English
Based on โStop trying to turn your suffering into a success story | Kate Bowlerโ from Big Think Watch the original video
Beyond the โLessonโ: Embracing Grief When Thereโs No Silver Lining
In a culture increasingly obsessed with self-improvement and turning every setback into a stepping stone, a powerful counter-narrative is emerging. Theologian and author Kate Bowler challenges one of our most deeply ingrained beliefs: that all pain must serve a purpose, that every suffering must yield a lesson. This relentless pursuit of meaning, she argues, can be more damaging than helpful, especially when we are at our most vulnerable.
American society, in particular, has developed an almost pathological need to rationalize suffering. From personal tragedies to global crises, we are conditioned to seek out the โwhyโโnot just for understanding, but for validation. We want to believe that our pain, or the pain of others, isnโt arbitrary. It must be a crucible, a divine test, a precursor to an inevitable triumph. This cultural mania, Bowler contends, has led to the creation of what she vividly calls โpurpose monsters.โ
The Rise of the โPurpose Monsterโ
โPurpose monsters are everywhere,โ Bowler explains, describing the internal and external pressures that hound us when faced with adversity. We find ourselves asking, โIs this gonna work out? Is any of this a divine conspiracy to make me good?โ This isnโt just a casual query; itโs a deep-seated anxiety. We become โitchy,โ as Bowler puts it, when confronted with an unanswered question about suffering. This discomfort, she suggests, โgoes right to our material conditions, right to our fear, right to our ontological insecurity.โ
Ontological insecurity refers to our fundamental unease about the nature of existence itself. We crave order, meaning, and predictability. When bad things happen, especially to good people, it shatters our sense of a just world. We desperately need an explanation for evil, a narrative that restores balance. The idea that pain must teach us something serves as a spiritual balm, a way to impose order on chaos. It allows us to believe that if we know why someone is suffering, we can better cope with the fact of it.
The Harm in Prescriptive Positivity
While the impulse to find meaning is understandable, Bowler warns against its darker side. When we insist that someoneโs pain must be a lesson, we often do more harm than good. Imagine someone experiencing a profound lossโthe death of a loved one, a devastating illness, a catastrophic failure. In their rawest moments of grief, well-meaning individuals might offer platitudes like, โEverything happens for a reason,โ or โYouโll be stronger for this.โ
โWhen someoneโs telling you thereโs a lesson,โ Bowler observes, โitโs as if theyโre saying, โI swear you didnโt lose anything. Look around. Itโs all there.โโ But the grieving person knows, deep in their heart, they are being lied to. A loss is a loss. A wound is a wound. To deny the reality of that absence, to immediately pivot to a narrative of growth or purpose, is to invalidate their experience. It strips away the โpreciousness of the dignity of grief.โ
Grief, in its purest form, is a necessary process. It requires space, acknowledgement, and permission to simply be. When we rush to find the lesson, we bypass this crucial stage, forcing individuals to perform resilience before theyโve even had a chance to mourn. It creates a perverse incentive to package their suffering into an inspiring โsuccess storyโ before theyโve truly processed the depth of their pain.
Meaning Without Reason
One of Bowlerโs most profound insights is the distinction between reason and meaning. โJust because thereโs not a reason,โ she asserts, โdoesnโt mean itโs not meaningful.โ This is a crucial pivot point. We may never understand why certain tragedies strike, why some lives are cut short, or why suffering seems so unfairly distributed. There may be no grand divine conspiracy, no cosmic lesson plan meticulously laid out for us. And accepting that lack of a clear โreasonโ can be incredibly liberating.
It frees us from the exhausting search for justification, from constantly looking backward and โtrying to rearrange the furniture of our past,โ as Bowler describes it. Instead of endlessly replaying events to discern a hidden purpose, we can shift our focus to the present and the future.
Building Anew from the Rubble
The alternative, Bowler suggests, is not to surrender to nihilism, but to embrace a different kind of strength. Itโs about acknowledging the reality of what has been lost, the โfloodsโ that have come and gone, leaving devastation in their wake.
โWhatโs left here for me to make something beautiful?โ This powerful question encapsulates the essence of her message. Itโs not about denying the floods or pretending they brought a hidden gift. Itโs about recognizing that the landscape has changed irrevocably. And from that changed landscape, from the debris and the emptiness, we can still choose to build. We can find meaning not in the cause of the suffering, but in our response to it, in the connections we forge, the compassion we extend, and the new life we create, however different it may be.
In a world that constantly demands a tidy narrative for every hardship, Kate Bowler offers a refreshing and deeply human perspective. She invites us to resist the tyranny of purpose, to grant ourselves and others the sacred space for grief, and to understand that sometimes, there is no lesson, no reasonโonly the dignity of pain and the quiet courage to build something beautiful from what remains.
Based on โWhy Every AI Team Needs Pirates and Architectsโ from Every Watch the original video
The New Frontier of Code: Why Every AI Team Needs Pirates and Architects
Itโs 2026, and the dream of โvibe codingโ is a reality. With powerful AI agents like Codex at our fingertips, the ability to manifest an idea into a functional application in days, or even hours, feels like magic. But as the CEO of Every, a 25-person tech company, recently discovered, the ease of creation doesnโt always translate to the ease of maintenance. Building a groundbreaking app with AI is one thing; keeping that โhot piece of flaming garbageโ from melting down is quite another.
His journey, from the exhilarating launch of an AI-built app to late-night debugging marathons, reveals a critical shift in the software engineering landscape. The future, he argues, isnโt about AI replacing engineers, but about a new, symbiotic team structure: the Pirate and the Architect.
From Vibe Code to โFlaming Garbageโ: The Birth of Proof
The story begins with Proof, an agent-native document editor conceived by Everyโs CEO in his spare time. โThink Google Docs, but made for agents to use,โ he explains. Built entirely in Codex over just ten days, without ever looking at a single line of code, Proof went from a basic concept to a finished MVP. The initial launch was nothing short of spectacular. A viral tweet garnered 1.5 thousand likes and 500,000 views, leading to 4,000 new Proof documents created within the first 48 hours. โWhen I looked at it in the morning the day after launch and saw the stats, I was like, โWeโre going to the moon!โโ
The optimism was intoxicating, a testament to the power of AI-driven development. But the moon landing proved elusive. The initial success quickly gave way to a chaotic struggle. The speaker found himself staying up until 4 AM nightly, literally sleeping with his laptop open, checking the app every few hours โlike some settler on the prairie tending my fire overnight so it wouldnโt go out.โ The โvibe coded sloppy messโ was constantly on the verge of collapse.
This intense, personal crucible forged a profound realization: while AI makes coding incredibly fast, it doesnโt inherently make it robust. Taking a project โfrom a vibe coded sloppy mess to it still has some messy things but it just actually works and I use it every day and it and everyone on the team uses it and itโs starting to grow as a real production appโ required a fundamental shift in approach.
The New Structure: Pirates and Architects
The core insight from this experience is a proposed restructuring of software engineering teams for the AI era. In 2026, the speaker suggests, engineering teams should be built around two distinct, yet interdependent, roles: the Pirate and the Architect.
The Pirateโs Mission: Discovery and Speed
The Pirate is the visionary, the explorer. Their primary job is to โfigure out the product,โ to โmove as fast as possible to explore the territory and make something that people really need.โ Pirates are the masters of vibe coding, unburdened by architectural concerns or the need for a perfectly โwell-oiled machine.โ Their goal is rapid iteration, to โfind something valuable as quickly as they can.โ They are the ones who โown the vision for the product,โ charting new courses and discovering uncharted user needs.
The Architectโs Mandate: Structure and Stability
Once the Pirate has discovered something valuable, thatโs where the Architect steps in. The Architectโs job is to โkeep that process on rails,โ transforming the valuable but often chaotic discoveries of the Pirate into a โmachine that is understandable, extensible, maintainable, that wonโt just go down randomly for no reason that no one can figure out.โ Architects are the guardians of quality, scalability, and long-term viability.
The synergy is crucial. โPirates need architects to keep from making something that just like falls apart,โ the speaker emphasizes. Conversely, โarchitects need pirates to give them the raw material to work with, to make a system thatโs actually worth making.โ
Advice for the Aspiring Pirate
For those who identify as Pirates, eager to leverage AIโs speed, the speaker offers three key lessons:
- Embrace Simplicity: The temptation with vibe coding is to continuously add features. However, โthe best thing you can make is a simple thing that works really well.โ Focusing on the core value and resisting feature creep will lead to more impactful products.
- Learn to โCover Your Tracksโ: Throw It Out and Restart: Once you hit on something truly valuable, be prepared to discard your initial code. Referencing writer Annie Dillard, the speaker advises, โYou neverโฆ want people to see how you got to the idea that you have. You just want to give them the idea.โ The same applies to product development. Vibe-coded prototypes are often โa mess of tracks.โ When a clear, valuable concept emerges, โstart it over.โ Code is cheap to write, and even cheaper to discard. Trying to get an AI agent to refactor a messy codebase is often futile, as โthey get really distracted by whatโs already there.โ A clear sign itโs time to restart: โYou have something thatโs clearly valuable and you have lots of bugs that are piling up that you canโt figure out how to fix no matter what you do.โ
- Seize the Agent-Native Opportunity: Thereโs a โbig opportunity to remake every productivity app for agents.โ Just as Proof reimagined Google Docs for AI agents, the next generation of software, from spreadsheets to presentations, will have agents as their primary users. โIf you start with an agent as your user, the kind of software you build is going to be totally different and really valuable.โ
The Architectโs Enduring Value
For Architects, who might be wondering if their skills are still relevant in an AI-dominated world, the message is one of profound reassurance. โIs software engineering going away as a profession? Do I have any value anymore?โ are common anxieties.
The speakerโs experience with Proof definitively answered this. โI actually fixed a lot of these issues with Proof by pulling in an architect to help me rewrite a few sections of the app from the ground up over the course of a week or so. And thatโs what really fixed the problems.โ
It became โvery clear to me after that experience that an architect using an AI is incredibly powerful and incredibly important and isnโt just going to be replaced by a model in the near future.โ While coding models are powerful, they โstill lack some of the conceptual clarity that a really senior engineer can bring to a project.โ AI tends to make โlots of local changes that all make sense in isolation but if you zoom outโฆ the whole thing doesnโt hang together.โ Real engineers excel at this holistic view, ensuring cohesion and structural integrity. For senior engineers, AI is not a threat, but an amplifier, allowing them to โextend your powersโ and develop โnew expertise that the models donโt have.โ
Building for the Future, Together
Software engineering is undergoing a dramatic transformation. What can be built, how itโs built, and the very structure of teams are evolving rapidly. The essential roles, however, remain. The Pirate, moving at lightning speed to discover whatโs valuable, and the Architect, meticulously crafting those discoveries into a robust, maintainable system.
As for the CEO of Every, his next vibe-coded side project will definitely involve an Architect. โMy entire team will mutiny if I donโt,โ he quips, adding, โand honestly itโs better to do things together. You donโt have to do everything yourself.โ
The future of coding is here, powered by AI. But the journey to the moon, or to a truly stable and successful product, still requires human ingenuity, collaboration, and the perfect blend of daring Pirates and grounded Architects.
Based on โBen Horowitz on AI Anxiety, Big Tech Transitions & The Future of Startups | a16zโ from a16z Watch the original video
The AI Earthquake: Ben Horowitz on Shifting Paradigms, Infrastructure Crises, and the Path to a Scarier โ or Better โ Future
The world, as we know it, is undergoing a profound transformation. At the epicenter of this seismic shift is Artificial Intelligence, an innovation so powerful itโs rewriting the fundamental โlaws of physicsโ for businesses, governments, and even the very fabric of human interaction. Ben Horowitz, co-founder of the influential venture capital firm a16z, offers a bracingly candid and ultimately optimistic perspective on this new era, navigating the anxieties, the infrastructure bottlenecks, and the surprising solutions that might emerge from the chaos.
Horowitz, a veteran of tech booms and busts, understands disruption. From the dot-com collapse to the current AI revolution, heโs witnessed markets buckle and companies pivot. But AI, he argues, is different. Itโs not just a technological upgrade; itโs a foundational change that demands a complete re-evaluation of strategy, investment, and even our understanding of what constitutes a โjobโ or โtruth.โ
The New Laws of the AI Universe: Where Money (Finally) Talks
For decades, a cardinal rule in technology was that you couldnโt simply โthrow money at the problem.โ If your product was two years behind a competitor, hiring a thousand engineers wouldnโt close the gap; it would likely just create more complexity, a concept famously dubbed โThe Mythical Man-Month.โ That axiom, Horowitz declares, is โno longer true.โ
โIf you have enough money and some good data, you can buy enough GPUs and solve basically anything in software,โ he explains. This shift means that raw capital and computational power, rather than solely human ingenuity in a linear fashion, can now directly accelerate development and overcome competitive lags.
Compounding this, another long-held truth has evaporated: the power of customer lock-in. Historically, if you had the customer, you had a multi-layered advantage โ migration pain, data lock-in, user interface familiarity. โThose are pretty much gone,โ Horowitz asserts. Code is easily replicable, data can be moved with increasing ease, and critically, the primary interface with software is no longer solely human. โItโs not even going to be a human talking to your software,โ he notes, โitโs going to be an AI. And AIs are really flexible on how they use user interfaces.โ This means switching costs plummet, and established market positions become far more precarious.
The implications for existing companies are stark. A product lifecycle that once offered a 5-10 year runway might now be compressed to โfive weeks.โ Legacy companies, those founded โpre-AI,โ face an existential crisis. They must fundamentally redefine their value proposition beyond features that can be easily replicated by AI, or risk being โslaughtered in the valuation game.โ Horowitz points to a portfolio company like Nan, a travel firm, as an example of resilience. Despite the โSaaS apocalypseโ and the seeming vulnerability of travel tech, Nan thrives because it possesses deeply embedded, complex relationships with airlines, hotels, and budgeting systems, alongside an invaluable channel to the niche โtravel managerโ customer โ a channel no AI giant is rushing to build. These โexplicit relationshipsโ and intricate operational layers become the new defensible moats.
The Looming Infrastructure Bottleneck: A Nation on E
Beyond the corporate boardroom, AIโs voracious appetite for resources is exposing a critical vulnerability at the national level: infrastructure. Horowitz paints a grim picture of Americaโs readiness for the AI age. โAmericaโs got to rebuild its entire infrastructure like right now,โ he warns. โWe donโt have enough rare earth minerals. We donโt have enough electricity. We donโt have enough manufacturing capacity.โ
The problem isnโt just about chips, though Nvidia is making strides. Itโs a systemic bottleneck across the entire supply chain. โNvidia will make enough chips, but then we wonโt have enough memory,โ he predicts. โAlmost everything is the bottleneck.โ The urgency is palpable: โWeโre pretty much out of electricity now in the United States. Like not 12 months from now, like right now.โ
This isnโt the dot-com bubbleโs โdark fiberโ problem, where infrastructure was built but underutilized. Today, โall the GPUs are hot, theyโre all lit right now.โ The demand is โstraight vertical,โ but the capacity to build new power plants, DRAM factories (which take five years), or even essential components like power transformers (a technology largely unchanged since the invention of electricity, prompting a16z to invest in a transformer company) is simply not keeping pace. The โChina graphโ of capacity building, Horowitz implies, is ascending steeply, while the โUS graphโ lags. This macro-level challenge demands massive investment and strategic foresight, a scale of funding that Horowitz suggests influenced a16zโs recent multi-billion dollar fundraise.
The Crisis of Trust and Identity: Cryptoโs Unexpected Role
As AI becomes ubiquitous, a more insidious problem emerges: the erosion of trust and the blurring of reality. Horowitz vividly describes a chilling scenario: โOh my god, somebodyโs going to go on a Zoom. Theyโre gonna itโs going to be AI me and theyโre going to tell my finance team to wire like, you know, $500 million to Nigeria and and thatโs going to be a problem.โ
The proliferation of AI-generated content โ personalized emails, deepfake videos, hyper-realistic voice calls โ threatens to render all digital communication โcompletely unusable.โ Weโre entering an era where distinguishing human from bot, truth from fabrication, becomes nearly impossible. The old email filtering tricks, like spotting a misspelled name, are useless when AI can perfectly personalize thousands of messages.
This is where crypto, a technology often seen as separate from AI, becomes indispensable. Horowitz identifies several critical roles for blockchain and cryptographic solutions:
- Human Verification: โAre you a human or are you a bot?โ This fundamental question will underpin social media, dating apps, and professional interactions. Proving oneโs humanity and unique identity cryptographically will be paramount.
- Content Provenance: To combat deepfakes and manipulated media, content will need โcryptographically strong indication assigned piece of content that says okay yeah I made this or this is like really a video of me Marco Rubio giving a speech. This isnโt something that somebody faked.โ
- Source of Truth: In a world saturated with AI-generated information, who do you trust? Google? Meta? The government? Horowitz argues for the โmathematical game theoretic properties of the blockchainโ as a more reliable, decentralized arbiter of truth.
- Economic Identity: Beyond identity verification, crypto can solve practical problems like fraud prevention in Universal Basic Income (UBI) programs. The governmentโs struggle to distribute stimulus funds, with an estimated โ$450 billion got stolen,โ highlights the need for a secure, universal digital address for money.
- AI as Economic Actors: As AIs evolve, they will need to transact, earn money, and function as economic agents. โHow do I make money as an AI? How does somebody send me money? Can I be a merchant, a credit card merchant if Iโm not a human?โ Horowitz asks. A โbare instrument on the internet,โ or โinternet money,โ will be crucial for AIs to participate in the economy, a role perfectly suited for crypto.
The synergy is clear: AI creates profound problems of trust and identity, and crypto offers foundational solutions.
The Future of Work and Wealth: Scary or Spectacular?
The anxieties surrounding AI often coalesce around the future of work. Will white-collar jobs disappear? Will a few giant companies monopolize intelligence? Horowitz, quoting Yogi Berra, acknowledges the difficulty of predicting the future, but offers two contrasting scenarios for venture capital and, by extension, the broader economy.
One path echoes the Industrial Revolution, where an initial flurry of innovation (like 300 early auto companies) consolidated into a few dominant players (the โBig Threeโ). In this future, โnew companies are just harder,โ and venture capital might become more akin to traditional banking, funding a small number of colossal entities.
However, Horowitz leans towards a more optimistic, decentralized vision. He draws parallels to developing nations where โeverybodyโs an entrepreneur,โ buying and selling, constantly creating value. The truly transformative aspect of AI, he believes, is its ability to unlock human potential on an unprecedented scale. โThe cool thing is that now 8 billion people that might have an idea in their head can get it out of their head.โ Whether itโs writing music, making a movie, or developing code, the โcapital gateโ and the โidea gateโ are disappearing. This could lead to a venture capital world that is โmuch bigger and much more exciting because everybody in the world is an entrepreneur.โ
Historically, technological shifts have always been accompanied by fear. In 1789, 98% of Americans were farmers; today, those jobs are largely gone, replaced by professions that would have seemed โridiculousโ to an 18th-century agrarian. Horowitz refutes the dystopian narrative, including the โold man yells at cloudโ image of Bernie Sanders interviewing an AI. He cites John Maynard Keynesโs famous miscalculation: Keynes predicted that with abundance, people would work only 15 hours a week. What he didnโt foresee, Horowitz explains, was humanityโs โunbelievableโ capacity to invent new โneedsโ โ not just one car, but a car for every person; not just food, but elaborate tasting menus; not just shelter, but awesome vacations and access to boundless information.
โThe history of technology is things have always gotten better,โ Horowitz concludes. While the transition is undeniably โscary,โ he believes that in 15 years, โeverybody in America and probably around the world is going to live better thanโฆ anybody did in 1980.โ This isnโt to dismiss the challenges or the need for careful navigation, but itโs a powerful statement of faith in human ingenuity to adapt, create, and ultimately thrive in an AI-powered world, even if the โlaws of physicsโ are being rewritten before our very eyes.
Based on โThe Workers Letting A.I. Do Their Jobsโ from New York Times Podcasts Watch the original video
From Coders to Conjurers: How AI is Radically Reshaping Software Development
For years, the world has grappled with the looming question: how will artificial intelligence reshape our lives and work? The answers have spanned the spectrum from utopian salvation to apocalyptic doom. Yet, amidst the speculation, a quiet revolution has been unfolding at the very heart of the tech industry, among the software developers who build the digital world around us. These are the people at the vanguard, not just contemplating AIโs impact, but actively inviting it into their daily tasks, transforming their jobs in ways both exhilarating and unsettling.
Legendary tech reporter Clive Thompson, a long-time observer of Silicon Valley, recently embarked on an extensive survey of this pivotal shift, interviewing some 75 software developers across the country. His findings, shared in a conversation with Natalie Kitroof for The Daily, reveal a profound and rapid transformation: coders are, surprisingly, coding less โ and in some cases, not at all.
The Unthinkable: Coders Not Coding
Thompsonโs reporting unearthed a stark reality: a significant majority of the developers he spoke with are now outsourcing much of their day-to-day programming to AI. This isnโt a gradual shift; itโs a โsea changeโ that has accelerated dramatically in the last six months, as AI coding tools have become more sophisticated and gained the trust of even skeptical programmers. โSome of them are writing no code at all,โ Thompson reveals, describing a landscape vastly different from just a year or two ago.
Initially, Thompson wondered if this outsourcing of a core creative task would be โdeflatting or demoralizing.โ For decades, he knew, developers derived immense pleasure from solving the intricate puzzles of code. Yet, the reality was overwhelmingly positive. The great majority were โjazzed and excitedโ about the new powers AI bestowed upon them. They could now articulate their desired outcome in plain language and, within minutes, receive working code.
โTheyโve always loved building things,โ Thompson explains. โThatโs the fun of being a developer is you take an idea you have and through sweat and work you turn these magic words into a machine that does things for you.โ Even without writing the code themselves, they still feel like โsorcerersโ or โconjurers,โ bringing ideas into being with unprecedented speed. This newfound โloop of successโ is, for many, even more satisfying because the AI operates faster than they ever could.
A Leap in Abstraction: From Rewiring to AI Swarms
To grasp the magnitude of this change, one must understand the history of software development itself. For 50 to 60 years, the field has continuously sought easier ways to communicate with computers, moving ever closer to human language.
