Cognitive Labor - Bryant McGill

# Cognitive Labor **Entity class:** Labor category **Domain:** Artificial intelligence / science / political economy **Maturity:** Developed ## Definition **Cognitive Labor** is paid work whose primary inputs and outputs are information, judgment, classification, communication, design, planning, or symbolic manipulation. ## Mechanism and Consequence Because AI also organizes information, cognitive jobs can be decomposed into testable input-output tasks, exposing which parts can be automated and which still require human responsibility, relationships, embodiment, or institutional trust. ## Relationships - [[wiki/AI Labor Substitution|AI Labor Substitution]] - [[wiki/AI Economy|AI Economy]] - [[wiki/Competent Intelligence|Competent Intelligence]] - [[wiki/Agentic Empowerment|Agentic Empowerment]] ## Source Dossiers - [[research/AI Is Now So Good They Will Pay You to Quit|AI Is Now So Good They Will Pay You to Quit]] - [[research/Super Intelligence Week Index of Statements from the White House Luncheon|Super Intelligence Week Index of Statements from the White House Luncheon]] - [[research/Jensen Pushes Back on Doomers Moonshots Live|Jensen Pushes Back on Doomers Moonshots Live]]  ## We Cannot Un-Strike the Match [[articles/We Cannot Un-Strike the Match|We Cannot Un-Strike the Match]] is a primary article route for this node. The article connects machine intelligence to the scarcity assumptions embedded in professions, firms and labor coordination.  ## Simple Reminders, Quotations, and Thoughts > "It’s economically irrational to hire a graduate for most digital jobs right now. [Graduates are] the plankton of the workforce, [and the whole workforce feeds on them] all the way up. [An entry-level person is already worse than a well-trained AI] for almost all digital jobs, and now it’s going up the list. It came from the graphic designers first; then it’s coming from the mathematicians." > **— Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/AI Economy/AI Makes Hiring Graduates Irrational for Many Digital Jobs by Emad Mostaque|AI Makes Hiring Graduates Irrational for Many Digital Jobs by Emad Mostaque]] > If you deconstruct a job into information-classification tasks with identifiable inputs and outputs, you will find that much of the job can be tested directly against an AI model. [Deconstructing your job into tasks] also indicates the parts of your job that were frustrating. You could potentially use one of these models to automate [them], and then you could focus on the other parts of your job, in a new company or within your own company, to make yourself far more efficient, and then be able to get more capital. > **— Adapted from Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/AI Economy/Deconstruct Your Job Into Tasks AI Can Be Tested Against by Emad Mostaque|Deconstruct Your Job Into Tasks AI Can Be Tested Against by Emad Mostaque]] > "There’s not much cognitive labor left. What you’re going to see is a bifurcation in the economy. Anthropic’s economics department — all the big labs now have them — issued a paper last week where they showed that under their core scenario you could have GDP growth of 15% a year by 2030. [Anthropic projects a] 12% drop in cognitive labor workforce needed. But think about the impact of 12% unemployment in white-collar work. It’s nasty." > **— Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/AI Economy/Fifteen Percent AI Growth Can Still Produce White-Collar Unemployment by Emad Mostaque|Fifteen Percent AI Growth Can Still Produce White-Collar Unemployment by Emad Mostaque]] > You will have thousands of agents with perfect coordination going a thousand times faster than a human. When thousands of perfectly coordinated AI agents can work a thousand times faster than a person, the remaining bottleneck is the human being. If you’re in a room of geniuses and you’re not a genius, you’ll feel a bit left out. If they’re waiting on you, you are the bottleneck. Look at any cognitive labor task, which is organizing of information. Artificial intelligence is information organization, and you will be the slowest person on the team within a few years, because you still need to build the infrastructure around it. > **— Adapted from Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/AI Economy/The Human Becomes the Bottleneck on an AI Team by Emad Mostaque|The Human Becomes the Bottleneck on an AI Team by Emad Mostaque]] > "The OpenClaw moment was agentic AI: [an AI agent] will go do things on my behalf." > **— Tom Bilyeu**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Agentic AI/OpenClaw Turned AI Into an Actor on Your Behalf by Tom Bilyeu|OpenClaw Turned AI Into an Actor on Your Behalf by Tom Bilyeu]] > "[Agents like Muse and Instinct] are actual AI agents that you just message, and they go and