋ถ๊ฐ โ๋๊ท๋ชจ ๋ณต์ง ๊ตญ๊ฐ ํ๋๋ฅผ ์ง์งํ ๋ฆฌ ์๋ ๊ณตํ๋น ํ๋ณด๋ค์ ์ฃผ๋ก ํ์ํ๋คโ๋ ์ ์์ ์ง์ ์ฑ์ ์๋ฌธ์ ์ ๊ธฐํฉ๋๋ค.
๊ฒฐ๋ก ์ ์ผ๋ก, ํ์ฌ AI ์ฐ์ ์ ๋์ค์ ์ถ์ ์์ฒญ๋ ํผ๋์ ์ผ๊ธฐํ ๊ฒ์ด๋ผ๊ณ ๋งํ๋ฉด์๋, ๊ทธ ํผ๋ ์ดํ์ ๋ํ ๊ตฌ์ฒด์ ์ธ ๊ณํ์ด ์๋ โ๊ฑฐ๋ฒ๋์ค(governance) ๋ฌธ์ โ์ ์ง๋ฉดํด ์์ต๋๋ค. ์งํ์๋ค์ ์ ๋ถ๊ฐ ์ด ๋ฌธ์ ์ ๋ํ ํด๋ต์ ๋ด๋์์ผ ํ์ง๋ง, ํ์ฌ ์ ๋ถ์ ๋ฅ๋ ฅ์ ๋ํด์๋ ๋น๊ด์ ์ธ ์๊ฐ์ ๋๋ฌ๋ ๋๋ค. ๊ทธ๋ผ์๋ ๋ถ๊ตฌํ๊ณ , AI์ ํํ์ด ๋ชจ๋ ์ฌ๋์๊ฒ ๋์๊ฐ๋๋ก ๋ณด์ฅํ ์ ์ฑ ๊ณํ์ ์ ์ํ๋ ์ ์น์ธ๋ค์๊ฒ๋ โ๋ฏธ๊ตญ ์ ์น์ ๋์ ๊ธฐํ์ ๊ธธ(wide open lane)โ์ด ์ด๋ ค ์๋ค๊ณ ๊ฐ์กฐํ๋ฉฐ, ๊ทธ๋ฌํ ๋ฆฌ๋์ญ์ด ์ญ์ฌ์ ์ด๋ฆ์ ์๊ธธ ๊ฒ์ด๋ผ๊ณ ์ญ์คํฉ๋๋ค.
์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ์ ๋ ๋ค๋ฅธ ์ง์ฐฉ: ๋ถ๋ฉธ์ ๊ฟ
ํ์บ์คํธ์ ํ๋ฐ๋ถ์์๋ ๊ธฐ์ ์ ๋๋ฆฌ์คํธ ์นด๋ผ ์ค์์ (Kara Swisher)๊ฐ CNN์์ ๋ฐฉ์ํ๋ ์๋ก์ด ๋คํ๋ฉํฐ๋ฆฌ โ์นด๋ผ ์ค์์ ๋ ์์ํ ์ด๊ณ ์ถ๋ค(Cara Swisher Wants to Live Forever)โ๋ฅผ ํตํด ์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ ์๋ฆฌํธ๋ค์ ๋ ๋ค๋ฅธ ์ง์ฐฉ์ธ โ์ฅ์(longevity)โ์ โ๋ถ๋ฉธ(immortality)โ์ ๋ํ ํ๊ตฌ๋ฅผ ๋ค๋ฃน๋๋ค.