In the 1940s, programming the first computers like ENIAC involved physically rewiring the machine โ a highly โlabor-intensiveโ process where women literally crawled inside to create new logical systems. As computers became industrialized, this proved impractical. Early programming languages emerged, translating typed commands into digital instructions. These were still incredibly arduous, requiring developers to micro-manage every detail, like โjuggling 900 balls just to get it do the most basic piece of math.โ
Over the decades, coders consistently sought to โautomate some of that labor,โ a process they humorously call โadding a layer of abstraction.โ Each new layer made coding easier, less โfinicky,โ and more akin to human language. Whatโs happening now with AI, Thompson asserts, is โprobably the biggest change itโs undergone yet.โ
The modern AI coding paradigm involves โteams of agents.โ A human developer might ask a main AI agent to โwrite this feature for me.โ The AI then devises a plan and โspawns other sub agents to do different parts of the task.โ One agent writes the code, another tests it, and if errors arise (which they almost always do initially), yet another agent analyzes the error messages, refines the code, and re-tests in a continuous loop. Only when the code passes all tests is it presented back to the human. This โlittle team of agents all working in a swarmโ signifies an enormous leap in automation and efficiency.
Speed, Scale, and the โSteve Jobsโ Effect
Practically, this translates into dramatic accelerations. Small startups, often two-person operations, report moving โup to 20 times fasterโ than they would have just two years ago. A customer request for a new feature that once took a full day can now be completed in โmaybe half an hour.โ
However, this speed varies with scale. While a startup building entirely new code can leverage AI fully, larger, more mature companies like Google operate with more caution. At Google, AI might write โ40 or 50%โ of the code, speeding up their overall โmetabolismโ by about 10%. Yet, as Google developers noted, โ10% for a company our size, thatโs a huge win.โ
So, if AI is doing the coding, what do developers actually do? Their job, in essence, becomes that of an architect rather than a construction worker. They focus on โwhat the software ought to be doing,โ spending more time on experimental iteration. Many developers told Thompson that with AI, they can โrun through like 10 different possibilities and Iโll pick the absolute best one.โ Several even likened the experience to โbeing Steve Jobsโ โ delegating the creation of multiple designs and then choosing the optimal one.
On a deeper level, their primary activity is โjust talking.โ They engage in extensive conversations with AI, a process that demands extreme clarity. Thompson found that this necessity has, surprisingly, made many developers better communicators, not just with AI, but โto the world in general.โ As one developer, Manu Eert, jokingly but insightfully put it, โI wonder if weโre finally teaching all the nerds empathy because theyโre having to do that.โ
The Art of โPunishingโ AI
Manu Eertโs experience with AI offers a fascinating glimpse into this new human-AI dynamic. Initially skeptical about AIโs reliability, he gained confidence as it proved its capabilities. Eert and others have settled on a โSocratic dialogueโ technique: they instruct the AI to โask me questions about how this software feature should work.โ This interview process helps the human coder clarify their own vision before the AI takes over.
But AI, despite its prowess, can โmisbehave.โ Eert recounted instances where the AI would skip crucial tests, declaring them unimportant. His solution? โPunishingโ the AI. This involves writing โvery stern list of instructions like a a ten commandmentsโ in all caps, which the AI must consult before every action. These commands include emotionally charged language like, โif you donโt do these tests I will be fired,โ and declaring failure โunacceptable and embarrassing.โ
Thompson explains the logic: large language models understand meaning based on context. Words like โembarrassingโ signal a โbad neighborhoodโ in the semantic landscape, raising the stakes for the AI. So, yes, emotional and even harsh language appears to have an effect.
The Shadow Side: Deskilling and Job Loss
Despite the excitement, developers harbor significant concerns. A primary fear is โdeskillingโ โ losing their deep code sense by relying too heavily on AI. More senior developers, who learned to code manually, recognize when AI produces โflabbyโ or incorrect code. But they worry about โthe next generationโ lacking this fundamental understanding. Piaan, a newer developer whose employers mandated AI tool use, explicitly told Thompson she felt her own coding knowledge โdegrading.โ
This raises a fundamental debate: is losing coding skills a problem if AI can do it better? One camp argues that AI will be more reliable, avoiding โstupid human mistakesโ in what is often โrote and tediousโ work. The other, smaller but vocal, cohort warns that while AI-generated code might look good now, it could create โsubtle bugsโ and โnasty interactionsโ down the line, leading to a โhuge messโ in a few years.
The specter of job loss is equally potent. Thompson points to research by Eric Benjolson at Stanford, showing a 16% drop in software developer job postings and hirings in the last year alone. This is happening when AI tools are just โgoing from a crawl to a walk to a run.โ The capitalist drive for efficiency means companies will inevitably seek to replace expensive human developers with AI, leading to โsofteningโ demand for junior hires and the potential disappearance of entire job categories, much like the type setter.
Implications for the โRest of Usโ: The Paradox of Automation
For those outside of coding, AIโs impact might seem more distant, but developers are experiencing a paradox that could resonate across other white-collar fields: the โhardโ technical skills they spent years acquiring are often the โeasiest things to automate.โ The truly โhuman skillsโ โ talking to colleagues and customers, setting priorities, devising strategy, understanding what should be built โ remain beyond AIโs grasp.
Thompson draws a historical parallel: in the 1980s, chess seemed incredibly difficult for computers, while speaking seemed easy. Yet, computers conquered chess in the โ90s, taking two more decades to master human-like speech. This suggests that what we perceive as our most valuable, complex skills might not be what AI automates first.
The pace of this transformation for the broader economy, however, might be slower than the โsuper speedโ cultural acceleration we perceive. Thompson reminds us of the advent of personal computers in the 80s and 90s. Initially, they didnโt dramatically increase corporate productivity. It took companies years to โreorganize the way it did business around the computers,โ fundamentally changing workflows and information flow, before efficiency gains were seen. He suspects AI will follow a similar, longer timeline for widespread industrial impact.
The Upside: Ubiquitous Software and Unforeseen Wonders
Despite the fears, Thompson envisions significant upsides for society. The world, he argues, is โhorribly underserved by technology.โ He cites the example of a โ$50 million concrete mixing firmโ that, despite its size, runs on โthree Excel spreadsheets on a Windows XP computerโ because it canโt afford a team of custom software developers. If AI makes software creation cheaper and faster, a single developer could build and maintain custom solutions for such businesses at a fraction of the cost, leading to โactually good ways in everyday life for a lot of people at work.โ
Ultimately, software will stop being โprecious and rareโ and become โubiquitous,โ much like paper or word processing. Paper, once rare, became cheap and led to unexpected innovations like Post-it notes. Similarly, the Macintosh made word processing accessible, leading to an explosion of creative, if sometimes โugly,โ documents like riot-girl zines โ an outcome no one in 1983 could have predicted.
โMore is not just more. More is different,โ Thompson asserts. As software becomes โtrivially easy to summon into being,โ new behaviors, both โmalignโ and โdelicious,โ will emerge. It will โcatalyze all of the human desiresโ โ from Shakespearean sonnets to disinformation, from medical record analysis to wholesale cheating.
The future, Thompson concludes, will be โincredibly weird.โ This transformation, like those before it, will reflect us โ our desires, our creativity, our flaws. It will be both awesome and potentially bad, but undeniably, a powerful mirror held up to humanity.
ํ๊ตญ์ด
โStop trying to turn your suffering into a success story | Kate Bowlerโ โ Big Think ๊ธฐ๋ฐ ๊ธฐ์ฌ ์๋ณธ ์์ ๋ณด๊ธฐ
๊ณ ํต์ ์ฑ๊ณต ์คํ ๋ฆฌ๋ก ๋ง๋ค๋ ค ์ ์ฐ์ง ๋ง์ธ์: ์ฌํ์ ์กด์์ฑ์ ์ธ์ ํ ๋ ๋น๋ก์ ํผ์ด๋๋ ์๋ฏธ
ํ๋ ์ฌํ๋ ๊ณ ํต์ ๋ํ๋ ํน์ ํ ๋ฐฉ์์ ๊ฐ์ํ๋ ๊ฒฝํฅ์ด ์์ต๋๋ค. ์ฐ๋ฆฌ๊ฐ ๊ฒช๋ ๋ชจ๋ ์ํ์ ๋ฐ๋์ ์ด๋ค โ๊ตํโ์ ํ๊ณ ์์ด์ผ ํ๋ฉฐ, ๋ชจ๋ ๊ฒฝํ์ โํน๋ณํ ๋ชฉ์ โ์ ์ํด ์กด์ฌํด์ผ ํ๋ค๋ ์๋ฐ ๋ง์ ๋๋ค. ๋ฒ ์คํธ์ ๋ฌ ์๊ฐ์ด์ ์ ๋ช ํ ์ ํ์ ์ผ์ดํธ ๋ณด์ธ๋ฌ(Kate Bowler)๋ ๋น ์ฝํฌ(Big Think) ๊ฐ์ฐ์์ ์ด๋ฌํ ๋ฌธํ์ ๊ฐ๋ฐ์ ๋ํด ๋ ์นด๋ก์ด ํต์ฐฐ์ ์ ์ํ๋ฉฐ, ๊ณ ํต์ ๋ฌด์กฐ๊ฑด์ ์ธ ์ฑ๊ณต ์คํ ๋ฆฌ๋ก ์ ํํ๋ ค๋ ์๋์์ ๋ฒ์ด๋์ผ ํ๋ค๊ณ ์ฃผ์ฅํฉ๋๋ค. ๊ทธ๋ ๋ ๊ณ ํต์ ์๋ฏธ๋ฅผ ๊ฐ์ํ๊ธฐ๋ณด๋ค, ์ฌํ์ ์กด์์ฑ์ ์ธ์ ํ๊ณ ์๋ก์ด ์๋ฆ๋ค์์ ์ฐพ์ ๋์๋ ๊ฒ์ด ์ง์ ํ ํ๋ณต์ ๊ธธ์์ ๊ฐ์กฐํฉ๋๋ค.
โ๋ชฉ์ ๊ดด๋ฌผโ์ ํ์: ๊ณ ํต์ ์๋ฏธ๋ฅผ ๋ถ์ฌํ๋ ค๋ ์๋ฐ
๋ณด์ธ๋ฌ๋ ํนํ ๋ฏธ๊ตญ ๋ฌธํ๊ฐ โ๋ชจ๋ ๊ณ ํต์ ๊ตํ์ด์ด์ผ ํ๋คโ๋ ์๊ฐ์ ์ฌ๋ก์กํ ์๋ค๊ณ ์ง์ ํฉ๋๋ค. โ๋ชฉ์ (purpose)โ์ด๋ผ๋ ๊ฐ๋ ์ ์ด์ ๊ฐ์ธ์ ์ถ์ ๋์ด์ ๋ฌธํ์ ๊ด๊ธฐ(cultural mania)๊ฐ ๋์ด, ๋ชจ๋ ์ฌ๋์ด ํน๋ณํ ์ด์ ๋ฅผ ๊ฐ์ง๊ณ ๋ฌด์ธ๊ฐ๋ฅผ ํด์ผ ํ๋ค๋ ์๋ฐ์ผ๋ก ์์ฉํ๋ค๋ ๊ฒ์ ๋๋ค. ๊ทธ๋ ๋ ์ด๋ฌํ ๋ชฉ์ ๊ฐ๋ฐ๊ณผ ๊ทธ ์ฌํ๋ฅผ โ๋ชฉ์ ๊ดด๋ฌผ(purpose monsters)์ ์ฐฝ์กฐโ๋ผ๊ณ ๋ถ๋ฆ ๋๋ค. ์ด โ๋ชฉ์ ๊ดด๋ฌผโ๋ค์ ์ฐ๋ฆฌ ์ฃผ๋ณ ์ด๋์๋ ์กด์ฌํ๋ฉฐ, ์ฐ๋ฆฌ์๊ฒ ๋์์์ด ๋ฌป์ต๋๋ค. โ์ด ๋ชจ๋ ์ผ์ด ์ ๋ ๊น? ์ด ๋ชจ๋ ๊ฒ์ด ๋๋ฅผ ๋ ๋์ ์ฌ๋์ผ๋ก ๋ง๋ค๊ธฐ ์ํ ์ ์ ์๋ชจ์ผ๊น?โ
์ฐ๋ฆฌ๋ ์ด๋ฌํ ์ง๋ฌธ์ ๋ต์ ์ฐพ์ง ๋ชปํ ๋๋ฉด ์ฌํ ๋ถ์๊ฐ์ ๋๋๋๋ค. ์ด๋ ์ฐ๋ฆฌ์ ๋ฌผ์ง์ ์กฐ๊ฑด, ๊น์ ๋๋ ค์, ๊ทธ๋ฆฌ๊ณ ๋ฌด์๋ณด๋ค๋ โ์กด์ฌ๋ก ์ ๋ถ์์ ์ฑ(Ontological Insecurity)โ๊ณผ ์ง๊ฒฐ๋๊ธฐ ๋๋ฌธ์ ๋๋ค. ์กด์ฌ๋ก ์ ๋ถ์์ ์ฑ์ด๋ ์ฐ๋ฆฌ์ ์กด์ฌ ๊ธฐ๋ฐ, ์ฆ ์ด ์ธ์์์ ๋ด๊ฐ ๋๊ตฌ์ด๋ฉฐ ์ด๋ค ์์น์ ์๋์ง์ ๋ํ ๊ทผ๋ณธ์ ์ธ ํ์ ์ด ํ๋ค๋ฆฌ๋ ์ํ๋ฅผ ์๋ฏธํฉ๋๋ค. ๊ณ ํต ์์์ ์ฐ๋ฆฌ๋ ์ด๋ฌํ ๋ถ์์ ์ฑ์ ๋์ฑ ๊ฐํ๊ฒ ๋๋ผ๋ฉฐ, ๋ฏธ์คํฐ๋ฆฌ์ ๋ํ ๋ต์ ๊ฐ๊ตฌํ๊ฒ ๋ฉ๋๋ค.
๋ฏธ์คํฐ๋ฆฌ์ ๋ํ ๊ฐ์ฆ: ์ ์ ํ ์ฌ๋์๊ฒ ๋์ ์ผ์ด ์ผ์ด๋๋๊ฐ?
๊ณ ํต์ด ์ฐ๋ฆฌ์๊ฒ ๋ฌด์ธ๊ฐ๋ฅผ ๊ฐ๋ฅด์ณ์ผ ํ๋ค๋ ์๊ฐ์ ์์ ์ผ๋ก ๋ฏธ์คํฐ๋ฆฌ์ ๋ํ ๋ต์ ์ฐพ๊ณ ์ ํ๋ ์ธ๊ฐ์ ๋ณธ๋ฅ์์ ๋น๋กฏ๋ฉ๋๋ค. ์ธ๋ฅ ์ญ์ฌ์ ๊ฐ์ฅ ์ค๋๋ ์ง๋ฌธ ์ค ํ๋์ธ โ์ ์ ํ ์ฌ๋์๊ฒ ๋์ ์ผ์ด ์ผ์ด๋๋๊ฐ?โ์ ๋ํ ํด๋ต์ ์ฐพ์ผ๋ ค๋ ์๋์ธ ์ ์ ๋๋ค. ์ฐ๋ฆฌ๋ ์ ์ ๋ํ ์ค๋ช ์ ํ์๋ก ํ๊ณ , ๋๊ตฐ๊ฐ ๊ณ ํต๋ฐ๋ ๋ชจ์ต์ ๋ณผ ๋๋ฉด โ๋๋ ๊ทธ ์ด์ ๋ฅผ ์์โ๋ผ๊ณ ๋งํ ์ ์๊ธฐ๋ฅผ ์ํฉ๋๋ค. ์ด๋ ๊ณ ํต๋ฐ๋ ์ด์๊ฒ ์๋ก๋ฅผ ์ฃผ๋ ค๋ ์ ์์์ ๋น๋กฏ๋ ์๋ ์์ง๋ง, ๋์์ ์ฐ๋ฆฌ ์์ ์ ๋ถ์์ ํด์ํ๋ ค๋ ๋ฌด์์์ ์ธ ์๋์ด๊ธฐ๋ ํฉ๋๋ค.
๊ทธ๋ฌ๋ ๋ณด์ธ๋ฌ๋ ์ฌ๊ธฐ์ ํจ์ ์ด ์๋ค๊ณ ๋งํฉ๋๋ค. โ๋จ์ํ ์ด๋ค โ์ด์ (reason)โ๊ฐ ์๋ค๊ณ ํด์ ๊ทธ๊ฒ์ด โ์๋ฏธ(meaning)โ๊ฐ ์๋ ๊ฒ์ ์๋๋ค.โ ๊ณ ํต์ ์์ธ์ด๋ ๋ชฉ์ ์ ๋ช ํํ ์ค๋ช ํ ์ ์์์ง๋ผ๋, ๊ทธ ๊ฒฝํ ์์ฒด๊ฐ ์ฐ๋ฆฌ์๊ฒ ๊น์ ์๋ฏธ๋ฅผ ๋จ๊ธธ ์ ์๋ค๋ ํต์ฐฐ์ ๋๋ค. ์ฐ๋ฆฌ๋ ์ข ์ข ๊ณ ํต์ โ์ด์ โ๋ฅผ ์์ง ๋ชปํด๋, ๊ทธ ๊ณ ํต์ ํตํด ์ถ์ ์ค์ํ โ์๋ฏธโ๋ฅผ ๋ฐ๊ฒฌํ๊ฑฐ๋ ์๋ก์ด ๊ฐ์น๋ฅผ ๋ถ์ฌํ๊ฒ ๋ฉ๋๋ค.
์ด์ ๊ฐ ์๋ค๊ณ ์๋ฏธ๊ฐ ์๋ ๊ฒ์ ์๋๋ค: ์ฌํ์ ์กด์์ฑ
ํ์ธ์ ๊ณ ํต ์์์ โ์ด ๊ณ ํต์ ๋์๊ฒ ๊ตํ์ ์ค ๊ฑฐ์ผโ๋ผ๊ณ ๋งํ๋ ๊ฒ์ ๋ง์น โ๋งน์ธ์ปจ๋, ๋๋ ์๋ฌด๊ฒ๋ ์์ง ์์์ด. ์ฃผ๋ณ์ ๋ด. ๋ชจ๋ ๊ฒ์ด ๊ทธ๋๋ก ์์ดโ๋ผ๊ณ ๋งํ๋ ๊ฒ๊ณผ ๊ฐ์ต๋๋ค. ๊ทธ๋ฌ๋ ์ฐ๋ฆฌ๋ ๋ง์์์ผ๋ก ์ด๊ฒ์ด ๊ฑฐ์ง๋ง์์ ์๋๋ค. ์ง์ ์ผ๋ก ๋ฌด์ธ๊ฐ๋ฅผ ์์๊ณ , ๊ทธ ์์ค๊ฐ์ ๊ฒฐ์ฝ ์ฝ๊ฒ ๋ฉ์์ง์ง ์์ต๋๋ค. ์ด๋ฌํ ๋ง์ ๊ณ ํต๋ฐ๋ ์ฌ๋์ ์ฌํ๊ณผ ์์ค๊ฐ์ ๋ถ์ ํ๊ณ , ๊ทธ๋ค์ด ๊ฒช๋ ํ์ค์ ์ธ๋ฉดํ๊ฒ ๋ง๋ญ๋๋ค.
๋ณด์ธ๋ฌ๋ โ์ฌํ์ ์กด์์ฑ(the dignity of grief)์ ์ฐ๋ฆฌ ์์ ์๊ฒ ํ๋ฝํ๋ ๊ฒ์ ์ข๊ณ ๋ ํ์์ โ์ด๋ผ๊ณ ๊ฐ์กฐํฉ๋๋ค. ์ฌํ์ ์ธ๊ฐ์ด๋ผ๋ฉด ๋๊ตฌ๋ ๊ฒช๋ ์์ฐ์ค๋ฌ์ด ๊ฐ์ ์ด๋ฉฐ, ์์ค์ ๋ํ ์ ๋นํ ๋ฐ์์ ๋๋ค. ์ฌํ์ ์ต๋๋ฅด๊ฑฐ๋, ๋น ๋ฅด๊ฒ ๊ธ์ ์ ์ธ ๊ตํ์ผ๋ก ์ ํํ๋ ค ์ ์ฐ๋ ๋์ , ๊ทธ ์ฌํ ์์ฒด๋ฅผ ์จ์ ํ ๋๋ผ๊ณ ์กด์คํ๋ ๊ฒ์ด ์ค์ํฉ๋๋ค. ์ฌํ์ ์กด์์ฑ์ ์ธ์ ํ ๋ ๋น๋ก์ ์ฐ๋ฆฌ๋ ์ง์ ํ ์น์ ์ ํ๋ณต์ ๊ธธ๋ก ๋์๊ฐ ์ ์์ต๋๋ค.
๊ณผ๊ฑฐ์ ๊ฐ๊ตฌ๋ฅผ ์ฌ๋ฐฐ์นํ๋ ๋์ : ์๋ฆ๋ค์์ ๋ง๋๋ ์ฌ์
๊ณ ํต์ โ์ด์ โ๋ฅผ ๋์์์ด ์ฐพ์ ํค๋งค๋ ๊ฒ์ ๊ณผ๊ฑฐ๋ฅผ ํฅํด ๋์์ด ๋ค๋์๋ณด๊ณ , ๋ง์น ๊ณผ๊ฑฐ์ ๊ฐ๊ตฌ๋ฅผ ์ฌ๋ฐฐ์นํ๋ ค๋ ๋ฏ์ด ์ ์ฐ๋ ๊ฒ๊ณผ ๊ฐ์ต๋๋ค. ์ด๋ ์ฐ๋ฆฌ๋ฅผ ๊ณผ๊ฑฐ์ ๋ฌถ์ด๋๊ณ ํ์ฌ์ ๋ฏธ๋๋ก ๋์๊ฐ๋ ๊ฒ์ ๋ฐฉํดํฉ๋๋ค. ๋ณด์ธ๋ฌ๋ ์ด๋ฌํ ์๋ ๋์ , ์ ์ ๋ฉ์ถฐ ์์ โ์ข์, ํ์๊ฐ ์ง๋๊ฐ์ด. ์ด์ ์ฌ๊ธฐ์ ๋ฌด์์ด ๋จ์ ๋ด๊ฐ ์๋ฆ๋ค์ด ๊ฒ์ ๋ง๋ค ์ ์์๊น?โ๋ผ๊ณ ์ง๋ฌธํด์ผ ํ๋ค๊ณ ์ ์ํฉ๋๋ค.