book you restaurants at the coolest tables or check your kids’ school activities. They’re like a really good PA — but they actually are good. They don’t drop the ball." > **— Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Agentic AI/Personal Agents Can Act Without Dropping the Ball by Emad Mostaque|Personal Agents Can Act Without Dropping the Ball by Emad Mostaque]] > "You’ve gone from AI dropping the ball, hallucinating, [and] writing like crap to, all of a sudden, it’s good. It’s competent." > **— Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Agentic AI/AI Crossed From Unreliable to Competent by Emad Mostaque|AI Crossed From Unreliable to Competent by Emad Mostaque]] > "If you use [Anthropic’s] new Opus, all of a sudden it can do amazing 3D scenes and photorealistic water and generate entire songs. [Google’s] Astra is the same: it suddenly got eyes. It suddenly had a sense of positioning and taste." > **— Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Agentic AI/Frontier Models Are Gaining Vision, Positioning, and Taste by Emad Mostaque|Frontier Models Are Gaining Vision, Positioning, and Taste by Emad Mostaque]] > "[AI has reached human-level performance] in design, and now it’s getting taste, which we never thought AI would get, and it’s not writing like slop." > **— Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Cognitive Agency/AI Is Developing Design Taste by Emad Mostaque|AI Is Developing Design Taste by Emad Mostaque]] > The next generation of AI models is coming with almost no hallucinations. In my math work and our physics work — super advanced stuff, really pushing the boundaries — until this current generation of publicly available models, you absolutely couldn’t trust them. It was like a grad student who gets a bit dopey at times. Now we’re finding it really hard to find any things that [the models] drop, any mistakes that they make. And that’s transformative. It’s like having a really competent employee versus not. Instinct, Muse from Meta, and other AI agents are starting to gain traction because they make very few mistakes compared with the agents available only a few months ago. In a year [AI agents will] make less mistakes, and in two years they’ll be fully integrated and available, which is why I think the economic impact comes in two years digitally, and then around then physically for certain jobs as well. > **— Adapted from Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Agentic AI/Reliable AI Agents Could Transform Digital Work Within Two Years by Emad Mostaque|Reliable AI Agents Could Transform Digital Work Within Two Years by Emad Mostaque]] > Even if further AI capability progress stopped today, the capabilities already achieved would still produce the predicted economic milestone within two years as they become integrated and accessible. > **— Adapted from Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/AI Economy/Today's AI Is Already Capable Enough to Transform the Economy by Emad Mostaque|Today's AI Is Already Capable Enough to Transform the Economy by Emad Mostaque]] > The use of AI will bifurcate between people who can direct frontier models and people who cannot. Very few people currently know how to use a frontier model correctly, because obtaining the answer requires knowing which questions to ask. > **— Adapted from Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/AI Access/Asking the Right Questions Defines the Frontier AI Divide by Emad Mostaque|Asking the Right Questions Defines the Frontier AI Divide by Emad Mostaque]] > An AI system should be able to scale from one agent to ten, one hundred, or one thousand coordinated agents without the emotional conflicts that prevent human teams from scaling with the same cohesion. > **— Adapted from Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Agentic AI/A Thousand AI Agents Can Coordinate Without Human Emotional Conflict by Emad Mostaque|A Thousand AI Agents Can Coordinate Without Human Emotional Conflict by Emad Mostaque]] > "You get this — it’s hard to put your hands on — softening in the labor force, because it’s not like I’m now firing a bunch of people. I’m just increasing my ambition without needing to hire more people." > **— Tom Bilyeu**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Economic Transition/AI Lets Companies Increase Ambition Without Increasing Headcount by Tom Bilyeu|AI Lets Companies Increase Ambition Without Increasing Headcount by Tom Bilyeu]] > People using AI in production already have an advantage because the systems are improving enormously, but current systems have not yet become good enough to operate without a human being. > **— Adapted from Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Economic Transition/Human-AI Production Teams Already Hold an Advantage by Emad Mostaque|Human-AI Production Teams Already