์ค์์ ๋ ์ด ๋คํ๋ฉํฐ๋ฆฌ ์ ๋ชฉ์ด โ๋๋ดโ์ด๋ผ๊ณ ๋งํ๋ฉฐ, ์์ ์ด ์ง์ ์ผ๋ก ์์ํ ์ด๊ณ ์ถ์ด ํ๋ ๊ฒ์ ์๋๋ผ๊ณ ๋ฐํ๋๋ค. ๊ทธ๋ ์ ๋ชฉํ๋ โ๋งค์ฐ ๊ธธ๊ณ ํจ์ฌ ๋์ ์ถ์ ์ฌ๋ ๋ฐ ๋์์ด ๋ ๋งํ ๊ฒ๋คโ์ ์ฐพ์๋ณด๋ ๊ฒ์ ๋๋ค. ์ค์์ ๋ ์คํฐ๋ธ ์ก์ค(Steve Jobs)์ ์ฃฝ์๊ณผ ๊ทธ์ โ์ฃฝ์์ด ๋๊ธฐ ๋ถ์ฌ๊ฐ ๋๋คโ๋ ์คํ ํผ๋ ์ฐ์ค์์ ์๊ฐ์ ๋ฐ์์ง๋ง, ์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ๊ฐ ๊ฐํ์ ๋จ์(intermittent fasting), ์๋ ํธ(Soylent) ๊ฐ์ โ์ ์ฒด ํดํน(body hacking)โ๋ถํฐ ํ๊ฐ์ (psychedelics), ๋ ํดํน(brain hacking)์ ์ด๋ฅด๊ธฐ๊น์ง ์ ์ฐจ โ๋ถ๋ฉธโ์ ์ง์ฐฉํ๋ ๊ฒ์ ๋ชฉ๊ฒฉํ์ต๋๋ค. ์ ์ฌํธ๋จผ, ๋๋ฆฌ ์๋ฆฌ์จ(Larry Ellison), ํผํฐ ํธ(Peter Thiel) ๋ฑ ๋ง์ ๊ฑฐ๋ฌผ๊ธ ์ธ์ฌ๋ค์ด ํญ๋ ธํ(anti-aging) ๊ธฐ์ ์ ํฌ์ํ๊ณ ์์ต๋๋ค.
์ค์์ ๋ ํฌ๋ฆฌ์คํผ(CRISPR), mRNA ๋ฐฑ์ , AI๋ฅผ ํ์ฉํ ์ ์ ์ ์ ํ ๋ถ์(gene folding) ๋ฑ โ์ค์ง์ ์ธ(real)โ ๊ธฐ์ ๋ฐ์ ๊ณผ ํจ๊ป, ๊ณ ์ ์ฐ์ ์ฑ๋ฒ(hyperbaric chamber), ์ฌ์ด๋ ํ ๋ผํผ(sound therapy), ์ ์๊ด ์น๋ฃ(red light therapy) ๊ฐ์ โํฐ๋ฌด๋์๋(ridiculous)โ ์๋๋ค์ด ํผ์ฌ๋์ด ์๋ค๊ณ ๋งํฉ๋๋ค. ๊ทธ๋ ๋ ์์ ์ด ์ง์ ์ฌ๋ฌ ๊ฐ์ง๋ฅผ ์๋ํ๋๋ฐ, ์ผํ๋ฏผ(ketamine)์ โ์ต์ ํ(optimization)โ ๋ชฉ์ ์ผ๋ก ์ฌ์ฉํ ๊ฒฝํ๋ ํฌํจ๋ฉ๋๋ค.