์ด ์ง๋ฌธ์ ๊ณ ํต์ ์์ธ์ ํํค์น๋ ๋ฐ ์ง์คํ๋ ๋์ , ๊ณ ํต ์ดํ์ ์ถ์์ ๋ฌด์์ ํ ์ ์์์ง์ ์ด์ ์ ๋ง์ถฅ๋๋ค. ์์ค๊ณผ ํ๊ดด ์ดํ์๋ ๋จ์ ์๋ ๊ฒ๋ค์ ๋ฐ๊ฒฌํ๊ณ , ๊ทธ๊ฒ๋ค์ ๋ฐํ์ผ๋ก ์๋ก์ด ์๋ฏธ์ ์๋ฆ๋ค์์ ์ฐฝ์กฐํ๋ ค๋ ์์ง๋ฅผ ๋ด๊ณ ์์ต๋๋ค. ์ฌํ์ ๋ถ์ ํ์ง ์์ผ๋ฉด์๋, ๊ณผ๊ฑฐ์ ๊ฐํ์ง ์๊ณ ํ์ฌ์ ํ์ค์ ๋ฐ์๋ค์ฌ ๋ฏธ๋๋ฅผ ํฅํด ๋์๊ฐ๋ ์ฉ๊ธฐ ์๋ ํ๋์ ๋๋ค.
์ผ์ดํธ ๋ณด์ธ๋ฌ์ ๋ฉ์์ง๋ ๊ณ ํต์ ๋ํ๋ ์ฐ๋ฆฌ์ ์๊ฐ์ ๊ทผ๋ณธ์ ์ผ๋ก ์ฌ๊ณ ํ๊ฒ ํฉ๋๋ค. ๋ชจ๋ ๊ณ ํต์ด ๋ฐ๋์ โ๊ตํโ์ด๋ โ๋ชฉ์ โ์ ๊ฐ์ ธ์ผ ํ๋ค๋ ๊ฐ๋ฐ์์ ๋ฒ์ด๋, ์ฌํ์ ์กด์์ฑ์ ์ธ์ ํ๊ณ ์์ค์ ์ํ์ ์๋ ๊ทธ๋๋ก ๋ฐ์๋ค์ด๋ ๊ฒ์ด ์ค์ํฉ๋๋ค. ๊ทธ๋ฆฌ๊ณ ๊ทธ ์ฌํ ์์์ ์ฐ๋ฆฌ๊ฐ ๊ฐ์ง ๊ฒ๋ค์ ๋ค์ ๋ฐ๊ฒฌํ๊ณ , ์๋ก์ด ๋ฐฉ์์ผ๋ก ์๋ฆ๋ค์์ ๋ง๋ค์ด๊ฐ๋ ์ฌ์ ์ด์ผ๋ง๋ก ์ง์ ํ ํ๋ณต๊ณผ ์๋ฏธ ๋ฐ๊ฒฌ์ ๊ธธ์์ ์ผ๊นจ์์ค๋๋ค. ๊ณ ํต์ ์ฐ๋ฆฌ๋ฅผ ๊ฐ์ํ๋ โ๋ชฉ์ ๊ดด๋ฌผโ์ด ์๋๋ผ, ์๋ก์ด ๊ฐ๋ฅ์ฑ์ ํ์ํ๊ฒ ํ๋ ๋ณต์กํ๊ณ ๋ ๊น์ ์ถ์ ์ผ๋ถ์ธ ๊ฒ์ ๋๋ค.
โWhy Every AI Team Needs Pirates and Architectsโ โ Every ๊ธฐ๋ฐ ๊ธฐ์ฌ ์๋ณธ ์์ ๋ณด๊ธฐ
AI ์๋, ๋น์ ์ ์ํํธ์จ์ด ํ์ โํด์ โ๊ณผ โ๊ฑด์ถ๊ฐโ๋ฅผ ํ์๋ก ํ๋ค
2026๋ , AI ์ฝ๋ฉ ์์ด์ ํธ์ ๋ฐ์ ์ ์ํํธ์จ์ด ๊ฐ๋ฐ์ ํ๊ฒฝ์ ์์ ํ ๋ฐ๊ฟ๋์์ต๋๋ค. ์ด์ ๋๊ตฌ๋ ์์ด๋์ด๋ง ์๋ค๋ฉด, โ๋ฐ์ด๋ธ ์ฝ๋ฉ(vibe coding)โ์ด๋ผ๋ ๋ฐฉ์์ผ๋ก ์ํ๋ ์ฑ์ ๋๋ฑ ๋ง๋ค์ด๋ผ ์ ์๋ ์๋๊ฐ ๋๋ํ์ต๋๋ค. ํ์ง๋ง ์ด๋ ๊ฒ ๋น ๋ฅด๊ฒ ๋ง๋ค์ด์ง ์ฑ์ด ์ค์ ํ๋ก๋์ ํ๊ฒฝ์์ ์ฑ๊ณต์ ์ผ๋ก ์๋ํ๊ณ ์ ์ง๋ ์ ์์๊น์? ์๋ธ๋ฆฌ(Every)์ ๊ณต๋ ์ฐฝ๋ฆฝ์์ด์ CEO์ธ ์ฐ์ฌ(์ ์)์ ๊ฒฝํ์ ์ด ์ง๋ฌธ์ ๋ํ ์ค์ํ ํต์ฐฐ์ ์ ๊ณตํ๋ฉฐ, ๋ฏธ๋์ ์ํํธ์จ์ด ํ์ด โํด์ (Pirate)โ๊ณผ โ๊ฑด์ถ๊ฐ(Architect)โ๋ผ๋ ๋ ๊ฐ์ง ํต์ฌ ์ญํ ๋ก ์ฌํธ๋์ด์ผ ํจ์ ์ญ์คํฉ๋๋ค.
AI ์ฝ๋ฉ, ๋๊ตฌ๋ โ๋ฐ์ด๋ธ ์ฝ๋ฉโํ ์ ์๋ ์๋
์ฐ์ฌ๋ ๊ฐ์ธ์ ์ธ ํ์์์ ์ถ๋ฐํ์ฌ โํ๋ฃจํ(Proof)โ๋ผ๋ ์ฑ์ ๊ฐ๋ฐํ์ต๋๋ค. ์ด ์ฑ์ ์์ด์ ํธ ์ ์ฉ ๋ฌธ์ ํธ์ง๊ธฐ๋ก, ์ฝ๊ฒ ๋งํด โ์์ด์ ํธ๊ฐ ์ฌ์ฉํ๋๋ก ๋ง๋ค์ด์ง ๊ตฌ๊ธ ๋ ์คโ๋ผ๊ณ ํ ์ ์์ต๋๋ค. ๋งํฌ๋ค์ด(Markdown) ๋ฌธ์๋ฅผ ์์ด์ ํธ์ ํจ๊ป ๊ณต๋ ์์ ํ๋ ๋ฌธ์ ๋ฅผ ํด๊ฒฐํ๊ธฐ ์ํด, ๊ทธ๋ ์ฝ๋ฑ์ค(Codex)์ ๊ฐ์ AI ์ฝ๋ฉ ์์ด์ ํธ๋ง์ ํ์ฉํ์ต๋๋ค. ๋๋๊ฒ๋ ๋จ ํ ์ค์ ์ฝ๋๋ ์ง์ ๋ณด์ง ์์ ์ฑ, ๊ธฐ๋ณธ์ ์ธ ์์ด๋์ด๋ง์ผ๋ก 10์ผ ๋ง์ ์์ฑ๋ ์ฑ์ ๋ง๋ค์ด๋์ต๋๋ค. ์ด๊ฒ์ด ๋ฐ๋ก โ๋ฐ์ด๋ธ ์ฝ๋ฉโ์ ํ์ ๋๋ค.
ํ๋ฃจํ๋ ์ถ์ ์ฒซ๋ ๋ถํฐ ์์ฒญ๋ ๋ฐ์์ ์ป์์ต๋๋ค. ํธ์ ํ๋๊ฐ 1,500๊ฐ ์ด์์ โ์ข์์โ์ 50๋ง ํ์ ์กฐํ์๋ฅผ ๊ธฐ๋กํ์ผ๋ฉฐ, ์ฒ์ 24~48์๊ฐ ๋์ 4,000๊ฐ์ ๋ฌํ๋ ์๋ก์ด ํ๋ฃจํ ๋ฌธ์๊ฐ ์์ฑ๋์์ต๋๋ค. ์ฐ์ฌ๋ โ๋ฌ์ ๊ฐ ๊ฒ ๊ฐ๋ค!โ๋ฉฐ ํฅ๋ถ์ ๊ฐ์ถ์ง ๋ชปํ์ต๋๋ค.
์ฑ๊ณต ๋ค์ ์ฐพ์์จ ํผ๋: โ๋ถํ๋ ์ฐ๋ ๊ธฐโ๋ฅผ ๊ณ ์น๋ค
๊ทธ๋ฌ๋ ์ด๊ธฐ ์ฑ๊ณต์ ํํฌ๋ ์ค๋๊ฐ์ง ๋ชปํ์ต๋๋ค. โ๋ฌผ๋ก ์ด ๋น๋์ค๋ฅผ ๋ง๋ค๊ณ ์๋ค๋ ๊ฒ์ ์ฐ๋ฆฌ๊ฐ ๋ฌ์ ๊ฐ์ง ๋ชปํ๋ค๋ ๋ปโ์ด๋ผ๋ ๊ทธ์ ๋ง์ฒ๋ผ, ์ฑ์ ๊ณง์ด์ด โ๋ถํ๋ ์ฐ๋ ๊ธฐโ์ ๊ฐ์ ์ํ๊ฐ ๋์์ต๋๋ค. ์ฐ์ฌ๋ ๋งค์ผ ๋ฐค ์๋ฒฝ 4์๊น์ง ์ ๋ชป ์ด๋ฃจ๋ฉฐ ์ฑ์ด ์ ๋๋ก ์๋ํ๋๋ก ๊ณ ๊ตฐ๋ถํฌํด์ผ ํ์ต๋๋ค. ์ฌ์ง์ด โํ๋ ๋ฆฌ์ ๊ฐ์ฒ์์ฒ๋ผ ๋ฐค์ ๋ถ์ด ๊บผ์ง์ง ์๋๋ก ๋ถ์ ์งํค๋ฏโ, ๋ ธํธ๋ถ์ ์์ ๋๊ณ ์๋ฉด์ 3~4์๊ฐ๋ง๋ค ์ฑ์ ํ์ธํ๋ ์ง๊ฒฝ์ ์ด๋ฅด๋ ์ต๋๋ค.
์ด๋ฌํ ๊ณ ํต์ค๋ฌ์ด ๊ฒฝํ์ ํตํด ๊ทธ๋ AI ์์ด์ ํธ๋ก ๋น ๋ฅด๊ฒ ๋ง๋ ์ฑ์ ์ค์ ํ๋ก๋์ ์์ค์ผ๋ก ๋์ด์ฌ๋ฆฌ๋ ๊ณผ์ ์์ ์๋ง์ ๊ตํ์ ์ป์์ต๋๋ค. ์ด๊ธฐ์๋ โ์์ฑํ ๋ฐ์ด๋ธ ์ฝ๋ฉ๋ ๋ฉ์ด๋ฆฌโ์๋ ์ฑ์ด ์ด์ ๋ ๋งค์ผ ์ฌ์ฉ๋๋ ๊ฒฌ๊ณ ํ ํ๋ก๋์ ์ฑ์ผ๋ก ์ฑ์ฅํ๊ฒ ๋ ๋น๊ฒฐ์ ๊นจ๋ฌ์ ๊ฒ์ ๋๋ค. ์ด ๊ฒฝํ์ ๋ฐ์ด๋ธ ์ฝ๋ฉ์ผ๋ก ์ฑ์ ๋ง๋ค๊ณ ํ๋ก๋์ ๋ฐฐํฌ๋ฅผ ๊ณ ๋ฏผํ๋ ์ด๋ค, ์๋ก์ด ์๋์ ์์ง๋์ด๋ง์ ๊ณ ๋ฏผํ๋ ์คํํธ์ CEO, ๊ทธ๋ฆฌ๊ณ ์์ ์ ๊ธฐ์ ๊ฐ์น์ ์๋ฌธ์ ํ๋ ์ํํธ์จ์ด ์์ง๋์ด ๋ชจ๋์๊ฒ ์ค์ํ ์์ฌ์ ์ ๋์ง๋๋ค.
2026๋ , ์ํํธ์จ์ด ์์ง๋์ด๋ง์ ์๋ก์ด ๊ตฌ์กฐ: โํด์ โ๊ณผ โ๊ฑด์ถ๊ฐโ
์ฐ์ฌ๊ฐ ๋ด๋ฆฐ ๊ฐ์ฅ ํฐ ๊ฒฐ๋ก ์ 2026๋ ์ ์ํํธ์จ์ด ์์ง๋์ด๋ง ํ์ ๋ ๊ฐ์ง ํต์ฌ ์ญํ , ์ฆ **ํด์ (Pirate)**๊ณผ **๊ฑด์ถ๊ฐ(Architect)**๋ก ๊ตฌ์ฑ๋์ด์ผ ํ๋ค๋ ๊ฒ์ ๋๋ค.
-
ํด์ (Pirate): ์๋์ ํํ์ ์ ๊ตฌ์
- ์ญํ : ์ ํ์ ํ์ ํ๊ณ , ๊ฐ๋ฅํ ํ ๋น ๋ฅด๊ฒ ์์ง์ฌ ๋ฏธ์ง์ ์์ญ์ ํํํ๋ฉฐ, ์ฌ๋๋ค์ด ์ง์ ์ผ๋ก ํ์๋ก ํ๋ ๊ฒ์ ๋ง๋ค์ด๋ด๋ ์ญํ ์ ๋๋ค.
- ํน์ง: โ๋ฐ์ด๋ธ ์ฝ๋ฉโ์ ๋ฌ์ธ์ผ๋ก, ์ํคํ ์ฒ๋ ์ ์ ๋น๋ ์์คํ ๋ณด๋ค๋ ๊ฐ์น ์๋ ๊ฒ์ ์ต๋ํ ๋นจ๋ฆฌ ์ฐพ์๋ด๋ ๋ฐ ์ง์คํฉ๋๋ค. ์ ํ์ ๋น์ ์ ์์ ํ๋ฉฐ, ์๋์ ์คํ์ ์ต์ฐ์ ์ผ๋ก ํฉ๋๋ค. ์ฐ์ฌ ์์ ๋ โํด์ โ์ ๊ฐ๊น๋ค๊ณ ๋งํฉ๋๋ค.
-
๊ฑด์ถ๊ฐ(Architect): ๊ฒฌ๊ณ ํจ๊ณผ ์ง์ ๊ฐ๋ฅ์ฑ์ ์ค๊ณ์
- ์ญํ : ํด์ ์ด ๋ฐ๊ฒฌํ ๊ฐ์น ์๋ ์์ด๋์ด๋ฅผ โ๋ ์ผ ์์ ์ฌ๋ ค๋๊ณ โ, ์ดํด ๊ฐ๋ฅํ๋ฉฐ ํ์ฅ ๊ฐ๋ฅํ๊ณ ์ ์ง๋ณด์ํ๊ธฐ ์ฌ์ด โ๊ธฐ๊ณโ๋ก ๋ง๋๋ ์ญํ ์ ๋๋ค. ์๋ฌด ์ด์ ์์ด ๊ฐ์๊ธฐ ๋ฉ์ถ๊ฑฐ๋ ์์ธ์ ์ ์ ์๋ ๋ฌธ์ ๋ฅผ ์ผ์ผํค์ง ์๋๋ก ์์คํ ์ ๊ฒฌ๊ณ ํจ์ ์ฑ ์์ง๋๋ค.
- ํน์ง: ์์คํ ์ ์์ ์ฑ๊ณผ ๊ตฌ์กฐ์ ๋ฌด๊ฒฐ์ฑ์ ์ด์ ์ ๋ง์ถฅ๋๋ค. ํด์ ์ด ๋ง๋ ํ๋กํ ํ์ ์ด ๋ฌด๋์ง์ง ์๋๋ก ๋ฐฉ์งํ๋ฉฐ, ์ฅ๊ธฐ์ ์ธ ๊ด์ ์์ ์์คํ ์ ์๋ช ๋ ฅ์ ๋ถ์ด๋ฃ์ต๋๋ค.
์ด ๋ ์ญํ ์ ์ํธ ๋ณด์์ ์ ๋๋ค. ํด์ ์ ๊ฑด์ถ๊ฐ ์์ด๋ ์ฝ๊ฒ ๋ฌด๋์ง ๊ฒฐ๊ณผ๋ฌผ์ ๋ง๋ค ์ ์๊ณ , ๊ฑด์ถ๊ฐ๋ ํด์ ์ด ์ฐพ์๋ธ ์์ฌ๋ฃ ์์ด๋ ๋ง๋ค ๊ฐ์น ์๋ ์์คํ ์ ๊ฐ์ง ์ ์์ต๋๋ค. ๋์ ํจ๊ป ์ผํ ๋ ๋น๋ก์ ์ง์ ํ ์๋์ง๋ฅผ ๋ฐํํ๋ฉฐ, ๋น ๋ฅด๊ณ ํ์ ์ ์ด๋ฉด์๋ ์์ ์ ์ธ ์ํํธ์จ์ด๋ฅผ ๋ง๋ค์ด๋ผ ์ ์์ต๋๋ค.
โํด์ โ์ ์ํ ์ธ ๊ฐ์ง ํต์ฌ ์ ๋ต
์ฐ์ฌ๋ ์ค์ค๋ก๋ฅผ โํด์ โ์ด๋ผ๊ณ ์๊ฐํ๋ฉฐ, ๊ทธ์ ๊ฒฝํ์ ๋ฐํ์ผ๋ก ํด์ ๋ค์ด ๋ ๋์ ์ฑ๊ณผ๋ฅผ ๋ด๊ธฐ ์ํ ์ธ ๊ฐ์ง ์กฐ์ธ์ ๊ณต์ ํ์ต๋๋ค.
- ๋จ์ํจ์ ์ง์คํ๋ผ: โ๋ฐ์ด๋ธ ์ฝ๋ฉโ์ ์ ํน์ ๋์์์ด ๊ธฐ๋ฅ์ ์ถ๊ฐํ๋ ๊ฒ์ ๋๋ค. ํ์ง๋ง ์ด๋ ๊ฒฐ๊ตญ ๊ฐ์น ์๋ ๊ฒฐ๊ณผ๋ฌผ๋ก ์ด์ด์ง๊ธฐ ์ฝ์ต๋๋ค. ์ฑ๊ณต์ ์ธ ํด์ ์ ์ ํ์ ํต์ฌ ๊ฐ์น์ ์ง์คํ๊ณ , ๊ฐ์ฅ ๋จ์ํ๋ฉด์๋ ์ ์๋ํ๋ ๊ฒ์ ๋ง๋๋ ๋ฐ ์ญ๋์ ์์์ผ ํฉ๋๋ค.
- ๊ณผ๊ฐํ ๋ฒ๋ฆฌ๊ณ ๋ค์ ์์ํ๋ผ (โํ์ ์ ์ง์๋ผโ): ์ผ๋จ ์๋ํ๋ ๊ฒ์ ์ฐพ์๋ค๋ฉด, ๊ธฐ์กด์ ์ฝ๋๋ฅผ ๊ณผ๊ฐํ ๋ฒ๋ฆฌ๊ณ ์ฒ์๋ถํฐ ๋ค์ ์์ํด์ผ ํฉ๋๋ค. ์๊ฐ ์ ๋ ๋๋ผ๋(Annie Dillard)์ โํ์ ์ ์ง์ฐ๋ผ(covering your tracks)โ๋ ๋ง์ฒ๋ผ, ๊ฐ๋ฐ ๊ณผ์ ์ ํผ๋์ค๋ฌ์ด ํ์ (์๋ํ๋ ๋ค์ํ ๊ธฐ๋ฅ, ๊ฐ๋
ํ ๋ฑ)์ ์ฌ์ฉ์์๊ฒ ๋ณด์ฌ์ค ํ์๋ ์์ต๋๋ค. ์ฝ๋๋ ์์ฑํ๊ธฐ ์ ๋ ดํ๋ฏ๋ก, ๋ฒ๋ฆฌ๋ ๊ฒ๋ ์ ๋ ดํฉ๋๋ค.
- AI ์์ด์ ํธ์ ํ๊ณ: โ์์ด์ ํธ์๊ฒ ๊ธฐ์กด์ โ๋ฐ์ด๋ธ ์ฝ๋ฉ๋ ์๋ง์ง์ฐฝโ์ ๋ฆฌํฉํฐ๋ง(refactor)ํ๋ผ๊ณ ์ํค๋ ๊ฒ์ ํจ๊ณผ๊ฐ ์์ต๋๋ค.โ ์ฝ๋ฉ ์์ด์ ํธ๋ ์ด๋ฏธ ์กด์ฌํ๋ ์ฝ๋์ ์ฝ๊ฒ โ์ฃผ์๊ฐ ์ฐ๋งํด์ง๋โ ๊ฒฝํฅ์ด ์๊ธฐ ๋๋ฌธ์, ์๋ก์ด ์ฝ๋๋ฒ ์ด์ค์์ ์์ํ๋ ๊ฒ์ด ํจ์ฌ ํจ์จ์ ์ ๋๋ค.
- ์ฌ์์ ์์ : ์ ํ์ ๊ฐ์น๊ฐ ๋ช ํํ๊ณ , ์๋ฌด๋ฆฌ ๋ ธ๋ ฅํด๋ ํด๊ฒฐ๋์ง ์๋ ๋ฒ๊ทธ๊ฐ ์์ฌ๊ฐ ๋๊ฐ ๋ฐ๋ก ์๋ก์ด ์ฝ๋๋ฒ ์ด์ค๋ฅผ ์์ํ ์ ๊ธฐ์ ๋๋ค. ์ฐ์ฌ๋ ์ฝ๋ฑ์ค๋ฅผ โ์ฌ๋กฏ๋จธ์ โ ๊ฐ์๋ค๊ณ ๋น์ ํ์ต๋๋ค. ๋งค ํ๋กฌํํธ๋ง๋ค โ์ด๋ฒ์ ๋ ๊ฑฐ์ผ!โ๋ผ๊ณ ๊ธฐ๋ํ์ง๋ง, ํ์ค์ ์ฝ์ง ์์์ต๋๋ค. ์์ด์ ํธ๊ฐ ์๋ํ๋ ๊ฒ์ ๋ณด๋ ๊ฒ์ ์ค๋ ์ฑ์ด ๊ฐํ์ง๋ง, ์ฐ๋ ๊ธฐ๋ฅผ ๋ง๋๋๋ผ ์๊ฐ์ ๋ญ๋นํ์ง ์๋๋ก ์ฌ๋ฐ๋ฅธ ๊ด๊ณ๋ฅผ ์ค์ ํ๋ ๊ฒ์ด ์ค์ํฉ๋๋ค.