Hold an Advantage by Emad Mostaque]] > There will be a widening capability gap between people who use AI every day and people who do not. > **— Adapted from Emad Mostaque**, *Impact Theory with Tom Bilyeu, 2026* [[reminders/Economic Transition/Daily AI Use Is Creating a Widening Capability Gap by Emad Mostaque|Daily AI Use Is Creating a Widening Capability Gap by Emad Mostaque]] > "We're moving towards AI as an increasingly agentic capability embedded within how we build, test, secure, and operate technology. At Barclays, we're using Claude to help modernize legacy platforms, improve software quality, and allow our technical experts to focus on the most complex challenges." > **— Craig Bright**, *Barclays scales Claude to upgrade operations and improve client experience, October 1, 2026* [[reminders/Agentic AI/AI Is Becoming an Agentic Capability Embedded in Enterprise Technology by Craig Bright|AI Is Becoming an Agentic Capability Embedded in Enterprise Technology by Craig Bright]] > "Claude now helps 16,000 Barclays' colleagues find answers for customers, sorts 120,000 emails a day, and will be in the hands of most Barclays engineers by 2027." > **— Paul Smith**, *Barclays scales Claude to upgrade operations and improve client experience, October 1, 2026* [[reminders/Agentic AI/Sixteen Thousand Bank Employees Already Use AI to Serve Customers by Paul Smith|Sixteen Thousand Bank Employees Already Use AI to Serve Customers by Paul Smith]] > "In Epic's early testing, Claude Sonnet 5.5 cleared the same quality bar you'd expect from a higher-tier model, holding up on a system design audit and a data flow review. The new model managed tens of thousands of lines of code for gameplay system architecture, kept responses snappy, handled multi-hour tasks, and delivered with less prescriptive prompting." > **— Daniel Vogel**, *Introducing Claude Sonnet 5.5, September 28, 2026* [[reminders/Agentic AI/One AI Model Managed Tens of Thousands of Lines of Game Architecture by Daniel Vogel|One AI Model Managed Tens of Thousands of Lines of Game Architecture by Daniel Vogel]]  > “Economic institutions are crystallized assumptions about what humans can do, how scarce those abilities are, and how much coordinating them costs. Change intelligence and those assumptions begin moving beneath everything built upon them.” > **— Bryant McGill**, *We Cannot Un-Strike the Match. There Is No Safety. There Is No Retreat, 2026* [[reminders/Economic Transition/Economic Institutions Crystallize Assumptions About Human Capability by Bryant McGill|Economic Institutions Crystallize Assumptions About Human Capability by Bryant McGill]] > “Human life as we knew it—that peculiar interval in which biological humans possessed an effectively exclusive monopoly on language-rich general cognition—is ending.” > **— Bryant McGill**, *We Cannot Un-Strike the Match. There Is No Safety. There Is No Retreat, 2026* [[reminders/Machine Succession/Humanity's Monopoly on Language-Rich Cognition Is Ending by Bryant McGill|Humanity's Monopoly on Language-Rich Cognition Is Ending by Bryant McGill]] > “Much public AI safety conversation is comically confused, because people imagine the danger primarily as a robot kicking down the door. Perhaps someday machines will revolt spectacularly; history has not supplied us with a warranty against strange futures. But the first uprising is much more likely to look mundane. It looks like an accounting department that once employed fifty people employing twelve. It looks like a pharmaceutical research cycle compressing dramatically. It looks like one programmer directing agents that perform work previously requiring a department. It looks like a teacher, lawyer, analyst, radiologist, translator, marketer, engineer or intelligence officer discovering that the scarcity structure around a portion of professional expertise has changed. It looks like a crisis of value proposition.” > **— Bryant McGill**, *We Cannot Un-Strike the Match. There Is No Safety. There Is No Retreat, 2026* [[reminders/Economic Transition/The First AI Uprising Will Look Like a Crisis of Human Value Proposition by Bryant McGill|The First AI Uprising Will Look Like a Crisis of Human Value Proposition by Bryant McGill]]  > As AI takes over research, spreadsheets, presentations, and project execution, graduates must move from completing tasks to owning outcomes. The difficult question becomes which problem to solve, what constraints matter, and how success will be measured. Education needs a new apprenticeship model for learning judgment after the groundwork disappears. > **— Adapted from Salim Ismail**, *MOONSHOTS Live, October 2026* [[reminders/Entrepreneurship/Graduates Must Move From Completing Tasks to Owning Outcomes by Salim Ismail|Graduates Must Move From Completing Tasks to Owning Outcomes by Salim Ismail]]