์ค์์ ๊ฐ ์ด๋ฌํ โ๊ฑด๊ฐ ๊ด๋ จโ ์ง์ฐฉ์ ๋ํด ๊ฐ์ฅ ๋นํ์ ์ธ ๋ถ๋ถ์ โ๋ถ์๋ค์ด ์๋ฒฝ์ฑ์ ์ถ๊ตฌํ๋ฉฐ ๋ง๋ ์ ๋๋ ์น๋ฃ๋ฒ์ ๋์ ๋ญ๋นํ๋ ๊ฒโ์ ๋๋ค. ๊ทธ๋ ๋ โ์ฐ๋ฆฌ ๋ชจ๋๊ฐ ๋ ์ค๋ ์ด ์ ์๋ ๊ฐ์ฅ ๊ฐ๋จํ ๋ฐฉ๋ฒ ์ค ํ๋๋ ๋ณดํธ์ ๊ฑด๊ฐ๋ณดํ(universal healthcare)โ์ด๋ผ๋ฉฐ, ํ๊ตญ์ ์๋ก ๋ค์์ต๋๋ค. ๋ฏธ๊ตญ์ ๋ค๋ฅธ ์ ์ง๊ตญ์ ๋นํด ๋ ๋ฐฐ์ ์๋ฃ๋น๋ฅผ ์ง์ถํ๋ฉด์๋ ๊ฒฐ๊ณผ๋ ์ตํ์๊ถ์ด๋ผ๋ ์ ์ ์ง์ ํ๋ฉฐ, ๋ถ์๋ค์ด ๊ฐ์ธ์ ์ธ ๋ถ๋ฉธ์ ์ง์ฐฉํ๋ ๋์ ๋ ํฐ ๋ฌธ๋ช ์ ๊ฑด๊ฐ์๋ ๋ฌด๊ด์ฌํ ํ๋๋ฅผ ๋นํํ์ต๋๋ค.
์ผ๋ถ์์๋ ๋ถ์๋ค์ด ์ด๋ฌํ ๊ธฐ์ ์ โ๊ธฐ๋ํผ๊ทธ(guinea pigs)โ๊ฐ ๋์ด ์ฑ๊ณตํ๋ฉด ๋์ค์๊ฒ๋ ํํ์ด ๋์์ฌ ๊ฒ์ด๋ผ๋ โ๋์ ํจ๊ณผ(trickle-down health)โ ์ฃผ์ฅ์ ํผ์น๊ธฐ๋ ํฉ๋๋ค. ๊ทธ๋ฌ๋ ์ค์์ ๋ ์ด๋ฌํ ์ฃผ์ฅ์ด ์ฃผ๋ก ๋ถ์ ์ธต ์ฌ์ฑ๋ค ์ฌ์ด์์ ํผ์ง GLP-1 ์ฝ๋ฌผ์ฒ๋ผ ์ด๋ฏธ ๋น๋จ๋ณ ํ์๋ค์๊ฒ ์ค๋ซ๋์ ์ฌ์ฉ๋์ด์จ ์น๋ฃ๋ฒ์ ์์ ๋ค๋ง์ โ์๋ก์ด ๋ฐ๊ฒฌโ์ฒ๋ผ ์ฌ๊ธฐ๋ ๊ฒฝํฅ์ด ์์์ ๊ผฌ์ง์ต๋๋ค.
๊ฒฐ๋ก ์ ์ผ๋ก, AI ๋ฐ๊ฐ๊ณผ ์ฅ์ ์ง์ฐฉ์ ์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ ์๋ฆฌํธ๋ค๊ณผ ์ผ๋ฐ ๋์ค ์ฌ์ด์ ๊น์ ๋ฌธํ์ , ๊ฒฝ์ ์ ๋จ์ ์ ์์งํฉ๋๋ค. ๊ธฐ์ ์ ๋ฐ์ ์ด ์ธ๋ฅ์๊ฒ ์ด๋ค ๋ฏธ๋๋ฅผ ๊ฐ์ ธ์ฌ์ง์ ๋ํ ์ง์งํ ๊ณ ๋ฏผ๊ณผ ํจ๊ป, ๊ทธ ํํ๊ณผ ์ํ์ ๊ณต์ ํ๊ฒ ๋ถ๋ฐฐํ๊ณ ํต์ ํ ์ ์๋ ๋ฏผ์ฃผ์ ๊ฑฐ๋ฒ๋์ค ์์คํ ์ ๊ตฌ์ถ์ด ์๊ธํ ๊ณผ์ ๋ก ๋ ์ค๋ฅด๊ณ ์์ต๋๋ค.