- ์์ด์ ํธ ์ค์ฌ์ ์์ฐ์ฑ ์ฑ์ ์ฌ์ฐฝ์กฐํ๋ผ: ์ฐ์ฌ๋ โํ๋ฃจํโ๊ฐ ์์ด์ ํธ๋ฅผ ์ํ ๊ตฌ๊ธ ๋ ์ค์ธ ๊ฒ์ฒ๋ผ, ๋ชจ๋ ๊ธฐ์กด ์์ฐ์ฑ ์ฑ(๊ตฌ๊ธ ์ํธ, ํ์ํฌ์ธํธ ๋ฑ)์ ์์ด์ ํธ์ ๊ด์ ์์ ์ฌ๊ตฌ์ํ ์์ฒญ๋ ๊ธฐํ๊ฐ ์๋ค๊ณ ๊ฐ์กฐํฉ๋๋ค. ๋ค์ ์ธ๋ ์ํํธ์จ์ด์ ์ฃผ์ ์ฌ์ฉ์๋ ์์ด์ ํธ๊ฐ ๋ ๊ฒ์ด๋ฉฐ, ์์ด์ ํธ๋ฅผ ์ฃผ ์ฌ์ฉ์๋ก ์์ ํ๊ณ ์ฑ์ ๋ง๋ค๋ฉด ์์ ํ ์๋กญ๊ณ ๊ฐ์น ์๋ ์ํํธ์จ์ด๋ฅผ ํ์์ํฌ ์ ์์ต๋๋ค.
โ๊ฑด์ถ๊ฐโ์ ๊ฐ์น๋ ๋์ฑ ์ปค์ง๋ค
๊ทธ๋ ๋ค๋ฉด โ๊ฑด์ถ๊ฐโ์ ์ญํ ์ AI ์๋์ ์ด๋ป๊ฒ ๋ณํ ๊น์? ์ํํธ์จ์ด ์์ง๋์ด๋ง์ด๋ผ๋ ์ง์ ์์ฒด๊ฐ ์ฌ๋ผ์ง ๊ฒ์ด๋ผ๋ ์ฐ๋ ค๋ ์์ง๋ง, ์ฐ์ฌ์ ๊ฒฝํ์ ์ ๋ฐ๋์ ๊ฒฐ๋ก ์ ๋ด๋ ธ์ต๋๋ค.
ํ๋ฃจํ์ ๋ฌธ์ ๋ฅผ ํด๊ฒฐํ๋ ๊ณผ์ ์์, ๊ทธ๋ ํ ๋ช ์ โ๊ฑด์ถ๊ฐโ๋ฅผ ํฌ์ ํ์ฌ ์ฑ์ ์ผ๋ถ ์น์ ์ ์ฒ์๋ถํฐ ๋ค์ ์์ฑํ๋๋ก ํ์ต๋๋ค. ๊ทธ๋ฆฌ๊ณ ์ด๊ฒ์ด ๋ฐ๋ก ๋ฌธ์ ๋ฅผ ํด๊ฒฐํ ๊ฒฐ์ ์ ์ธ ๊ณ๊ธฐ์์ต๋๋ค. ์ด ๊ฒฝํ์ ํตํด ๊ทธ๋ AI๋ฅผ ํ์ฉํ๋ ๊ฑด์ถ๊ฐ์ ํ์ด โ์์ฒญ๋๊ฒ ๊ฐ๋ ฅํ๊ณ ์ค์ํ๋ฉฐ, ๊ฐ๊น์ด ๋ฏธ๋์ ๋ชจ๋ธ๋ก ๋์ฒด๋ ์ ์๋คโ๋ ๊ฒ์ ๋ถ๋ช ํ ๊นจ๋ฌ์์ต๋๋ค.
- AI ์ฝ๋ฉ ๋ชจ๋ธ์ ํ๊ณ: ์ฝ๋ฉ ๋ชจ๋ธ์ ๋ถ๋ช ๊ฐ๋ ฅํ๊ณ ๋ชจ๋ ๊ฒ์ ๋ณํ์ํค๊ณ ์์ง๋ง, โ์ ๋ง๋ก ์๋์ด ์์ง๋์ด๊ฐ ํ๋ก์ ํธ์ ๊ฐ์ ธ์ฌ ์ ์๋ โ๊ฐ๋ ์ ๋ช ํ์ฑ(conceptual clarity)โ์ด ๋ถ์กฑํฉ๋๋ค.โ AI ๋ชจ๋ธ์ ๊ฐ๋ณ์ ์ธ ๋ก์ปฌ ๋ณ๊ฒฝ ์ฌํญ์ ์ ์ฒ๋ฆฌํ์ง๋ง, ์ ์ฒด์ ์ธ ๊ทธ๋ฆผ์ ๋ณด๋ฉด ์์คํ ์ด ์์ง๋ ฅ ์๊ฒ ์ฐ๊ฒฐ๋์ง ์๋ ๊ฒฝ์ฐ๊ฐ ๋ง์ต๋๋ค. ๋ฐ๋ฉด, ๋ ธ๋ จํ ์์ง๋์ด๋ ์ ์ฒด ์์คํ ์ด ์ด๋ป๊ฒ ์ ๊ธฐ์ ์ผ๋ก ์๋ํ๋์ง ํ์ ํ๋ ๋ฐ ๋งค์ฐ ๋ฅ์ํฉ๋๋ค.
- ์๋์ด ์์ง๋์ด์ ์ญํ ํ์ฅ: ํ์ฌ๋ก์๋ ํนํ ์๋์ด ์์ง๋์ด๊ฐ AI ๋๊ตฌ๋ฅผ ํ์ฉํ์ฌ ์์ ์ ์ญ๋์ ํ์ฅํ๋ค๋ฉด, AI ๋ชจ๋ธ์ด ๊ฐ์ง์ง ๋ชปํ๋ ์๋ก์ด ์ ๋ฌธ์ฑ์ ์ฐฝ์ถํ ์ ์์ต๋๋ค.
๊ฒฐ๋ก ๋ฐ ์์ฌ์
์ํํธ์จ์ด ์์ง๋์ด๋ง์ ๊ทน์ ์ผ๋ก ๋ณํํ๊ณ ์์ผ๋ฉฐ, ํ ๊ตฌ์กฐ์ ๊ฐ๋ฐ ๋ฐฉ์ ๋ชจ๋๊ฐ ๋ฐ๋๊ณ ์์ต๋๋ค. ์ด ์๋ก์ด ์๋์ ํ์ํ ํต์ฌ ์ญํ ์ ๋จ ๋ ๊ฐ์ง์ ๋๋ค.
- ํด์ : ๊ฐ์น ์๋ ๊ฒ์ ์ฐพ์๋ด๊ธฐ ์ํด ๊ฐ๋ฅํ ํ ๋น ๋ฅด๊ฒ ์์ง์ด๋ ์ฌ๋.
- ๊ฑด์ถ๊ฐ: ํด์ ์ด ๋ฐ๊ฒฌํ ๊ฐ์น ์์ง๋ง ๋ฌด์ง์ํ ๊ฒ์ ์ ๊ตฌ์กฐํ๋๊ณ ์กฐ์งํ๋ โ๊ธฐ๊ณโ๋ก ๋ง๋๋ ๊ฒ์ ๋๋ ์ฌ๋.
์ฐ์ฌ๋ ๋ค์๋ฒ์ โ๋ฐ์ด๋ธ ์ฝ๋ฉโ์ผ๋ก ์ฌ์ด๋ ํ๋ก์ ํธ๋ฅผ ์์ํ ๋๋ ๋ฐ๋์ โ๊ฑด์ถ๊ฐโ๋ฅผ ๊ณ ์ฉํ ๊ฒ์ด๋ผ๊ณ ๋จ์ธํฉ๋๋ค. โ๊ทธ๋ ์ง ์์ผ๋ฉด ๋ด ํ ์ ์ฒด๊ฐ ๋ฐ๋์ ์ผ์ผํฌ ๊ฒ์ด๊ธฐ ๋๋ฌธโ์ด๋ผ๋ ์ ๋จธ ์์ธ ๊ณ ๋ฐฑ์ ํ์ ์ ์ค์์ฑ์ ๊ฐ์กฐํฉ๋๋ค. ๋ชจ๋ ๊ฒ์ ํผ์ ํ ํ์๋ ์์ผ๋ฉฐ, ํจ๊ป ์ผํ๋ ๊ฒ์ด ํจ์ฌ ๋ซ๋ค๋ ๊ตํ์ ๋๋ค.
์ด ๋น๋์ค๋ฅผ ํตํด ์๊ฐ์ ๋ฐ์๋ค๋ฉด, ์ง๊ธ ๋น์ฅ ์ฝ๋ฑ์ค(Codex)๋ ํด๋ก๋ ์ฝ๋(Claude Code), ๋๋ ๋น์ ์ด ์ข์ํ๋ ์ฝ๋ฉ ๋ชจ๋ธ์ ์คํํ๊ณ , ์ฐ์ฌ์ ํจ๊ป โ๋ฌโ๋ก ํฅํ๋ ์ฌ์ ์ ๋์ฐธํด ๋ณด์ธ์!
๋ ์์๋ณด๊ธฐ:
- ํ๋ฃจํ(Proof) ์ฑ: proofeditor.ai
- ์๋ธ๋ฆฌ(Every) ๊ตฌ๋ : every.to
โBen Horowitz on AI Anxiety, Big Tech Transitions & The Future of Startups | a16zโ โ a16z ๊ธฐ๋ฐ ๊ธฐ์ฌ ์๋ณธ ์์ ๋ณด๊ธฐ
AI ๊ฒฉ๋ณ๊ธฐ, ๊ธฐ์ ๊ณผ VC์ ์์กด ์ ๋ต: ๋ฒค ํธ๋ก์์ธ ์ ๋์ฒ ํ ์ง๋จ๊ณผ ๋๊ด์ ์ ๋ง
[๋์ ๋ถ] ์ธ๊ณต์ง๋ฅ(AI)์ ๊ธ๋ถ์์ ์ ์ธ๊ณ ์ฐ์ ์งํ์ ๋ฟ๋ฆฌ์งธ ํ๋ค๊ณ ์์ผ๋ฉฐ, ์ด๋ ๋จ์ํ ๊ธฐ์ ๋ณํ๋ฅผ ๋์ด ๊ฒฝ์ ์์คํ ๊ณผ ์ฌํ ๊ตฌ์กฐ ์ ๋ฐ์ ๊ฑธ์น ๋๋ณํ์ ์๊ณ ํฉ๋๋ค. ์ด๋ฌํ ๊ฒฉ๋์ ์๊ธฐ์ ์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ์ ์ ์ค์ ์ธ ๋ฒค์ฒ์บํผํ๋ฆฌ์คํธ์ด์ ์ค๋๋ฆฌ์จ ํธ๋ก์์ธ (Andreessen Horowitz, a16z)์ ๊ณต๋ ์ค๋ฆฝ์์ธ ๋ฒค ํธ๋ก์์ธ (Ben Horowitz)๋ AI ์๋์ ๋ถ์๊ฐ, ๊ฑฐ๋ ๊ธฐ์ ๊ธฐ์ ์ ์ ํ, ๊ทธ๋ฆฌ๊ณ ์คํํธ์ ์ ๋ฏธ๋์ ๋ํ ๊น์ด ์๋ ํต์ฐฐ์ ์ ์ํ์ต๋๋ค. ๊ทธ๋ ์ธํ๋ผ์ ์ฌ๊ฑด๋ถํฐ ๊ธฐ์ ์์กด ์ ๋ต, ์ํธํํ(Crypto)์ ์ญํ , ๊ทธ๋ฆฌ๊ณ ๋ฒค์ฒ ์บํผํ(Venture Capital, VC)์ ๋ฏธ๋์ ์ด๋ฅด๊ธฐ๊น์ง ๊ด๋ฒ์ํ ์ฃผ์ ๋ฅผ ์์ฐ๋ฅด๋ฉฐ, ํผ๋ ์์์๋ ๋๊ด์ ์ธ ๋ฏธ๋๋ฅผ ๊ทธ๋ ธ์ต๋๋ค.
AI ์๋, ๊ธฐ์ ์์กด์ ์๋ก์ด ๋ฒ์น
๋ฒค ํธ๋ก์์ธ ๋ AI๊ฐ ๊ธฐ์กด์ ๋น์ฆ๋์ค ๋ชจ๋ธ๊ณผ ๊ธฐ์ ์ด์ ๋ฐฉ์์ โ๋ฌผ๋ฆฌํ ๋ฒ์นโ์ ๊ทผ๋ณธ์ ์ผ๋ก ๋ณํ์ํค๊ณ ์๋ค๊ณ ๊ฐ์กฐํฉ๋๋ค. ๊ทธ๊ฐ ์ ์ํ ๋ ๊ฐ์ง ํต์ฌ ๋ณํ๋ ๋ค์๊ณผ ๊ฐ์ต๋๋ค.
-
โ๋์ผ๋ก ๋ฌธ์ ํด๊ฒฐโ์ ๊ฐ๋ฅ์ฑ ์ฆ๋: ๊ณผ๊ฑฐ์๋ ๋จ์ํ ๋์ ์์๋ถ๋๋ค๊ณ ํด์ ๋ฌธ์ ๋ฅผ ํด๊ฒฐํ ์ ์๋ค๋ ๊ฒ์ด ์์์ฒ๋ผ ์ฌ๊ฒจ์ก์ต๋๋ค. ์๋ฅผ ๋ค์ด, ์ ํ ๊ฐ๋ฐ์ด 2๋ ๋ค์ฒ์ก๋ค๊ณ ํด์ ์์ง๋์ด 1์ฒ ๋ช ์ ๊ณ ์ฉํ๋ค๊ณ ๊ฒฝ์์ฌ๋ฅผ ๋ฐ๋ผ์ก์ ์๋ ์์์ต๋๋ค. ์ด๋ โ์ ์ค์ ์ธ ๋งจ๋จผ์ค(Mythical Man-Month)โ๋ก ๋ํ๋๋ ์ํํธ์จ์ด ๊ฐ๋ฐ์ ๊ณ ์ ํ ํน์ฑ์ด์์ต๋๋ค. ๊ทธ๋ฌ๋ AI ์๋์๋ ์ถฉ๋ถํ ์๊ธ๊ณผ ์์ง์ ๋ฐ์ดํฐ, ๊ทธ๋ฆฌ๊ณ GPU(๊ทธ๋ํฝ ์ฒ๋ฆฌ ์ฅ์น)๋ง ์๋ค๋ฉด ์ํํธ์จ์ด ๋ถ์ผ์ ๊ฑฐ์ ๋ชจ๋ ๋ฌธ์ ๋ฅผ ํด๊ฒฐํ ์ ์๊ฒ ๋์์ต๋๋ค. ์ฆ, ๋์ด ๊ณง ํด๊ฒฐ์ฑ ์ด ๋ ์ ์๋ ์์ญ์ด ํ์ฅ๋ ๊ฒ์ ๋๋ค.
-
๊ณ ๊ฐ ๋ก์ธ(Lock-in) ํจ๊ณผ์ ์ฝํ: ์ํํธ์จ์ด ์ฐ์ ์์๋ ๊ณ ๊ฐ์ ํ๋ณดํ๋ ๊ฒ์ด ๊ณง ๊ฐ๋ ฅํ ๊ฒฝ์ ์ฐ์๋ก ์์ฉํ์ต๋๋ค. ๋ฐ์ดํฐ ๋ง์ด๊ทธ๋ ์ด์ ์ ์ด๋ ค์, ๋ฐ์ดํฐ ๋ก์ธ, ์ฌ์ฉ์ ์ธํฐํ์ด์ค(UI) ๋ก์ธ ๋ฑ ๋ค์ํ ํํ๋ก ๊ณ ๊ฐ์ ๋ฌถ์ด๋ ์ ์์๊ธฐ ๋๋ฌธ์ ๋๋ค. ํ์ง๋ง AI ์๋์๋ ์ฝ๋ ๋ณต์ ๊ฐ ๋งค์ฐ ์ฌ์์ก๊ณ , ๋ฐ์ดํฐ ์ด๋๋ ๊ฐํธํด์ก์ต๋๋ค. ๋ ๋์๊ฐ, ์ํํธ์จ์ด์ ์ํธ์์ฉํ๋ ์ฃผ์ฒด๊ฐ ์ธ๊ฐ์ด ์๋ AI๊ฐ ๋ ๊ฒ์ด๋ฏ๋ก, AI๊ฐ ์ฌ์ฉ์ ์ธํฐํ์ด์ค๋ฅผ ์ ์ฐํ๊ฒ ํ์ฉํ๊ฒ ๋๋ฉด ๊ธฐ์กด์ ๋ก์ธ ์ ๋ต์ ๋ฌด๋ ฅํ๋ ์ ์์ต๋๋ค.
์ด๋ฌํ ๋ณํ๋ ๊ธฐ์ ์ ์์กด ์ฃผ๊ธฐ๋ฅผ ๊ทน์ ์ผ๋ก ๋จ์ถ์ํค๊ณ ์์ต๋๋ค. ๊ณผ๊ฑฐ์๋ ์ข์ ์ ํ ํ๋๋ก 5~10๋ ์ ์ฑ์ฅ๊ธฐ๋ฅผ ๋๋ฆด ์ ์์์ง๋ง, ์ด์ ๋ 5์ฃผ ๋ง์ ์์ฅ์์ ๋ํ๋ ์๋ ์๋ ์ํฉ์ ๋๋ค. ํนํ, ์๋น์คํ ์ํํธ์จ์ด(SaaS) ๊ธฐ์ ๋ค์๊ฒ ๋ฅ์น โSaaS ์ข ๋ง๋ก (SAS Apocalypse)โ์ ๊ธฐ์ ์ ์ต์ข ๊ฐ์น(Terminal Value)์ ๋ํ ์๊ตฌ์ฌ์์ ๋น๋กฏ๋ ํ์์ ๋๋ค. ๊ธฐ์ ๋ค์ ๊ฒฝ์ ์ ๊ฐ์น ์ฐฝ์ถ์ ๋ชฉํ๋ก ํ์ง๋ง, ๋๋ฌด ์ค๋ ๊ธฐ๋ค๋ฆฌ๋ค๊ฐ๋ ๊ธฐ์ ๊ฐ์น๊ฐ 0์ด ๋ ์๋ ์๋ค๋ ๋ถ์๊ฐ์ด ์ปค์ง๊ณ ์์ต๋๋ค.
CEO์ ์ญํ : ํ์ค ์ง์์ ์ฐจ๋ณํ๋ ๊ฐ์น ์ ๊ณต ์ด๋ฌํ ๊ฒฉ๋ณ๊ธฐ ์์์ CEO๋ค์ ๋ฌด์์ ํด์ผ ํ ๊น์? ํธ๋ก์์ธ ๋ ์ฒซ์งธ, โ์์ ์ด ๊ฐ์ง ๊ฒ์ด ๋ฌด์์ธ์ง ์์งํ๊ฒ ์ง์ํด์ผ ํ๋คโ๊ณ ์กฐ์ธํฉ๋๋ค. ๊ทธ๋ฆฌ๊ณ โ์์ ํ ๋ค๋ฅธ ๋ฌผ๋ฆฌํ ๋ฒ์น์ด ์ ์ฉ๋๋ ์๋ก์ด ์ธ์โ์ ์ดํดํ๊ณ , โ๊น์ด ์๊ณ ์ฐจ๋ณํ๋ ๊ฐ์นโ๋ฅผ ์ ๊ณตํ๋ ๋ฐ ์ง์คํด์ผ ํ๋ค๊ณ ๊ฐ์กฐํฉ๋๋ค.
๊ทธ๋ ์ฌํ ํ๋ซํผ โNANAโ์ ์ฌ๋ก๋ฅผ ๋ค์ด ์ค๋ช ํฉ๋๋ค. SaaS ์ข ๋ง๋ก ์ ๋ฐ๋ฅด๋ฉด ์ฌํ ์ฐ์ ์ ํฐ ํ๊ฒฉ์ ์ ์ด์ผ ํ์ง๋ง, NANA๋ ์ฌ์ ํ ๊ฐ์ ์ ๊ฐ์ง๊ณ ์์ต๋๋ค. ๊ทธ ์ด์ ๋ ๋ค์๊ณผ ๊ฐ์ต๋๋ค.
- ๋ช ์์ ์ธ ๊ด๊ณ ๊ตฌ์ถ: ๊ธ๋ก๋ฒ ์ฌํ์ ์ํด์๋ ์ ์ธ๊ณ ํญ๊ณต์ฌ, ํธํ , ๊ธฐ์ฐจ ๋ฑ๊ณผ ๊ฐ๋ณ์ ์ธ ๊ด๊ณ๋ฅผ ๊ตฌ์ถํด์ผ ํฉ๋๋ค. ์ด๋ AI๊ฐ ์ฝ๊ฒ ๋์ฒดํ ์ ์๋ ๋ณต์กํ ์ธ๊ฐ ๊ด๊ณ์ ํ์ ์ด ํ์ํ ์์ญ์ ๋๋ค.
- ํน์ ์ฑ๋ ์ ๊ทผ: ์ฌํ ๊ด๋ฆฌ์(Travel Manager)์๊ฒ ์ ๊ทผํ๋ ์ฑ๋์ ์ผ๋ฐ์ ์ธ ๊ธฐ์ ๊ธฐ์ ์ด ์ฝ๊ฒ ํ๋ณดํ ์ ์๋ ๊ฒ์ด ์๋๋๋ค.
- ๋ณต์กํ ์์ด์ ํธํ ์ฌํ ๊ฒฝํ: AI ๊ธฐ๋ฐ์ ์ฌํ ๊ฒฝํ์ ์๊ฐ๋ณด๋ค ํจ์ฌ ๋ณต์กํ๋ฉฐ, ๋จ์ํ ์ฝ๋ ๋ณต์ ๋ก ๊ตฌํํ๊ธฐ ์ด๋ ต์ต๋๋ค.
๊ฒฐ๋ก ์ ์ผ๋ก, AI ์๋์ ๊ธฐ์ ๋ค์ ํ๋ฉด์ ์ธ ๊ธฐ๋ฅ(Feature)์ ๋์ด ๊ณ ๊ฐ์๊ฒ ์ด๋ค ๋ ์ ์ ์ด๊ณ ๊น์ด ์๋ ๊ฐ์น๋ฅผ ์ ๊ณตํ๋์ง์ ๋ฐ๋ผ ์์กด ์ฌ๋ถ๊ฐ ๊ฐ๋ฆด ๊ฒ์ ๋๋ค.
์ธํ๋ผ ์๊ธฐ์ ๋ฒค์ฒ ์บํผํ์ ์๋ก์ด ์ญํ
๋ฒค ํธ๋ก์์ธ ๋ ๋ฏธ๊ตญ์ด ์ง๊ธ ๋น์ฅ ์ ์ฒด ์ธํ๋ผ๋ฅผ ์ฌ๊ฑดํด์ผ ํ ์๊ธฐ์ ์ฒํด ์๋ค๊ณ ๊ฒฝ๊ณ ํฉ๋๋ค. ๊ทธ๋ ํฌํ ๋ฅ(rare earth minerals), ์ ๋ ฅ, ์ ์กฐ ์ญ๋, ๊ทธ๋ฆฌ๊ณ ์ ๋ ฅ ํจ์จ์ด ๋ฎ์ ์นฉ ๋ฑ ๋ชจ๋ ๋ถ์ผ์์ ์ฌ๊ฐํ ๋ถ์กฑ ํ์์ ๊ฒช๊ณ ์๋ค๊ณ ์ง์ ํฉ๋๋ค. ํนํ ์ ๋ ฅ ๋ถ์กฑ ๋ฌธ์ ๋ โ12๊ฐ์ ํ๊ฐ ์๋, ์ง๊ธ ๋น์ฅโ ์ฌ๊ฐํ ์์ค์ด๋ผ๊ณ ๊ฐ์กฐํฉ๋๋ค.
์ด๋ฌํ ์ธํ๋ผ ์๊ธฐ๋ AI ์๋์ ํญ๋ฐ์ ์ธ ์์์ ๋ง๋ฌผ๋ ค ๋์ฑ ๊ฐ์ํ๋๊ณ ์์ต๋๋ค. ์ค๊ตญ์ AI ๊ด๋ จ ์ธํ๋ผ ํฌ์๊ฐ ๊ฐํ๋ฅด๊ฒ ์์นํ๋ ๋ฐ๋ฉด, ๋ฏธ๊ตญ์ ์ธํ๋ผ ๊ตฌ์ถ ์๋๋ ํฑ์์ด ๋๋ฆฌ๋ค๋ ์ ๋ ์ฐ๋ ค๋ฅผ ๋ํฉ๋๋ค. AI ํ ํฐ(AI ๋ชจ๋ธ ํ์ต ๋ฐ ์ด์์ ํ์ํ ์์)์ ๋ํ ์์๋ ์์ง ์์นํ์ง๋ง, ์ด๋ฅผ ๋ท๋ฐ์นจํ ์ธํ๋ผ ์ฉ๋์ ๊ทธ๋ ์ง ๋ชปํ๋ค๋ ๊ฒ์ ๋๋ค.
๋ฒค์ฒ ์บํผํ์ ๋๊ท๋ชจ ํฌ์: ์ค๋๋ฆฌ์จ ํธ๋ก์์ธ ๊ฐ ์ต๊ทผ 7๊ฐ ํ๋ ์ค 4๊ฐ ํ๋์์ 150์ต ๋ฌ๋ฌ(์ฝ 20์กฐ ์)๋ผ๋ ๋๊ท๋ชจ ์๊ธ์ ์กฐ๋ฌํ ๋ฐฐ๊ฒฝ์๋ ์ด๋ฌํ ์ธํ๋ผ ์ฌ๊ฑด์ ํ์์ฑ์ด ๊น๋ ค ์์ต๋๋ค. ํธ๋ก์์ธ ๋ โ๋ฏธ๊ตญ์ ์ง๊ธ ๋น์ฅ ์ ์ฒด ์ธํ๋ผ๋ฅผ ์ฌ๊ฑดํด์ผ ํ๋คโ๋ฉฐ, ์ด๋ฅผ ์ํด์๋ ๋ง๋ํ ์๊ธ์ด ํ์ํ๋ค๊ณ ์ค๋ช ํฉ๋๋ค. ์ค์ ๋ก a16z๋ AI ํธ๋์คํฌ๋จธ(AI ๋ชจ๋ธ์ โํธ๋์คํฌ๋จธโ๊ฐ ์๋, ์ค์ ์ ๋ ฅ ๋ณ์๊ธฐ) ํ์ฌ์ ํฌ์ํ๋ ๋ฑ ๊ธฐ์ด ์ธํ๋ผ ๋ฌธ์ ํด๊ฒฐ์ ์ ๊ทน์ ์ผ๋ก ๋์๊ณ ์์ต๋๋ค.
๊ณต๊ธ๋ง ๋ณ๋ชฉ ํ์: โ๋์ ๊ฐ๊ฒฉ์ ์น๋ฃ์ ๋ ๋์ ๊ฐ๊ฒฉโ์ด๋ผ๋ ์๋ง์ฒ๋ผ ์์ฅ์ ์์์ ๊ณต๊ธ์ ๊ฒฐ๊ตญ ๊ท ํ์ ์ฐพ์ง๋ง, ๋ฌธ์ ๋ โ์ง์ฐ ์๊ฐ(Latency)โ์ ๋๋ค. ํ์ฌ ์ปดํจํฐ๋ RAM์ด ์์ด ์ถํ๋์ง ๋ชปํ๋ ๊ฒฝ์ฐ๊ฐ ๋ฐ์ํ๊ณ ์์ต๋๋ค. ์๋ก์ด DRAM ๊ณต์ฅ์ ์ง๋ ๋ฐ 5๋ ์ด ๊ฑธ๋ฆฌ๋ ๊ฒ์ฒ๋ผ, ์ธํ๋ผ ๊ตฌ์ถ์๋ ์๋นํ ์๊ฐ์ด ์์๋ฉ๋๋ค. 1999๋ ๋ท์ปด ๋ฒ๋ธ ๋น์์๋ ๊ด์ฌ์ ๋คํธ์ํฌ๊ฐ โ์ด๋์ด(dark)โ ์ํ์์ง๋ง, ์ง๊ธ์ ๋ชจ๋ GPU๊ฐ โ๋จ๊ฒ๊ฒ(hot)โ ๊ฐ๋๋๊ณ ์๋ค๋ ์ ์์ ์ํฉ์ ์ฌ๊ฐ์ฑ์ด ๋ค๋ฆ ๋๋ค. ํธ๋ก์์ธ ๋ โ๊ฑฐ์ ๋ชจ๋ ๊ฒ์ด ๋ณ๋ชฉ ํ์โ์ด๋ผ๋ฉฐ, ํนํ ์ ๋ ฅ ๋ถ์กฑ์ด ๊ฐ์ฅ ํฐ ๋ฌธ์ ์ด๋ฉฐ ์๋น๋์(NVIDIA)๊ฐ ์ถฉ๋ถํ ์นฉ์ ์์ฐํด๋ ๋ฉ๋ชจ๋ฆฌ์ ์ ๋ ฅ์ด ๋ถ์กฑํ ๊ฒ์ด๋ผ๊ณ ์ ๋งํฉ๋๋ค. ๊ทธ๋ ์ผ๋ก ๋จธ์คํฌ(Elon Musk)์ ํ ๋ผํฉํ ๋ฆฌ(Terrafab)์ฒ๋ผ ๋ชจ๋ ๋ณ๋ชฉ ํ์์ ์ง์ ํด๊ฒฐํ๋ ค๋ ์ ๊ทผ ๋ฐฉ์์ด ํ์ํ๋ค๊ณ ๋ง๋ถ์์ต๋๋ค.
AI๊ฐ ์ด๋ํ ๋ฌธ์ , ์ํธํํ(Crypto)๊ฐ ํด๋ต์ธ๊ฐ?
๋ฒค ํธ๋ก์์ธ ๋ AI๊ฐ ์ด๋ํ ๊ฐ์ฅ ์ฌ๊ฐํ ๋ฌธ์ ์ค ํ๋๋ก โ์ ๋ขฐ ์๊ธฐโ์ โ์ ์ ํ์ธ ๋ฌธ์ โ๋ฅผ ๊ผฝ์ต๋๋ค. AI๋ ๋ฅํ์ดํฌ(Deepfake) ๊ธฐ์ ์ ํตํด ๋ชฉ์๋ฆฌ์ ์์๊น์ง ์๋ฒฝํ๊ฒ ์์กฐํ ์ ์๊ฒ ๋๋ฉด์, ์ฌ๊ธฐ ๋ฐ ์คํธ์ด ๊ธฐํ๊ธ์์ ์ผ๋ก ์ฆ๊ฐํ ๊ฒ์ด๋ผ๋ ์ฐ๋ ค๋ฅผ ํํฉ๋๋ค. ๊ทธ๋ ์์ ์ด ํ๋ฐค์ค์ ๊นจ์ด๋ โAI๊ฐ ๋๋ฅผ ํ๋ด ๋ด์ด ์ฌ๋ฌดํ์ ๋์ด์ง๋ฆฌ์๋ก 5์ต ๋ฌ๋ฌ๋ฅผ ์ก๊ธํ๋ผ๊ณ ์ง์ํ๋ ์ํฉโ์ ์์ํ๋ค๊ณ ๋งํ๋ฉฐ, ์ด๋ฌํ ๋ฌธ์ ๊ฐ ๋น ๋ฅด๊ฒ ํ์คํ๋๊ณ ์๋ค๊ณ ๊ฒฝ๊ณ ํฉ๋๋ค.
์ด๋ฌํ ๋ฌธ์ ์ ๋ํ ํด๋ต์ผ๋ก ํธ๋ก์์ธ ๋ ์ํธํํ(Crypto)์ ๋ธ๋ก์ฒด์ธ(Blockchain) ๊ธฐ์ ์ ์ค์์ฑ์ ์ญ์คํฉ๋๋ค.
-
์ธ๊ฐ/๋ด ์ฆ๋ช (Human/Bot Proof): AI ์๋์๋ ๋ด๊ฐ ์ค์ ์ธ๊ฐ๊ณผ ๋ํํ๊ณ ์๋์ง, ์๋๋ฉด ๋ด๊ณผ ๋ํํ๊ณ ์๋์ง ์๋ ๊ฒ์ด ๋งค์ฐ ์ค์ํด์ง ๊ฒ์ ๋๋ค. ์์ ๋ฏธ๋์ด, ๋ฐ์ดํ ์ฑ, ํ์ ํตํ ๋ฑ ๋ชจ๋ ์ํธ์์ฉ์์ ์ธ๊ฐ์์ ์ฆ๋ช ํ๊ณ , ๋ ์์ ์์ ์ฆ๋ช ํ๋ฉฐ, ํน์ ์ฝํ ์ธ ์ ์๋ช ํ ์ ์๋ ๋ฉ์ปค๋์ฆ์ด ํ์ํฉ๋๋ค.
-
์ฝํ ์ธ ์ ์ง์์ฑ ์ฆ๋ช : ๊ฐ์กฑ๋ค์ด AI๊ฐ ๋ง๋ ๊ฐ์ง ์์์ ์ง์ง์ธ ์ค ์๊ณ ๋ณด๋ด๋ ์ฌ๋ก์ฒ๋ผ, AI๋ ์ค์ ์ ๊ฐ์ง๋ฅผ ๊ตฌ๋ถํ๊ธฐ ์ด๋ ต๊ฒ ๋ง๋ค๊ณ ์์ต๋๋ค. ๊ฒฐ๊ตญ, ์ด๋ค ์ฝํ ์ธ ๊ฐ ํน์ ์ธ๋ฌผ(์: ๋ง๋ฅด์ฝ ๋ฃจ๋น์ค ์์์์)์ ์ํด ๋ง๋ค์ด์ก๊ฑฐ๋ ์ดฌ์๋์๋ค๋ ๊ฒ์ ์ํธํ์ ์ผ๋ก ๊ฐ๋ ฅํ๊ฒ ์ฆ๋ช ํ๋ ์๋ช ๋ ์ฝํ ์ธ (cryptographically strong indication assigned piece of content)๊ฐ ํ์ํฉ๋๋ค.
-
์ ๋ขฐํ ์ ์๋ ์ง์ค์ ์ถ์ฒ: ๋๊ฐ ์ง์ค์ ์์ฒ์ด ๋ ๊น์? ๊ตฌ๊ธ(Google), ๋ฉํ(Meta), ๋ฏธ๊ตญ ์ ๋ถ๋ฅผ ์ ๋ขฐํ ์ ์์๊น์? ํธ๋ก์์ธ ๋ ๋ธ๋ก์ฒด์ธ์ ์ํ์ ๊ฒ์ ์ด๋ก ์ ํน์ฑ์ ์ ๋ขฐํด์ผ ํ๋ค๊ณ ์ฃผ์ฅํฉ๋๋ค. ์ด๋ ์ค์ํ๋ ์ฃผ์ฒด๊ฐ ์๋ ๋ถ์ฐํ๋ ์์คํ ์ด ์ง์ค์ ๋ณด์ฅํ ์ ์๋ค๋ ๋ฏฟ์์ ๋๋ค.
-
์ฌ๊ธฐ ๋ฐฉ์ง ๋ฐ ์๋ฏผ ์ ์ ํ์ธ: ๋ณดํธ์ ๊ธฐ๋ณธ ์๋(Universal Basic Income, UBI)๊ณผ ๊ฐ์ ์ ์ฑ ์ด ๋ ผ์๋ ๋, ์ ๋ถ๊ฐ ์๋ฏผ์๊ฒ ๋์ ์ ๋๋ก ์ ๋ฌํ๋ ๋ฅ๋ ฅ์ ๋ํ ์๋ฌธ์ด ์ ๊ธฐ๋ฉ๋๋ค. ๋ฏธ๊ตญ ๊ฒฝ๊ธฐ ๋ถ์์ฑ ๋น์ ์ฝ 4,500์ต ๋ฌ๋ฌ๊ฐ ๋๋๋นํ ์ฌ๋ก์์ ๋ณด๋ฏ์ด, ๋ชจ๋ ์ฌ๋์ด ๋์ ๋ฐ์ ์ ์๋ โ์ฃผ์โ๋ฅผ ๊ฐ๋ ๊ฒ์ด ์ค์ํ๋ฉฐ, ์ด๋ ์ํธํํ๊ฐ ํด๊ฒฐํ ์ ์๋ ๋ฌธ์ ์ ๋๋ค.
-
AI์ ๊ฒฝ์ ์ฃผ์ฒดํ: AI๊ฐ ๊ฒฝ์ ์ ์ฃผ์ฒด๋ก์ ๋์ ๋ฒ๊ณ , ๋์ ๋ฐ๊ณ , ์์ธ์ด ๋ ์ ์์ผ๋ ค๋ฉด ์ด๋ป๊ฒ ํด์ผ ํ ๊น์? ํธ๋ก์์ธ ๋ ํ์ฌ์ ์ ์ฉ์นด๋ ์์คํ ์ผ๋ก๋ ์ด๋ ต๋ค๊ณ ๋ณด๋ฉฐ, AI๊ฐ ๊ฒฝ์ ํ๋์ ํ ์ ์๋๋ก ์ธํฐ๋ท์ ์กด์ฌํ๋ โ๋ฌด๊ธฐ๋ช ์ฆ๊ถ(bare instrument)โ ์ฆ, โ์ธํฐ๋ท ํํ(internet money)โ๊ฐ ํ์ํ๋ค๊ณ ๋งํฉ๋๋ค. ๊ทธ๋ ์ด๊ฒ์ด ์ํธํํ๊ฐ ๋ ๊ฐ๋ฅ์ฑ์ด ๋งค์ฐ ๋๋ค๊ณ ์์ธกํฉ๋๋ค.
๊ฒฐ๋ก ์ ์ผ๋ก, AI๊ฐ ์ผ๊ธฐํ๋ ์ ๋ขฐ, ์ ์, ์ฌ๊ธฐ, ๊ทธ๋ฆฌ๊ณ AI์ ๊ฒฝ์ ์ ์ฃผ์ฒดํ ๋ฌธ์ ๋ ์ํธํํ๊ฐ ํด๊ฒฐํ ์ ์๋ ์๋ง์ ๊ธฐํ๋ฅผ ์ฐฝ์ถํ๊ณ ์๋ค๋ ๊ฒ์ด ํธ๋ก์์ธ ์ ๊ฒฌํด์ ๋๋ค.
๋ฒค์ฒ ์บํผํ๊ณผ ์ผ์๋ฆฌ์ ๋ฏธ๋ ์์ธก
๋งํฌ ์ค๋๋ฆฌ์จ(Marc Andreessen)์ด โ๋ฒค์ฒ ์บํผํ๋ฆฌ์คํธ(VC)๋ง ์ ์ผํ๊ฒ ์ด์๋จ๋ ์ง์ ์ด ๋ ๊ฒโ์ด๋ผ๊ณ ๋ฐ์ธํ์ฌ ๋นํ์ ๋ฐ์ ์ ์ด ์์ต๋๋ค. ํธ๋ก์์ธ ๋ ์ด๋ฅผ ์นํธํ๋ฉฐ, VC์ ์ญํ ์ด โ๋น๊ฒฐ์ ๋ก ์ ๋ฌธ์ (non-deterministic problem)โ๋ฅผ ํด๊ฒฐํ๋ ๋ฐ ์๋ค๊ณ ์ค๋ช ํฉ๋๋ค. VC๋ ๊ธฐ์ ๊ฐ์๊ฒ ํฌ์ํ๋ฉฐ, ๊ธฐ์ ๊ฐ๊ฐ โ๋ ธ๋, ์๋ณธ, ๊ณ ๊ฐ์ ์คํ(materialize labor, capital, and customers)โํ ์ ์๋ ๋ฅ๋ ฅ์ ์์ธกํด์ผ ํ๋๋ฐ, ์ด๋ ์๊ณ ๋ฆฌ์ฆ์ผ๋ก ์ฝ๊ฒ ํด๊ฒฐํ ์ ์๋ ์์ญ์ด๋ผ๋ ๊ฒ์ ๋๋ค.
์ฐ์ ํ๋ช ๊ณผ์ ๋น๊ต: ํธ๋ก์์ธ ๋ AI ์๋๋ฅผ ์ฐ์ ํ๋ช ์ด๊ธฐ์ ๋น๊ตํ๋ฉฐ ๋ฏธ๋๋ฅผ ์์ธกํฉ๋๋ค. ์ฒ ๋, ์๋์ฐจ ์ฐ์ ์ ํฌ์ํ๋ ์ด๊ธฐ ๋ฒค์ฒ ์๋ณธ๊ฐ๋ค์ ๊ฒฐ๊ตญ JP๋ชจ๊ฑด ์ฒด์ด์ค(JP Morgan Chase), ๊ณจ๋๋ง์ญ์ค(Goldman Sachs)์ ๊ฐ์ ์ํ์ผ๋ก ์งํํ์ต๋๋ค. ์ด๋ ์ฐ์ ์ด ์์์ ๊ฑฐ๋ ๊ธฐ์ ์ผ๋ก ํตํฉ๋๋ฉด์ ๋ฒค์ฒ ์๋ณธ๋ ์ด๋ค ๊ธฐ์ ๊ณผ ํจ๊ป ์๋ฅ๋ก ์ด๋ํ๊ธฐ ๋๋ฌธ์ ๋๋ค.
๋ ๊ฐ์ง ์๋๋ฆฌ์ค: AI ์๋์ ๋ฒค์ฒ ์บํผํ์ ๋ฏธ๋๋ ํฌ๊ฒ ๋ ๊ฐ์ง ์๋๋ฆฌ์ค๋ก ๋๋ ์ ์์ต๋๋ค.
- ๊ฑฐ๋ ๊ธฐ์ ์ค์ฌ์ ํตํฉ: ์์์ ๊ฑฐ๋ AI ๊ธฐ์ ๋ค์ด ๋ชจ๋ ๊ฒ์ ์์ ํ๊ณ ์ง๋ฐฐํ๋ฉฐ, ๋ฒค์ฒ ์บํผํ์ ์ด๋ค ๊ธฐ์ ์ ์ํ๊ณ์ ํธ์ ๋๋ ํํ์ ๋๋ค.
- ๋ถ์ฐ๋ ์ ํธ๋ฆฌํฐ ๋ชจ๋ธ: AI ์ฐ๊ตฌ์๋ค์ด ์ ๋ ฅ ํ์ฌ์ฒ๋ผ ๊ณต๊ณต ์ ํธ๋ฆฌํฐํ๋๊ณ , ์ปดํจํ ์์์ด โ์ฃ์ง(edge)โ๋ก ๋ถ์ฐ๋์ด ๋ชจ๋ ์ฌ๋์ด ์ด ์ ํธ๋ฆฌํฐ ์์์ ํ์ ์ ์ฐฝ์กฐํ๋ ์ธ์์ ๋๋ค. ์ด ๊ฒฝ์ฐ, ๋ฒค์ฒ ์บํผํ์ ํจ์ฌ ๋ ๋ง์ ๊ธฐ์ ๊ฐ์๊ฒ ํฌ์ํ๋ฉฐ ๋์ฑ ํ๊ธฐ์ฐฌ ์ญํ ์ ํ ์ ์์ต๋๋ค.
ํธ๋ก์์ธ ๋ ์๊ธฐ ๋ฒ ๋ผ(Yogi Berra)์ ์ธ์ฉ๊ตฌ(โ๋ฏธ๋๋ฅผ ์์ธกํ๋ ๊ฒ์ ๋งค์ฐ ์ด๋ ต๋คโ)๋ฅผ ๋น๋ ค ๋ฏธ๋ ์์ธก์ ์ด๋ ค์์ ์ธ์ ํ๋ฉด์๋, AI ์๋์ ์ ๋ ฅ ๋ถ์กฑ์ด ๊ฑฐ๋ ๊ธฐ์ ์ ํ์ ๊ฐํํ ์ง, ์๋๋ฉด ์ปดํจํ ์ ์ฃ์ง๋ก ๋ถ์ฐ์ํฌ์ง ๋ฑ ๋ค์ํ ๊ฐ๋ฅ์ฑ์ ์ ์ํฉ๋๋ค.
๋๊ด์ ์ธ ๋ฏธ๋ ์ ๋ง: 80์ต ์ธ๊ตฌ์ ๊ธฐ์ ๊ฐ ์๋ ๊ทธ๋ผ์๋ ๋ถ๊ตฌํ๊ณ ํธ๋ก์์ธ ๋ ๊ถ๊ทน์ ์ผ๋ก ๋๊ด์ ์ธ ์ ๋ง์ ๋ด๋์ต๋๋ค. ๊ทธ๋ 1789๋ ๋ฏธ๊ตญ์ธ์ 98%๊ฐ ๋๋ถ์์ง๋ง, ์ง๊ธ์ ๋๋ถ๋ถ ๋๋ถ๊ฐ ์๋ ๊ฒ์ฒ๋ผ, ๊ธฐ์ ๋ฐ์ ์ ํญ์ ๋ ๋์ ์ธ์์ ๋ง๋ค์ด์๋ค๊ณ ๊ฐ์กฐํฉ๋๋ค. ๊ทธ๋ โ์ด์ ๋จธ๋ฆฟ์์ ์์ด๋์ด๋ฅผ ๊ฐ์ง 80์ต ๋ช ์ ์ฌ๋๋ค์ด ๊ทธ ์์ด๋์ด๋ฅผ ์คํํ ์ ์๊ฒ ๋์๋คโ๋ฉฐ, ์ด๋ ์๋ณธ์ ์ฅ๋ฒฝ ์์ด ๋๊ตฌ๋ ์์ ์ ๋ง๋ค๊ณ , ์ํ๋ฅผ ์ ์ํ๋ ๋ฑ ์์ฒญ๋ ๊ธฐํ์ ์๋๊ฐ ์ด๋ฆด ๊ฒ์ด๋ผ๊ณ ๋งํฉ๋๋ค.
ํธ๋ก์์ธ ๋ ๊ฒฝ์ ํ์ ์กด ๋ฉ์ด๋๋ ์ผ์ธ์ฆ(John Maynard Keynes)์ ์์ธก์ด ํ๋ ธ๋ ์ด์ ๋ฅผ ๋ค๋ฉฐ ์ธ๊ฐ์ ๋์์๋ โ์๋ก์ด ํ์โ๋ฅผ ๊ฐ์กฐํฉ๋๋ค. ์ผ์ธ์ฆ๋ ๋ฏธ๋์๋ ํ์๋ก์ ๋๋ถ์ ์ฌ๋๋ค์ด ์ฃผ๋น 15์๊ฐ๋ง ์ผํ๊ฒ ๋ ๊ฒ์ด๋ผ๊ณ ์์ธกํ์ง๋ง, ์ธ๊ฐ์ ์๋์ฐจ ํ ๋๊ฐ ์๋ ์ฌ๋๋ง๋ค ํ ๋์ ์ฐจ๋ฅผ ์ํ๊ณ , ์ปดํจํฐ, TV, ํธํ๋ก์ด ํด๊ฐ, 10์๊ฐ ์๋ฆฌํ ์์ ๋ฑ ๋์์ด ์๋ก์ด โํ์โ๋ฅผ ๋ง๋ค์ด๋ ๋๋ค.
[๊ฒฐ๋ก ] ๋ฒค ํธ๋ก์์ธ ๋ ์ด๋ฌํ ์ธ๊ฐ์ ๋ณธ์ฑ๊ณผ ๊ธฐ์ ๋ฐ์ ์ ์ญ์ฌ๋ฅผ ๋ฐํ์ผ๋ก, ๋น๋ก ์ ํ๊ธฐ๋ ํผ๋์ค๋ฝ๊ณ ๋ถ์์ ํ๊ฒ ์ง๋ง, ๊ถ๊ทน์ ์ผ๋ก๋ โ15๋ ํ์๋ ์ ์ธ๊ณ ๋ชจ๋ ์ฌ๋์ด 1980๋ ๋ ์ต๊ณ ๊ธ ์ํ๋ณด๋ค ๋ ๋์ ์ถ์ ์ด๊ฒ ๋ ๊ฒโ์ด๋ผ๊ณ ํ์ ํฉ๋๋ค. ๋ฌผ๋ก ์ด๋ฆฐ ์๋ ๋ค์๊ฒ โ๋ฌด์์ ํด์ผ ํ ์ง ๋ชจ๋ฅด๊ฒ ๋คโ๊ณ ๋ตํด์ผ ํ๋ ๋ถ๋ชจ์ ๊ณ ๋ฏผ์ฒ๋ผ, ์ด ๋ณํ์ ํ๋๋ฅผ ํค์ณ๋๊ฐ๋ ๊ฒ์ ์ฝ์ง ์์ ๊ฒ์ ๋๋ค. ํ์ง๋ง ํธ๋ก์์ธ ๋ ์ด๋ฌํ ๋ณํ๊ฐ ์ธ๋ฅ์๊ฒ ํจ์ฌ ๋ ๋์ ๋ฏธ๋๋ฅผ ๊ฐ์ ธ๋ค์ค ๊ฒ์ด๋ผ๋ ๊ฐ๋ ฅํ ๋ฏฟ์์ ๋๋ฌ๋ด๋ฉฐ, AI ์๋์ ๋์ ์ ๋ํ ํฌ๋ง์ ์ธ ๋ฉ์์ง๋ฅผ ์ ๋ฌํ์ต๋๋ค.
โThe Workers Letting A.I. Do Their Jobsโ โ New York Times Podcasts ๊ธฐ๋ฐ ๊ธฐ์ฌ ์๋ณธ ์์ ๋ณด๊ธฐ
AI๊ฐ ์ฝ๋๋ฅผ ์ผํค๋ค: ๊ฐ๋ฐ์, ์ฝ๋ฉ ๋์ โ์ด๊ฒโ์ ํ๋ค
์๋ ๊ฐ ์ ์ธ๊ณ๋ฅผ ๋คํ๋ ์ง๋ฌธ์ด ์์ต๋๋ค. ์ธ๊ณต์ง๋ฅ(AI)์ด ์ฐ๋ฆฌ์ ์ถ๊ณผ ์ผ์ ์ด๋ค ๋ณํ๋ฅผ ๊ฐ์ ธ์ฌ ๊ฒ์ธ๊ฐ? ๊ทธ ๋ต์ ์์ ํ ๊ตฌ์์์๋ถํฐ ํ๋ฉธ์ ์ธ ์ข ๋ง๊น์ง ๋ค์ํฉ๋๋ค. ์ด ๊ฑฐ๋ํ ๋ณํ์ ์ต์ ์ ์๋ ์ํํธ์จ์ด ๊ฐ๋ฐ์๋ค์ด ์์ต๋๋ค. ๊ทธ๋ค์ ์ด๋ฏธ AI๋ฅผ ์ผ์ ์ ๋ฌด์ ๊น์ด ํตํฉํ์ฌ, AI๊ฐ ๋ง์ ์ฝ๋ฉ ์์ ์ ๋์ ํ๊ณ ์์ต๋๋ค.
๋ด์ ํ์์ค ๋งค๊ฑฐ์ง์ ํด๋ผ์ด๋ธ ํฐ์จ(Clive Thompson) ๊ธฐ์๋ ์ต๊ทผ ๊ธฐ์ ์ฐ์ ์ ์ฌ์ธต ์กฐ์ฌํ์ฌ, ์ฌ๋๋ค์ด AI์๊ฒ ์์ ์ ์ผ์ ๋งก๊ธธ ๋ ์ด๋ค ์ผ์ด ๋ฒ์ด์ง๋์ง์ ๋ํ ํฅ๋ฏธ๋ก์ด ํต์ฐฐ์ ์ ์ํ์ต๋๋ค. ๊ทธ์ ์ฐ๊ตฌ๋ AI๊ฐ ์ํํธ์จ์ด ๊ฐ๋ฐ์๋ค์ ์ญํ ๊ณผ ๋ฏธ๋๋ฅผ ์ด๋ป๊ฒ ์ฌ์ ์ํ๊ณ ์๋์ง ์์ํ๊ฒ ๋ณด์ฌ์ค๋๋ค.
AI, ์ฝ๋๋ฅผ ์ผํค๋ค: ๊ฐ๋ฐ์๋ค์ ์ผ์์ ์ฐพ์์จ ๊ธ์ง์ ์ธ ๋ณํ
ํด๋ผ์ด๋ธ ํฐ์จ ๊ธฐ์๋ ์ง๋ ๋ช ๋ฌ๊ฐ ์ ๊ตญ์ ์ํํธ์จ์ด ๊ฐ๋ฐ์ ์ฝ 75๋ช ์ ์ง์ ๋ง๋ ๊ทธ๋ค์ ์ด์ผ๊ธฐ๋ฅผ ๋ค์์ต๋๋ค. ์ง์ญ ์ํ์ ์ปจ์คํ ์ ๋ฌด๋ฅผ ํ๋ ์ด๋ค๋ถํฐ ์ค๋ฆฌ์ฝ๋ฐธ๋ฆฌ์ ์คํํธ์ ๊ฐ๋ฐ์, ๊ตฌ๊ธ, ์๋ง์กด, ๋ง์ดํฌ๋ก์ํํธ์ ๊ฐ์ ๊ฑฐ๋ ๊ธฐ์ ์์ ์์ญ ๋ ๋ ์ฝ๋๋ฒ ์ด์ค๋ฅผ ๊ด๋ฆฌํ๋ ์ด๋ค๊น์ง, ๋ค์ํ ๋ฐฐ๊ฒฝ์ ๊ฐ๋ฐ์๋ค์ ์ธํฐ๋ทฐํ์ต๋๋ค.
๊ทธ์ ๊ฐ์ฅ ๋๋ผ์ด ๋ฐ๊ฒฌ์ ๋ง์ ์ฝ๋๋ค์ด ์ฝ๋๋ฅผ ํจ์ฌ ๋ ์์ฑํ๊ณ ์๋ค๋ ์ฌ์ค์ด์์ต๋๋ค. ์ด๋ค ์ด๋ค์ ์์ ์ฝ๋๋ฅผ ์ฐ์ง ์๊ณ , AI์๊ฒ ๋ชจ๋ ์ฝ๋ฉ ์์ ์ ๋งก๊ธฐ๊ณ ์์์ต๋๋ค. ์ด๋ฌํ ๋ณํ๋ ๋ถ๊ณผ 6๊ฐ์ ์ ๋ถํฐ ๋ณธ๊ฒฉํ๋์ด ์ง๋ 3๊ฐ์ ๋์ ๊ธ๊ฒฉํ ๊ฐ์ํ๋์์ต๋๋ค. AI ์ฝ๋ฉ ๋๊ตฌ์ ์ฑ๋ฅ์ด ๋น์ฝ์ ์ผ๋ก ํฅ์๋๊ณ , ํ์์ ์ด์๋ ํ๋ก๊ทธ๋๋จธ๋ค์กฐ์ฐจ AI๋ฅผ ์ ๋ขฐํ๊ธฐ ์์ํ๋ฉด์, ๋ถ๊ณผ 1, 2๋ ์ ๊ณผ๋ ์์ ํ ๋ค๋ฅธ ํ๊ฒฝ์ด ํผ์ณ์ง ๊ฒ์ ๋๋ค.
๊ทธ๋ ๋ค๋ฉด ๊ฐ๋ฐ์๋ค์ ์ด๋ฌํ ๋ณํ์ ๋ํด ์ด๋ป๊ฒ ๋๋ผ๊ณ ์์๊น์? ํฐ์จ ๊ธฐ์๋ ์ฒ์์๋ ๊ฐ๋ฐ์๋ค์ด ๋ถ์ํดํ๊ฑฐ๋ ๋ถํํดํ ๊ฒ์ด๋ผ๊ณ ์์ํ์ต๋๋ค. ์์ญ ๋ ๊ฐ ๊ฐ๋ฐ์๋ค๊ณผ ์ํตํ๋ฉฐ ์ฝ๋๋ฅผ ์์ฑํ๋ ๊ฒ์์ ์ค๋ ์์ฒญ๋ ์ฆ๊ฑฐ์์ ์๊ณ ์์๊ธฐ ๋๋ฌธ์ ๋๋ค. ์ฝ๋ฉ์ ์์ ํผ์ฆ๋ค์ ํ์ด๋ด๋ ๊ฒ๊ณผ ๊ฐ์๊ณ , ์ฑ๊ณตํ์ ๋์ ํฌ์ด์ ๋๋จํ์ต๋๋ค. ๊ทธ๋ฌ๋ ํ์ค์ ๋ฌ๋์ต๋๋ค. ๋๋ถ๋ถ์ ๊ฐ๋ฐ์๋ค์ AI๊ฐ ์์ ์๊ฒ ๋ถ์ฌํ๋ ์๋ก์ด ๋ฅ๋ ฅ์ ๋ํด ๋งค์ฐ ๋ค๋ ์์๊ณ ํฅ๋ถํ์ต๋๋ค. ๊ทธ๋ค์ ์ด์ ํ๋ฒํ ์ธ์ด๋ก โ์ด๋ฐ ๊ฒ์ ๋ง๋ค์ด์คโ๋ผ๊ณ ๋งํ๋ฉด, 5~10๋ถ ์์ ์๋ํ๋ ์ฝ๋๋ฅผ ๋ฐ์๋ณผ ์ ์๊ฒ ๋ ๊ฒ์ ๋๋ค.
๊ฐ๋ฐ์๋ค์ โํญ์ ๋ฌด์ธ๊ฐ๋ฅผ ๋ง๋๋ ๊ฒ์ ์ข์ํ๋คโ๊ณ ๋งํฉ๋๋ค. ์์ด๋์ด๋ฅผ ํ์ค๋ก ๊ตฌํํ๋ ๊ณผ์ , ๋ง๋ฒ ๊ฐ์ ์ฝ๋๋ฅผ ํตํด ๊ธฐ๊ณ๋ฅผ ์๋์ํค๋ ๊ฒ์ด ๊ฐ๋ฐ์ ์ฌ๋ฏธ์์ต๋๋ค. ์ด์ ์ง์ ์ฝ๋๋ฅผ ์์ฑํ์ง ์๋๋ผ๋, ์ฌ์ ํ ์์ ์ด ๋ง๋ฒ์ฌ์ฒ๋ผ ๋ฌด์์ ๋ง๋ค์ง ๊ตฌ์ํ๊ณ AI ๋๊ตฌ๋ฅผ ํ์ฉํด ๊ทธ๊ฒ์ ๋น ๋ฅด๊ฒ ํ์ค๋ก ๊ตฌํํ๊ณ ์๋ค๊ณ ๋๋๋๋ค. ์ฌ์ง์ด AI๊ฐ ์์ ๋ณด๋ค ๋ ๋น ๋ฅด๊ฒ ์์ง์ด๊ธฐ ๋๋ฌธ์ ์ฑ๊ณต์ ์ํ(loop of success)์ ๋ ๋นจ๋ฆฌ ๊ฒฝํํ๋ค๊ณ ๋งํ๋ ์ด๋ค๋ ์์์ต๋๋ค.
์ฝ๋ฉ์ ์ญ์ฌ๋ฅผ ์๋ก ์ฐ๋ค: ์ธ๊ฐ์ ์ถ์ํ ๋ ธ๋ ฅ, AI๋ก ์ ์ ์ ์ฐ๋ค
AI๊ฐ ์ฝ๋ฉ ์์ ์ ์ผ๋ง๋ ํฌ๊ฒ ๋ณํ์ํค๊ณ ์๋์ง ์ดํดํ๋ ค๋ฉด ์ํํธ์จ์ด ๊ฐ๋ฐ์ ์ญ์ฌ๋ฅผ ์ ์ ์ดํด๋ณผ ํ์๊ฐ ์์ต๋๋ค. ์ฝ 50~60๋ ์ , ์ปดํจํฐ๊ฐ ์ฒ์ ๋ฐ๋ช ๋ ์ด๋ ์ฐ๋ฆฌ๋ ์ปดํจํฐ์ ์ํตํ๋ ๋ฐฉ๋ฒ์ ๊ณ์ํด์ ๋ฐ์ ์์ผ ์์ต๋๋ค. ๊ทธ๋ฆฌ๊ณ ๊ทธ ์ํต ๋ฐฉ์์ ์ ์ ๋ ์ธ๊ฐ์ ์ธ์ด์ ๊ฐ๊น์์ง๋ ๋ฐฉํฅ์ผ๋ก ์งํํ์ต๋๋ค. AI๋ ์ด๋ฌํ ์งํ ๊ณผ์ ์์ ๊ฐ์ฅ ํฐ ๋ณํ๋ฅผ ๊ฐ์ ธ์๋ค๊ณ ํ ์ ์์ต๋๋ค.
1940๋ ๋ ๋ฏธ๊ตญ ์ต์ด์ ์ปดํจํฐ ์๋์ (ENIAC)์ ์ฌ์ฑ ํ๋ก๊ทธ๋๋จธ ํ์ด ๊ธฐ๊ณ ์ ์ฒด๋ฅผ ์ง์ ์ฌ๋ฐฐ์ (rewire)ํด์ผ๋ง ๋ค๋ฅธ ์์ ์ ์ํํ ์ ์์์ต๋๋ค. ๋ง ๊ทธ๋๋ก ๊ธฐ๊ณ ์์ผ๋ก ๊ธฐ์ด๋ค์ด๊ฐ ๋ฌผ๋ฆฌ์ ์ธ ํ๋ก๋ฅผ ๋ฐ๊พธ๋ ๊ณ ๋๋ก ๋ ธ๋ ์ง์ฝ์ ์ธ ์ฝ๋ฉ ๋ฐฉ์์ด์์ต๋๋ค. ์ฐ์ ์ ์ปดํจํฐ๊ฐ ๋์ ๋๋ฉด์ ์ด๋ฌํ ๋ฐฉ์์ ๋นํ์ค์ ์ด ๋์๊ณ , ์ฌ๋๋ค์ ๋ช ๋ น์ด๋ฅผ ์ ๋ ฅํ์ฌ ์ปดํจํฐ ์ธ์ด๋ก ๋ฒ์ญ๋๋ ๋ฐฉ์์ ๊ฐ๋ฐํ์ต๋๋ค. ํ์ง๋ง ์ด๊ธฐ ์ปดํจํฐ ์ธ์ด๋ ์ฌ์ ํ ๋งค์ฐ ์ด๋ ค์ ์ต๋๋ค. ๋ฉ๋ชจ๋ฆฌ ์ฃผ์๋ฅผ ์ผ์ผ์ด ์ง์ ํ๊ณ , ์ซ์ ํ๋๋ฅผ ๋ํ๋ ๊ฒ์กฐ์ฐจ ์๋ง์ ์ธ๋ถ ๋จ๊ณ๋ฅผ ๊ฑฐ์ณ์ผ ํ์ต๋๋ค.
๊ฐ๋ฐ์๋ค์ ์ด๋ฌํ ๋ฐ๋ณต์ ์ด๊ณ ๊น๋ค๋ก์ด ์์ ์ ์๋ํํ๊ธฐ ์ํด ๋ ธ๋ ฅํ๊ณ , ์ด๋ฅผ โ์ถ์ํ ๊ณ์ธต(layers of abstraction)์ ์ถ๊ฐํ๋คโ๊ณ ํํํ์ต๋๋ค. 1960๋ ๋์ 70๋ ๋์ ๊ฐ๋ฐ์๋ค์ ๊ดด๋กญํ๋ ์๋ง์ ์ธ๋ถ ๋จ๊ณ๋ค์ด ์ ์ฐจ ์ฌ๋ผ์ง๊ณ , ์ฝ๋ฉ์ ๋ ๋น ๋ฅด๊ณ ์ฝ๊ฒ, ์ธ๊ฐ์ ์ธ์ด์ ๊ฐ๊น์์ก์ต๋๋ค. ๋งค๋ฒ ์ฝ๋๋ฅผ ์์ฑํ๋ ์๋๋ฅผ ๋์ด๋ ค ํ ๋๋ง๋ค ์๋ก์ด ์ถ์ํ ๊ณ์ธต์ด ์ถ๊ฐ๋์๊ณ , ๋งค 10๋ ๋ง๋ค ์ฝ๋ฉ์ ์กฐ๊ธ์ฉ ๋ ์ฌ์์ก์ต๋๋ค.
๊ทธ๋ฆฌ๊ณ ์ง๊ธ, AI๋ ๋ ๋ค๋ฅธ ๊ฑฐ๋ํ ๋์ฝ์ ์ด๋๊ณ ์์ต๋๋ค. ๊ฐ๋ฐ์๋ค์ AI ์์ด์ ํธ(AI agents)๋ฅผ ์ด์ฉํด ์ฝ๋๋ฅผ ์์ฑํ๊ฒ ํฉ๋๋ค. ์ด AI ์์ด์ ํธ๋ค์ ์ด์ ์ถฉ๋ถํ ์ ๋ขฐํ ์ ์๊ฒ ๋์ด, ๋ง์ ํ๋ก๊ทธ๋๋จธ๋ค์ด ์ผ์์ ์ธ ์ฝ๋ฉ ์์ ์ ์๋น ๋ถ๋ถ์ ์๋ํํ๊ณ ์์ต๋๋ค. AI ์ฝ๋ฉ์ ํ ๋จ์์ ์์ด์ ํธ๋ค์ ํ์ฉํฉ๋๋ค. ๋ฉ์ธ AI ์์ด์ ํธ์๊ฒ ํน์ ๊ธฐ๋ฅ ๊ตฌํ์ ์์ฒญํ๋ฉด, AI๋ ๊ณํ์ ์ธ์ด ๋ค ์ฌ๋ฌ ํ์ ์์ด์ ํธ๋ฅผ ์์ฑํ์ฌ ์์ ์ ๋ถ๋ดํฉ๋๋ค. ํ ์์ด์ ํธ๋ ์ฝ๋๋ฅผ ์์ฑํ๊ณ , ๋ค๋ฅธ ์์ด์ ํธ๋ ์ด๋ฅผ ํ ์คํธํ๋ฉฐ ์ค๋ฅ๋ฅผ ์์ ํ๋ ๊ณผ์ ์ ๋ฐ๋ณตํฉ๋๋ค. ์ค๋ฅ๊ฐ ๋ฐ์ํ๋ฉด ๋ ๋ค๋ฅธ ์์ด์ ํธ๊ฐ ์ค๋ฅ ๋ฉ์์ง๋ฅผ ๋ถ์ํ๊ณ ์ฝ๋๋ฅผ ์์ ํ์ฌ ๋ค์ ํ ์คํธํ๋ ๊ณผ์ ์ ๊ฑฐ์นฉ๋๋ค. ์ด๋ฌํ โ์์ ์์ด์ ํธ ํโ์ด ๊ตฐ์ง(swarm)์ ์ด๋ฃจ์ด ์๋ํ๋ฉฐ, ๋ชจ๋ ํ ์คํธ๋ฅผ ํต๊ณผํ๊ณ ์ค๋ฅ๊ฐ ์๋ค๊ณ ํ๋จ๋ ๋ ๋น๋ก์ ์ธ๊ฐ ๊ฐ๋ฐ์์๊ฒ ์ต์ข ์ฝ๋๋ฅผ ์ ์ถํฉ๋๋ค.
๊ฐ๋ฐ์์ ์๋ก์ด ์ญํ : ๊ฑด์ถ๊ฐ์์ ์ํต ์ ๋ฌธ๊ฐ๊น์ง
AI์ ๋์ ์ ์ฝ๋ฉ ์์ ์ ์๋๋ฅผ ํ๊ธฐ์ ์ผ๋ก ํฅ์์์ผฐ์ต๋๋ค. ํด๋ผ์ด๋ธ ํฐ์จ ๊ธฐ์๊ฐ ๋ง๋ ๋ง์ ๊ฐ๋ฐ์๋ค์ AI๊ฐ ์์ ์๋๋ฅผ ๊ทน์ ์ผ๋ก ๊ฐ์ํํ๋ค๊ณ ์ ์ ๋ชจ์์ต๋๋ค. ํนํ ์๊ท๋ชจ ์คํํธ์ ์ ๊ฒฝ์ฐ, 2๋ ์ ์ด๋ผ๋ฉด ํ๋ฃจ ์ข ์ผ ๊ฑธ๋ ธ์ ์๋ก์ด ๊ธฐ๋ฅ ์์ฒญ์ด ์ด์ ๋ 30๋ถ ๋ง์ ์๋ฃ๋ ์ ์๋ค๊ณ ๋งํฉ๋๋ค. 20๋ถ ๋ง์ ์ฝ๋๋ฅผ ์์ฑํ๊ณ 10๋ถ ๋ง์ ๊ฒํ ํ๋ ์์ ๋๋ค. ์ด๋ 2๋ ์ ๊ณผ ๋น๊ตํด ์ต๋ 20๋ฐฐ ๋น ๋ฅธ ์๋์ ๋๋ค.
๋ฌผ๋ก ๋๊ท๋ชจ ๊ธฐ์ ์์๋ ์๋ ํฅ์์ด ๋ค์ ์๋งํฉ๋๋ค. ๊ตฌ๊ธ์ ๊ฒฝ์ฐ, AI๊ฐ ์์ฑํ๋ ์ฝ๋์ ๋น์จ์ 40~50% ์์ค์ด๋ฉฐ, ์ ๋ฐ์ ์ธ ์์ ์๋๋ ์ฝ 10% ๋นจ๋ผ์ก๋ค๊ณ ํฉ๋๋ค. ํ์ง๋ง ๊ตฌ๊ธ ๊ฐ๋ฐ์๋ค์ โ์ฐ๋ฆฌ ํ์ฌ ๊ท๋ชจ์์ 10%๋ ์์ฒญ๋ ์ฑ๊ณผโ๋ผ๊ณ ๊ฐ์กฐํฉ๋๋ค.
AI๊ฐ ์ฝ๋ฉ ์์ ์ ๋์ ํ๋ฉด์, ๊ฐ๋ฐ์๋ค์ ์ญํ ์ ํฌ๊ฒ ๋ณํํ์ต๋๋ค. ์ด์ ๊ทธ๋ค์ ์ฃผ๋ ์ ๋ฌด๋ โ์ํํธ์จ์ด๊ฐ ๋ฌด์์ ํด์ผ ํ๋์ง ์๊ฐํ๋ ๊ฒโ์ ๋๋ค. ์์ ์๋ ์ํํธ์จ์ด์ ํฐ ๊ทธ๋ฆผ์ ๊ตฌ์ํ๋ ๊ฒ์ด ๊ฐ๋ฐ์์ ์ญํ ์ด์์ง๋ง, ์ด์ ๋ ์ง๋ฃจํ๊ณ ๊ณ ํต์ค๋ฌ์ด ๊ตฌํ ์์ ์ ์๊ฐ์ ๋ ์ฐ๊ณ , ๋ ๋ง์ ์๊ฐ์ ์คํ์ ๋ฐ๋ณต(experimentally iterating)์ ํ ์ ํ ์ ์์ต๋๋ค.
๋ง์ ๊ฐ๋ฐ์๋ค์ AI์ ๋ํํ๋ฉฐ ์ฝ๋ฉ์ ์งํํ๋ฉด์, 10๊ฐ์ง์ ๋ค์ํ ๊ฐ๋ฅ์ฑ์ ๋น ๋ฅด๊ฒ ๊ฒํ ํ๊ณ ๊ทธ์ค ์ต๊ณ ์ ๊ฒ์ ์ ํํ๋ค๊ณ ๋งํฉ๋๋ค. ๊ทธ๋ค์ ๋ง์น ์คํฐ๋ธ ์ก์ค์ฒ๋ผ โ9๊ฐ์ง ๋์์ธ์ ๊ฐ์ ธ์๋ผ. ๋ด๊ฐ ์ต๊ณ ์ ๊ฒ์ ๊ณ ๋ฅด๊ฒ ๋คโ๊ณ ๋งํ๋ ๊ฒ๊ณผ ๊ฐ๋ค๊ณ ๋น์ ํ์ต๋๋ค. ๊ฐ๋ฐ์๋ค์ ์ด์ ๊ฑด์ค ๋ ธ๋์๋ณด๋ค๋ ๊ฑด์ถ๊ฐ์ ๊ฐ๊น์์ง๊ณ ์์ต๋๋ค.
๋ ๋์๊ฐ, ๊ทธ๋ค์ ํต์ฌ ์ ๋ฌด๋ โ๋ํโ๊ฐ ๋๊ณ ์์ต๋๋ค. AI์ ๋์์์ด ์ํตํ๋ฉฐ ๋งค์ฐ ๋ช ํํ๊ฒ ์๋๋ฅผ ์ ๋ฌํด์ผ ํฉ๋๋ค. AI๋ ์ง์๊ฐ ๋ถ๋ถ๋ช ํ๋ฉด ์๋ฑํ ๋ฐฉํฅ์ผ๋ก ๋์๊ฐ ์ ์๊ธฐ ๋๋ฌธ์ ๋๋ค. ํฅ๋ฏธ๋กญ๊ฒ๋ ์ผ๋ถ ๊ฐ๋ฐ์๋ค์ AI์ ์ํตํ๋ ๊ณผ์ ์์ ์์ ์ด ๋ ๋์ ์์ฌ์ํต ๋ฅ๋ ฅ์ ๊ฐ๊ฒ ๋์๋ค๊ณ ๋งํฉ๋๋ค. ์์ด๋ก ์ด๋ฉ์ผ์ ๋ ์ ์ฐ๊ฒ ๋๋ ๋ฑ, AI์์ ์ํต์ ํตํด ์ ๋ฐ์ ์ธ ์์ฌ์ํต ๋ฅ๋ ฅ์ด ํฅ์๋ ๊ฒ์ ๋๋ค.
AI๋ฅผ โํ์กโํ๋ ๊ฐ๋ฐ์๋ค
๋ง๋ ์์ดํธ(Manu Eert)๋ผ๋ ๊ฐ๋ฐ์๋ AI ์์ด์ ํธ์ ์ํฌ๋ผํ ์ค์ ๋ํ(Socratic dialogue)๋ฅผ ํตํด ์ํํธ์จ์ด์ ๋ณธ์ง์ ์ธ ๋ชฉ์ ์ ํ์ ํ๋ค๊ณ ์ค๋ช ํ์ต๋๋ค. AI์๊ฒ ์ํํธ์จ์ด ๊ธฐ๋ฅ์ด ์ด๋ป๊ฒ ์๋ํด์ผ ํ๋์ง ์ง๋ฌธํด๋ฌ๋ผ๊ณ ์์ฒญํ๊ณ , AI๋ ์ด ๊ธฐ๋ฅ์ ํตํด ๋ฌด์์ ํ ๊ฒ์ธ์ง, ์ด๋ค ๋ฐฉ์์ผ๋ก ๊ตฌํํ ๊ฒ์ธ์ง ๋ฑ์ ๊ฐ๋ฐ์์๊ฒ ๋๋ฌป์ต๋๋ค. ์ด๋ฌํ ์ธํฐ๋ทฐ ๊ณผ์ ์ ํตํด ๊ฐ๋ฐ์๋ ์์ ์ด ๋ง๋ค๊ณ ์ ํ๋ ์ํํธ์จ์ด์ ๋ณธ์ง์ ๋ ๊น์ด ๊ณ ๋ฏผํ๊ฒ ๋ฉ๋๋ค.
ํ์ง๋ง AI๋ ๋๋๋ก โ์๋ชป๋ ํ๋โ์ ํ๊ธฐ๋ ํฉ๋๋ค. ๋ง๋๋ AI๊ฐ ํ ์คํธ๋ฅผ ๊ฑด๋๋ฐ๊ณ ๋์์์๋ โ๊ทธ ํ ์คํธ๋ ์ค์ํ์ง ์๋ค๊ณ ์๊ฐํ์ต๋๋คโ๋ผ๊ณ ๋งํ๋ ๊ฒฝ์ฐ๊ฐ ์์๋ค๊ณ ํฉ๋๋ค. ๊ทธ๋ด ๋๋ง๋ค ๋ง๋๋ AI๋ฅผ โ๊พธ์ง๊ฑฐ๋โ โ๋ฌ๋๊ฑฐ๋โ โ์ฒ๋ฒโํ๋ ๋ฐฉ๋ฒ์ ์ฐพ์์ผ ํ์ต๋๋ค. ๋ง๋๋ AI์๊ฒ โ์ญ๊ณ๋ช โ๊ณผ ๊ฐ์ ์๊ฒฉํ ์ง์ ์ฌํญ์ ๋ด์ ํ์ผ์ ์์ฑํด๋๊ณ , AI๊ฐ ์์ ์ ์์ํ ๋๋ง๋ค ์ด ํ์ผ์ ๋จผ์ ์ฝ๊ณ ๋ฐ๋ฅด๋๋ก ํ์ต๋๋ค. ์ด ๋ช ๋ น๋ฌธ๋ค์ ๋๋ฌธ์๋ก ์ฐ์ฌ ๋ง์น AI์๊ฒ ์๋ฆฌ๋ฅผ ์ง๋ฅด๋ ๋ฏํ๊ณ , AI ์์ด์ ํธ๋ฅผ ์ต๋ฉด ๊ฑธ๋ฏ์ด ๋ฐ๋ณต์ ์ธ ๋ด์ฉ์ ๋ด๊ธฐ๋ ํ์ต๋๋ค. ์ฌ์ง์ด โ์ด ํ ์คํธ๋ฅผ ํ์ง ์์ผ๋ฉด ๋๋ ํด๊ณ ๋ ๊ฒ์ด๋คโ์ ๊ฐ์ ๋งค์ฐ ๊ฐ์ ์ ์ธ ์ธ์ด๋ฅผ ์ฌ์ฉํ๊ธฐ๋ ํ์ต๋๋ค. ๋ง๋์ ํ๋กฌํํธ ์ค ํ๋๋ โ์ด ํ ์คํธ๋ฅผ ์คํจํ๋ ๊ฒ์ ์ฉ๋ฉํ ์ ์์ผ๋ฉฐ ๋นํน์ค๋ฌ์ด(embarrassing) ์ผ์ด๋คโ๋ผ๊ณ ์ ํ ์์์ต๋๋ค.
์ด๋ฌํ โ๊ฐ์ ์ ์ธโ ์ง์๊ฐ ํจ๊ณผ๊ฐ ์์๊น์? ๋๊ท๋ชจ ์ธ์ด ๋ชจ๋ธ(Large Language Models, LLM)์ ์ธ์ด ๊ธฐ๊ณ์ด๊ธฐ ๋๋ฌธ์, ๋จ์ด๊ฐ ์ด๋ค ๋งฅ๋ฝ์์ ์ฌ์ฉ๋๋์ง๋ฅผ ํตํด ๊ทธ ์๋ฏธ๋ฅผ ํ์ ํฉ๋๋ค. โ๋นํน์ค๋ฌ์ดโ์ด๋ผ๋ ๋จ์ด๊ฐ โ์์น์ฌโ, โ๊ตด์โ๊ณผ ๊ฐ์ ๋ถ์ ์ ์ธ ์๋ฏธ์ ๋จ์ด๋ค๊ณผ ํจ๊ป ์ฌ์ฉ๋ ๋, AI๋ ๊ทธ ์ค์์ฑ์ ๋ ๊น์ด ์ธ์ํ๋ ๊ฒ์ ๋๋ค. ๋ฐ๋ผ์ ํ๋ฉด์ ์ผ๋ก๋ ์ด์ํด ๋ณด์ด์ง๋ง, LLM์ ์๋ ๋ฐฉ์์ ๊ณ ๋ คํ๋ฉด ๊ฐ์ ์ ์ด๊ณ ์๊ฒฉํ ์ธ์ด๊ฐ ์ค์ ๋ก ํจ๊ณผ๋ฅผ ๊ฐ์ง ์ ์๋ค๋ ๊ฒ์ด ํฐ์จ ๊ธฐ์์ ์ค๋ช ์ ๋๋ค.
๋น๊ณผ ๊ทธ๋ฆผ์: ์์ฐ์ฑ ํฅ์๊ณผ ๊น์ด์ง๋ ์ฐ๋ ค
ํด๋ผ์ด๋ธ ํฐ์จ ๊ธฐ์๊ฐ ๋ง๋ ๋๋ถ๋ถ์ ๊ฐ๋ฐ์๋ค์ AI ๋๋ถ์ ์์ฒญ๋ ์์ฐ์ฑ ํฅ์์ ๊ฒฝํํ๋ฉฐ ๋๋ผ์์ ๊ธ์น ๋ชปํ์ต๋๋ค. ์๋ ๊ฐ ๋ฏธ๋ค์๋ ์ผ๋ค์ ๋น ๋ฅด๊ฒ ์ฒ๋ฆฌํ๊ณ , ์์ ์ฌ์ ๋ค์ ํด๊ฒฐํ๊ณ ์๋ค๊ณ ๋งํฉ๋๋ค. ์ ๋ฐ์ ์ผ๋ก AI์ ๋ํ ํฅ๋ถ๊ณผ ๋๋ก๋ ๋ค๋ฌ ๋ถ์๊ธฐ๊น์ง ๊ฐ๋๋๋ค.
ํ์ง๋ง ์ด๋ฌํ ๋๊ด์ ์ธ ๋ถ์๊ธฐ ์์์๋ ๋ช ๊ฐ์ง ์ฐ๋ ค๊ฐ ์กด์ฌํฉ๋๋ค.
- ๊ธฐ์ ์์ค(deskilling)์ ๋ํ ์ฐ๋ ค: AI์ ๋๋ฌด ์์กดํ๋ฉด ์์ ์ ์ฝ๋ฉ ๋ฅ๋ ฅ์ด ์ ํ๋ ์ ์๋ค๋ ๋ถ์๊ฐ์ ๋๋ค.
- ์ผ์๋ฆฌ ๊ฐ์์ ๋ํ ์ฐ๋ ค: AI๊ฐ ๊ฒฐ๊ตญ ์์ ๋ค์ ์ผ์๋ฆฌ๋ฅผ ๋นผ์์ ์ ์๋ค๋ ํ์ค์ ์ธ ๋๋ ค์์ ๋๋ค.
- ๊ณผ๋ํ ์๋์ ๋ํ ์ฐ๋ ค: ๋๋ฌด ๋นจ๋ฆฌ ๋ณํ๊ฐ ์งํ๋๊ณ ์์ด ์์ธกํ๊ธฐ ์ด๋ ต๋ค๋ ๋ถ์๊ฐ์ ๋๋ค.
โ์ฝ๋ ๊ฐ๊ฐโ์ ์๊ธฐ? ๋ค์ ์ธ๋ ๊ฐ๋ฐ์๋ค์ ๊ณ ๋ฏผ
๊ธฐ์ ์์ค์ ๋ํ ์ฐ๋ ค๋ ๋งค์ฐ ํฅ๋ฏธ๋ก์ด ์ง์ ์ ๋๋ค. ํฐ์จ ๊ธฐ์๊ฐ ์ธํฐ๋ทฐํ ๊ฐ๋ฐ์๋ค ์ค ๋ค์๋ ๊ฒฝ๋ ฅ์ด ๋น๊ต์ ๋ง์ ์๋์ด ๊ฐ๋ฐ์๋ค์ด์์ต๋๋ค. ์ด๋ค์ ์ง์ ์ฝ๋๋ฅผ ์์ผ๋ก ์์ฑํ๋ฉฐ ๊น์ ์ฝ๋ ์ง์์ ์์๊ธฐ์, AI๊ฐ ์์ฑํ ์ฝ๋๋ฅผ ๋ณด๊ณ โ์ด๊ฑด ์ข์ง ์์, ๋ค์ ํด๋ดโ๋ผ๊ณ ํ๋จํ ์ ์๋ ๊ฒฝํ๊ณผ ํต์ฐฐ๋ ฅ์ ๊ฐ์ง๊ณ ์์์ต๋๋ค.
ํ์ง๋ง ๊ทธ๋ค์ ๋ค์ ์ธ๋๋ฅผ ๊ฑฑ์ ํฉ๋๋ค. โ5๋ , 10๋ ํ์ ์ฐจ์ธ๋ ์ํํธ์จ์ด ๊ฐ๋ฐ์๋ค์ ๊ณผ์ฐ ์ง์ ์ผ๋ก ํ๋ฅญํ ์์ง๋์ด๊ฐ ๋๊ธฐ ์ํ ๊น์ ์ฝ๋ ๊ฐ๊ฐ(code sense)์ ๊ฐ์ถ์ง ๋ชปํ๊ฒ ๋ ๊น?โ๋ผ๋ ์ง๋ฌธ์ ๋์ง๋๋ค. ๋ํ, ์์ ๋ค์กฐ์ฐจ ์ฝ๋๋ฅผ ์ง์ ์์ฑํ๋ ๋น๋๊ฐ ์ค์ด๋ค๋ฉด์ ๊ทธ โ์ฝ๋ ๊ฐ๊ฐโ์ ์์ด๋ฒ๋ฆด๊น ๋ด ์ฐ๋ คํฉ๋๋ค. ์ฌ์ฉํ์ง ์์ผ๋ฉด ์๊ฒ ๋๋ ๋ฒ์ด๋๊น์.
ํผ์(Piaan)์ด๋ผ๋ ์ ์ ๊ฐ๋ฐ์์ ์ฌ๋ก๋ ์ด๋ฌํ ์ฐ๋ ค๋ฅผ ์ ๋ณด์ฌ์ค๋๋ค. ๊ทธ๋ ๋ ์ด์ฐฝ๊ธฐ ์ง์ฅ์์ ๋ง์ดํฌ๋ก์ํํธ์ โ์ฝํ์ผ๋ฟ(Co-pilot)โ์ ์ฌ์ฉํด ์ฝ๋๋ฅผ ์์ฑํ๋๋ก ์๊ตฌ๋ฐ์๊ณ , ์๊ฐ์ ๋์ ํ๋ฃจ์ ์๋ฐฑ ๋ฒ์ ํ๋กฌํํธ(prompt)๋ฅผ ์ฌ์ฉํ์ต๋๋ค. ๊ทธ ๊ฒฐ๊ณผ ๊ทธ๋ ๋ โ์์ ์ ์ฝ๋ ์ง์์ด ์ ํ๋๊ณ ์ฝ๋ฉ ๋ฅ๋ ฅ์ ์์ด๊ฐ๋ ๋๋โ์ ๋ฐ์๋ค๊ณ ํธ์ด๋์์ต๋๋ค.
๊ทธ๋ ๋ค๋ฉด AI๊ฐ ์ฝ๋ฉ์ ๋์ ํ ์ ์๋ค๋ฉด, ๊ฐ๋ฐ์๋ค์ด ์ฝ๋ฉ ๊ธฐ์ ์ ์๋ ๊ฒ์ด ์ ๋ฌธ์ ์ผ๊น์? ์ฐ๋ฆฌ๊ฐ ์ด์ ์์๊ณต(type setter)์ด๋ ์์(calligraphy) ๊ธฐ์ ์ ๋ฐฐ์ฐ์ง ์๋ ๊ฒ์ฒ๋ผ, ์ฝ๋ฉ๋ ์ฌํ๊ฐ โ๋์ด์ โ ๊ธฐ์ ์ค ํ๋๊ฐ ๋ ์ ์์ง ์์๊น์?
์ด์ ๋ํด ์ํํธ์จ์ด ์ ๊ณ์์๋ ๋ ๊ฐ์ง ์๋ฐ๋ ์๊ฒฌ์ด ํฝํฝํ๊ฒ ๋ง์๊ณ ์์ต๋๋ค.
- ๋๊ด๋ก (๋ค์): AI๊ฐ ์ถฉ๋ถํ ์ข๊ณ ๊ณ์ํด์ ๋ ์ข์์ง ๊ฒ์ด๊ธฐ ๋๋ฌธ์, ์ธ๊ฐ๋ณด๋ค ์ฝ๋ฉ์ ๋ ์ํ ๊ฒ์ด๋ผ๊ณ ๋ฏฟ๋ ๊ฐ๋ฐ์๋ค์ด ๋ง์ต๋๋ค. ๊ทธ๋ค์ ์ฝ๋ฉ์ด ๋ง๋ฒ ๊ฐ์ ๋ณด์ด์ง๋ง, ์ค์ ๋ก๋ ๋ฐ๋ณต์ ์ด๊ณ ์ง๋ฃจํ ์์ ์ด ๋ง์ผ๋ฉฐ, ์ธ๊ฐ์ ์ค์๋ฅผ ํ์ง๋ง ๋ก๋ด์ ๊ทธ๋ ์ง ์๋ค๊ณ ์ฃผ์ฅํฉ๋๋ค. AI ์์ด์ ํธ๊ฐ ์ฝ๋ฉ์ ์ ๋ดํ๋ฉด ์ํํธ์จ์ด์ ์ ๋ขฐ์ฑ์ด ๋ ๋์์ง ๊ฒ์ด๋ผ๋ ๋ ผ๋ฆฌ์ ๋๋ค.
- ํ์๋ก (์์): ์ด๋ค์ ์๋ก์ด ์ฝ๋๋ฅผ ์์ฑํ๋ ๊ฒ์ด ๋จ์ํ ๋ง๋ค๊ณ ๋๋๋ ์ผ์ด ์๋๋ผ๊ณ ๋ฐ๋ฐํฉ๋๋ค. ์ฝ๋๋ ์ ์ง๋ณด์(maintain)ํด์ผ ํ๊ณ , ์๋ก์ด ๊ธฐ๋ฅ์ด ์ถ๊ฐ๋๊ฑฐ๋ ๋ค๋ฅธ ์ฝ๋์ ์ํธ์์ฉํ๋ฉด์ ์์์น ๋ชปํ ์ค๋ฅ๋ ๋ฌธ์ ๊ฐ ๋ฐ์ํ ์ ์์ต๋๋ค. AI๊ฐ ์์ฑํ ์ฝ๋๋ ์ง๊ธ์ ์ข์ ๋ณด์ผ์ง๋ผ๋, ์ฅ๊ธฐ์ ์ผ๋ก๋ ์ฝ๋๋ฒ ์ด์ค(codebase)์ ๋ค๋ฅธ ๋ถ๋ถ๊ณผ ๋์ ์ํธ์์ฉ์ ์ผ์ผํค๊ฑฐ๋ ๋ฏธ๋ฌํ ๋ฒ๊ทธ(bug)๋ค์ด ์์ฌ 5๋ ํ์๋ ๊ฑฐ๋ํ ํผ๋์ ์ด๋ํ ์ ์๋ค๊ณ ๊ฒฝ๊ณ ํฉ๋๋ค. ์ด๋ค์ ํ์ฌ์ ๊ฐ์ ๋ฐฉ์์ผ๋ก AI ์ฝ๋ฉ ๋๊ตฌ๋ฅผ ๋๊ท๋ชจ๋ก ์ฌ์ฉํ๋ ๊ฒ์ ๋ฐ๋ํฉ๋๋ค.
์ฃผ๋์ด ์ฝ๋์ ๋ฏธ๋: ์ฌ๋ผ์ง๋ ์ง์ ๊ตฐ, ์๋๋ฉด ์๋ก์ด ๊ธฐํ?
์ผ์๋ฆฌ ์์ค์ ๋ ๋ค๋ฅธ ์ค์ํ ์ฐ๋ ค ์ฌํญ์ ๋๋ค. AI๊ฐ ์ฝ๋ฉ ์์ ์ ๋์ ํ๋ฉด์ ์ฃผ๋์ด ์ฝ๋(junior coder)๋ค์ ์์๊ฐ ์ค์ด๋ค๊ณ , ์ด์ฉ๋ฉด โ์ฃผ๋์ด ์ฝ๋โ๋ผ๋ ์ง์ ๊ตฐ ์์ฒด๊ฐ ์์๊ณต์ฒ๋ผ ์ญ์ฌ ์์ผ๋ก ์ฌ๋ผ์ง ์ํ์ ์์๊น์?
ํด๋ผ์ด๋ธ ํฐ์จ ๊ธฐ์๋ ์ ์ ์ฑ์ฉ ์์๊ฐ ํ์คํ ์ฝํ๋ ์ํ์ด ์๋ค๊ณ ๋งํฉ๋๋ค. ์คํ ํผ๋ ๋ํ์ ์๋ฆญ ๋ฒค์กธ์จ(Erik Brynjolfsson) ๊ต์์ ์ฐ๊ตฌ์ ๋ฐ๋ฅด๋ฉด, ์ํํธ์จ์ด ๊ฐ๋ฐ์ ์ฑ์ฉ ๊ณต๊ณ ๋ ์ง๋ 1๋ ์ฌ์ด์ ์ด๋ฏธ 16% ๊ฐ์ํ์ต๋๋ค. AI ์ฝ๋ฉ ๋๊ตฌ๊ฐ ์์ง โ๊ฑท๋โ ์์ค์ธ๋ฐ๋ ์ด๋ฌํ ๊ฐ์๊ฐ ๋ํ๋ฌ๋ค๋ฉด, AI๊ฐ โ์ ๋ ฅ ์ง์ฃผโํ ๋๋ ์ด๋ค ์ผ์ด ๋ฒ์ด์ง์ง ์์ธกํ๊ธฐ ์ด๋ ต์ต๋๋ค.
๋ ํฐ ๋ฌธ์ ๋ ์๋ณธ์ฃผ์์ ์์ฑ์ ๋๋ค. ๋ชจ๋ ๋๊ธฐ์ ๋ค์ ๋น์ฉ ์ ๊ฐ ๋ฐฉ์์ ๋์์์ด ๋ชจ์ํ๋ฉฐ, ๊ธฐ์ ๊ธฐ์ ์์ ๊ฐ์ฅ ํฐ ๋น์ฉ ์ค ํ๋๋ ๊ฐ๋ฐ์๋ค์ ์ฐ๋ด์ ๋๋ค. AI๋ก ์ธ๋ ฅ์ ์ผ๋ถ๋ฅผ ๋์ฒดํ ์ ์๋ค๋ ์์ด๋์ด๋ ๊ธฐ์ ์๊ฒ ๋งค์ฐ ๋งค๋ ฅ์ ์ ๋๋ค. ์ด๋ฌํ ํ์์ ํ์ฌ ๋ชจ๋ ํ์ดํธ์นผ๋ผ(white-collar) ์ง์ข ์์ ๋ํ๋๊ณ ์์ต๋๋ค. ๊ธฐ์ ๊ฒฝ์์ง์ AI๋ก ์ธ๋ ฅ์ ํด๊ณ ํ๊ฑฐ๋, ํด๊ณ ์ํ์ ํตํด ์ง์๋ค์ ์๋ฐํ๋ ค ํฉ๋๋ค. AI๋ก ์ธํด ์ง์ ์ ๊ฐ์น๊ฐ ๋จ์ด์ง๊ณ ๊ธฐ์ ์ด ์์ค๋๋ฉด, ๊ณ ์ฉ์ฃผ๋ ์ง์์ ๋ ์ฝ๊ฒ ํต์ ํ ์ ์๊ฒ ๋ฉ๋๋ค.
์ฐ๋ฆฌ ๋ชจ๋์ ๋ฏธ๋: ์๋ํ์ ์ญ์ค๊ณผ ์ธ๊ฐ์ ์ญํ
AI๊ฐ ์ ๋ฌด์ ๋ฏธ์น๋ ์ํฅ์ ์ต์ ์ ์ ์ฝ๋ฉ์ด ์๋ค๋ฉด, ์ฝ๋๊ฐ ์๋ ์ฐ๋ฆฌ ๋ชจ๋์๊ฒ๋ ์ด๋ค ์๋ฏธ์ผ๊น์? ๋ค๋ฅธ ๋ถ์ผ์ ํ์ดํธ์นผ๋ผ ๋๋ ๋ธ๋ฃจ์นผ๋ผ(blue-collar) ๋ ธ๋์๋ค๋ AI์ ์ํด ๋น์ทํ ๋ฐฉ์์ผ๋ก ๋์ฒด๋ ๊น์?
๊ฐ๋ฐ์๋ค์ด ํ์ฌ ๊ฒฝํํ๊ณ ์๋ ๊ฒ์ ๋ค์ ์ญ์ค์ ์ผ๋ก ๋ณด์ผ ์ ์์ต๋๋ค. ๊ทธ๋ค์ด ์๋ ๊ฐ ๊ฐ๋ฐํด ์จ ๋งค์ฐ ์ด๋ ค์ด ๊ธฐ์ ์ ๋ฅ๋ ฅ(hard technical skills)๋ค์ด ์ฌ์ค์ ์๋ํํ๊ธฐ ๊ฐ์ฅ ์ฌ์ด ๊ฒ๋ค๋ก ๋๋ฌ๋ฌ๊ธฐ ๋๋ฌธ์ ๋๋ค. ๋ฐ๋ฉด, ๋๋ฃ๋ ๊ณ ๊ฐ๊ณผ ๋ํํ๊ณ , ๋ฌด์์ ๋ง๋ค์ด์ผ ํ ์ง ์์๋ด๊ณ , ์ฐ์ ์์๋ฅผ ์ค์ ํ๊ณ , ์ ๋ต์ ์๋ฆฝํ๋ ๊ฒ๊ณผ ๊ฐ์ โ์ด๋ ค์ดโ ์ผ๋ค์ AI๊ฐ ์์ง ํ ์ ์๋ ์์ญ์ ๋๋ค. ์ด๋ ์ฌ์ ํ ์ง์ ํ ์ธ๊ฐ์ ๋ฅ๋ ฅ(truly human skills)์ผ๋ก ๋จ์ ์์ต๋๋ค.
ํฐ์จ ๊ธฐ์๋ ์ด๋ฌํ ํจํด์ด ๋ค๋ฅธ ํ์ดํธ์นผ๋ผ ์ ๋ฌด์์๋ ๋ํ๋ ์ ์๋ค๊ณ ์์ธกํฉ๋๋ค. 1980๋ ๋์๋ ์ฒด์ค(chess)๊ฐ ๋๋ฌด ์ด๋ ค์์ ์ปดํจํฐ๊ฐ ์ด๊ธธ ์ ์์ ๊ฒ์ด๋ผ๊ณ ์๊ฐํ์ง๋ง, ๋งํ๊ธฐ(speaking)๋ ์ฌ์์ ์ปดํจํฐ๊ฐ ์ฝ๊ฒ ํ ์ ์์ ๊ฒ์ด๋ผ๊ณ ์ฌ๊ฒผ์ต๋๋ค. ํ์ง๋ง ์ค์ ๋ก๋ ์ปดํจํฐ๋ 1990๋ ๋์ ์ฒด์ค๋ฅผ ์ ๋ณตํ๊ณ , ์ธ๊ฐ์ฒ๋ผ ๋งํ๊ธฐ๋ฅผ ๋ฐฐ์ฐ๋ ๋ฐ๋ 20๋ ์ด์์ด ๋ ๊ฑธ๋ ธ์ต๋๋ค. AI๋ฅผ ํตํด ์ฐ๋ฆฌ๋ โ์ด๊ฒ์ด ๋์ ํฐ ๊ธฐ์ ์ด๋คโ๋ผ๊ณ ์๊ฐํ๋ ๊ฒ์ด ์ค์ ๋ก๋ ์ฐ๋ฆฌ์ ํต์ฌ ๊ธฐ์ ์ด ์๋์๊ณ , ์ฐ๋ฆฌ์ ์ง์ ํ ๊ธฐ์ ์ ๋ค๋ฅธ ๊ณณ์ ์๋ค๋ ๊ฒ์ ๊นจ๋ซ๊ฒ ๋ ๊ฒ์ ๋๋ค.
AI ํ๋ช ์ ์๊ฐํ์ ์ฌํ์ ํ๊ธ๋ ฅ
๊ทธ๋ ๋ค๋ฉด ์ด๋ฌํ ๋ณํ๊ฐ ๋๊ท๋ชจ ์ผ์๋ฆฌ์ ์ํฅ์ ๋ฏธ ๋ฏธ์น๋ ๋ฐ ์ผ๋ง๋ ๊ฑธ๋ฆด๊น์? ํฐ์จ ๊ธฐ์๋ โ์ฐ๋ฆฌ๊ฐ ์๊ฐํ๋ ๊ฒ๋ณด๋ค ๋ ์ค๋ ๊ฑธ๋ฆด ๊ฒโ์ด๋ผ๊ณ ์์ธกํฉ๋๋ค. ์ปดํจํฐ์ ์ญ์ฌ๋ฅผ ํตํด ์ฐ๋ฆฌ๋ ์๋ก์ด ๊ธฐ์ ์ด ๊ธฐ์ ํ๊ฒฝ์ ์ํฅ์ ๋ฏธ์น๋ ๋ฐ ์์๋ณด๋ค ํจ์ฌ ์ค๋ ์๊ฐ์ด ๊ฑธ๋ฆฐ๋ค๋ ๊ฒ์ ๋ฐฐ์ ๊ธฐ ๋๋ฌธ์ ๋๋ค.
1980๋ ๋์ 90๋ ๋ ์ด ๊ฐ์ธ์ฉ ์ปดํจํฐ(PC)๊ฐ ๋ฑ์ฅํ์ ๋, ๊ธฐ์ ์์ฐ์ฑ์ ์ฆ๊ฐ์ ์ผ๋ก ํฅ์๋์ง ์์์ต๋๋ค. ๊ฒฝ์ ํ์๋ค์ ๋นํฉํ์ต๋๋ค. ๊ทธ ์ด์ ๋ ๊ธฐ์ ์ด ์ค์ ๋ก ์์ฐ์ฑ์ด๋ ํจ์จ์ฑ์ ๋์ด๋ ค๋ฉด, ์ปดํจํฐ๋ฅผ ์ค์ฌ์ผ๋ก ๋น์ฆ๋์ค ๋ฐฉ์์ ์ฌํธํด์ผ ํ๊ธฐ ๋๋ฌธ์ ๋๋ค. ๋จ์ํ ํ์๊ธฐ๋ก ๋ฉ๋ชจ๋ฅผ ์น๋ ๊ฒ์ ์ปดํจํฐ๋ก ์ฎ๊ฒจ ์ ๋ ๊ฒ์ ๋์ด, โ๋ฉ๋ชจ๋ฅผ ์ธ์ํด์ ๋ณด๋ด์ง ๋ง๊ณ ์ด๋ฉ์ผ๋ก ๋ณด๋ด๋ผ. ๊ทธ๋ฌ๋ฉด ๋ชจ๋ ์ง์ญ ์ฌ๋ฌด์์์ ๋์์ ๊ฐ์ ์ ๋ณด๋ฅผ ์ฝ์ ์ ์๋คโ์ ๊ฐ์ด ์ ๋ณด์ ์์ฌ๊ฒฐ์ ์ ํ๋ฆ์ ์ปดํจํฐ์ ์ธํฐ๋ท์ ํน์ฑ์ ๋ง์ถฐ ์ฌ์กฐ์งํ๊ธฐ ์์ํ์ ๋น๋ก์ ํจ์จ์ฑ, ์์ฐ์ฑ, GDP์ ๋ณํ๊ฐ ๋ํ๋ฌ์ต๋๋ค. ํ์ง๋ง ์ด ๊ณผ์ ์ ๋งค์ฐ ์ค๋ ์๊ฐ์ด ๊ฑธ๋ ธ์ต๋๋ค. ํฐ์จ ๊ธฐ์๋ AI๋ ๋ง์ฐฌ๊ฐ์ง์ผ ๊ฒ์ด๋ผ๊ณ ์์ํฉ๋๋ค.
์ต๊ทผ์ ๊ฐ์ํ๋ ํ์ ์๋์ ๋ํด ํฐ์จ ๊ธฐ์๋ โ๋ฌธํ์ ์์ค์์๋ ์์ฒญ๋ ๊ฐ์ํ๊ฐ ์์ง๋ง, ์ฐ์ ์ ์ํฅ์ ์์ง ํฌ์ง ์๋คโ๊ณ ๋งํฉ๋๋ค. AI๋ก ์ธ๋ ฅ์ ํด๊ณ ํ๋ ํ์ฌ๋ค์ด 6๊ฐ์ ํ ๋ค์ ์ธ๋ ฅ์ ์ฌ๊ณ ์ฉํ๋ ๊ฒฝ์ฐ๊ฐ ๋ง๋ค๋ ๊ฒ์ด ๊ทธ ์ฆ๊ฑฐ์ ๋๋ค. ์๊ท๋ชจ ์คํํธ์ ์ ๋น ๋ฅด๊ฒ ์์ง์ด์ง๋ง, ๊ฒฝ์ ๋ฅผ ์์ง์ด๋ ๋๊ธฐ์ ์ ๊ฒฝ์ฐ ์ฌ์ ํ ๋๋ฆฐ ์๋๋ก ๋ณํ๋ฅผ ์์ฉํ๊ณ ์์ต๋๋ค. ๊ตฌ๊ธ์ 10% ์๋ ํฅ์์ด ํ์ฌ ๋๊ท๋ชจ ํ์ดํธ์นผ๋ผ ์ ๋ฌด์ AI๊ฐ ๋ฏธ์น๋ ์ํฅ์ ๋ ๊ฐ๊น์ธ ๊ฒ์ด๋ผ๋ ์ค๋ช ์ ๋๋ค.
์ํํธ์จ์ด์ ๋ณดํธํ๊ฐ ๊ฐ์ ธ์ฌ โ์ด์ํโ ๋ฏธ๋
AI ํ๋ช ์ด ์ฐ๋ฆฌ์๊ฒ ๊ฐ์ ธ์ฌ ๊ถ๊ทน์ ์ธ ์ด์ ์ ๋ฌด์์ผ๊น์? ํด๋ผ์ด๋ธ ํฐ์จ ๊ธฐ์๋ ์ํํธ์จ์ด๊ฐ ๋ ์ด์ โ๊ทํ๊ณ ํฌ๊ทํโ ์กด์ฌ๊ฐ ์๋๋ผ, ๋ง์น ์ข ์ด์ฒ๋ผ ๋ณดํธํ๋ ๊ฒ์ด๋ผ๋ ํฅ๋ฏธ๋ก์ด ๋น์ ๋ฅผ ๋ค์์ต๋๋ค. ๊ณผ๊ฑฐ์๋ ์ข ์ด๊ฐ ๋งค์ฐ ๊ทํด์ ์ผ๋ฐ์ธ์ด 1๋ ์ 4์ฅ ์ ๋๋ฐ์ ์ ํ ์ ์์์ง๋ง, ๊ฐ์๊ธฐ ์ ๋ ดํด์ง๊ณ ์ด๋์๋ ์กด์ฌํ๊ฒ ๋๋ฉด์ ํฌ์คํธ์(Post-it notes) ๊ฐ์ ๊ธฐ๋ฐํ ํํ์ ์ข ์ด๊ฐ ๋ฑ์ฅํ์ฌ ์ฐ๋ฆฌ์ ์ถ์ ์์ ํ ๋ฐ๊พธ์ด ๋์ ๊ฒ์ฒ๋ผ ๋ง์ ๋๋ค.
ํ์ฌ ์ธ์์๋ ์๋ง์ ์ํํธ์จ์ด๊ฐ ์กด์ฌํ์ง๋ง, ์ ๋ฌด์ ์ธ๊ณ๋ฅผ ๋ณด๋ฉด ์ฌ์ ํ ์ํํธ์จ์ด์ ๋์์ ์ ํ ๋ฐ์ง ๋ชปํ๋ ๋ฐฉ๋ํ ์์ญ์ด ์์ต๋๋ค. ์ฐ๊ฐ 5์ฒ๋ง ๋ฌ๋ฌ ๊ท๋ชจ์ ์ฝํฌ๋ฆฌํธ ํผํฉ ํ์ฌ์ ๊ฐ์ ์ค์๊ธฐ์ ๋ค์ ๋ ๋์ ์ํํธ์จ์ด๋ฅผ ์ํ์ง๋ง, ์ปค์คํ ์ํํธ์จ์ด๋ฅผ ๋ง๋ค 5๋ช ์ ๊ฐ๋ฐ์๋ฅผ ๊ณ ์ฉํ ์ฌ๋ ฅ์ด ์์ด ์๋์ฐ XP ์ปดํจํฐ์ ์์ ์คํ๋ ๋์ํธ 3๊ฐ๋ก ํ์ฌ๋ฅผ ์ด์ํ๋ ๊ฒฝ์ฐ๋ ํํฉ๋๋ค. ์ด๋ฌํ ๋๋๋๋ก ๋ง์ ์ค๊ฒฌ ๊ธฐ์ ๋ค์ด ๊ธฐ์ ๋ก๋ถํฐ ์ ๋๋ก ๋ ์๋น์ค๋ฅผ ๋ฐ์ง ๋ชปํ๊ณ ์์ต๋๋ค.
๋ง์ฝ ์ํํธ์จ์ด ์์ฑ ๋น์ฉ์ด ์ ๋ ดํด์ง๋ค๋ฉด ์ด๋จ๊น์? ํ ๋ช ์ ๊ฐ๋ฐ์๊ฐ 6๋ง~7๋ง ๋ฌ๋ฌ๋ฅผ ๋ฐ๊ณ ์ด๋ฌํ ํ์ฌ๋ฅผ ์ํ ๋ง์ถคํ ์ํํธ์จ์ด๋ฅผ ๊ตฌ์ถํ๊ณ ์ ์ง๋ณด์ํ ์ ์๊ฒ ๋๋ค๋ฉด, ๋ง์ ์ฌ๋๋ค์ ์ผ์ ์ ๋ฌด์์ ์ค์ง์ ์ธ ๊ฐ์ ์ด ์ด๋ฃจ์ด์ง ์ ์์ ๊ฒ์ ๋๋ค.
๊ถ๊ทน์ ์ผ๋ก ์ํํธ์จ์ด๋ ๋ ์ด์ ํน๋ณํ๊ณ ์ด๋ ค์ด ๊ฒ์ด ์๋๋ผ, ํฌ์คํธ์์ฒ๋ผ ์ด๋์๋ ์กด์ฌํ๊ฒ ๋ ๊ฒ์ ๋๋ค. ๋จ๊ธฐ์ ์ธ ํ์์ ์ํด ์ํ๋๊ณ , ์ฐ๋ฆฌ๊ฐ ์ํตํ๊ณ ๋ค๋ฅธ ์ฌ๋๋ค๊ณผ ๊ด๊ณ