AI Won’t Make Our Minds Useless According to 400 Years of History

Context

This is a solo NLW episode built around Venkatesh Rao's essay "Cognitive Mercantilism," with NLW adding his own product and labor-market interpretation. The central question is whether abundant AI makes human minds less relevant, or whether it expands what people want, build, and work toward. The historical analogy matters for PMs because it changes how to think about market size, user behavior, and AI strategy. If intelligence gets cheaper the way food production once did, the biggest product opportunities may come from newly created demand, not just lower operating costs.

The Big Idea

AI may matter less as a job-substitution machine and more as a demand-expansion machine.

When intelligence becomes cheaper and easier to access, people do not necessarily become useless. They often become more ambitious about what they want to do.

Venkatesh Rao's historical claim is that abundance often changes human desires, not just production economics. NLW's product implication is that PMs should stop treating today's level of demand as fixed, because AI can change what users find worth attempting in the first place.

Key Insights

Scarcity stories misread AI

Venkatesh Rao's core critique is that many AI debates still treat intelligence like a scarce resource to hoard, protect, and ration.

• He calls this cognitive mercantilism.
• In that worldview, intelligence is something embodied in elite brains, expert institutions, and protected knowledge products.
• That leads leaders to focus on talent hoarding, intellectual property control, and limiting who gets access.
• For PMs, this mindset pushes strategy toward cost control and access restriction, while blinding teams to demand creation and market expansion.

Demand grows with abundance

The essay borrows from the history of capitalism to make a demand-side point. Pre-capitalist mercantilist thinking assumed wealth was finite, but capitalism emerged when societies began to see that wealth could be created and that ordinary people could become sources of demand, not just labor.

That matters for AI because many product plans still assume the size of demand is fixed. If intelligence becomes far cheaper, easier, and more available, the relevant question is not only how efficiently work gets done. The deeper question is what new forms of consumption, curiosity, and activity become worth doing at all.

Cheap intelligence changes desire

Venkatesh Rao's strongest claim is that abundant intelligence does not just produce more answers, reports, or software. It changes what people can want.

• Previously unreachable projects become feasible.
• Previously unaffordable curiosities become worth pursuing.
• Previously specialized domains become explorable by ordinary users.

For PMs, this is a major shift in product framing. The best AI opportunities may not be "same workflow, lower cost." They may be "new workflow, new user, new reason to engage."

AI can raise ambition

The essay uses a historical example to explain why AI may make people work harder, not less.

• Joyce Appleby's account describes annual working days in Britain rising from about 250 to 300 before the Industrial Revolution.
• The point was not simple coercion. New desirable goods gave people a reason to work more voluntarily.
• Venkatesh Rao argues AI could create an analogous effect for cognitive effort. When the payoff from attention rises, people may choose to think harder and attempt more.

NLW adds a complementary lens. He argues that AI and agents make more of the "infinite backlog" feel actionable, which creates both anxiety and excitement. For PMs, this suggests that user engagement with AI may show up as expanded ambition, not just time savings.

Access may beat job protection

The episode suggests that AI-era politics may center less on preserving every current cognitive role and more on securing access to intelligence and the ability to use it.

In the same way industrial societies did not ultimately organize around protecting farm labor forever, an AI-rich society may organize around who gets a fair share of cognitive abundance. For PMs, this means products that broaden access to high-quality intelligence could matter more than products designed mainly to defend expert bottlenecks.

Transition pain still matters

NLW is explicit that this optimism is not a dismissal of AI safety or transition costs. He does not argue that abundance guarantees a painless shift.

The important distinction is between:

• trying to manage the transition, and
• trying to prevent the transition entirely.

For PMs, that means a serious AI strategy needs two layers at once. It needs a long-run view of new opportunity and a short-run plan for trust, adoption, training, and disruption.

Mental Models & Frameworks

Cognitive mercantilism

This is Venkatesh Rao's name for the scarcity mindset behind much AI anxiety. It maps old mercantilist beliefs about wealth onto modern beliefs about intelligence.

Its logic has four parts:

• Intelligence is treated as finite and embodied in elite producers.
• Ordinary people are seen as untrustworthy users of intelligence.
• Strategy focuses on controlling the supply of intelligence and keeping it valuable.
• Most people are assumed to have little intellectual upside beyond routine cognitive labor.

Use this model when a team is over-fixated on protecting expertise, restricting access, or treating AI mainly as a threat to status hierarchies.

Cognitive industrious revolution

Venkatesh Rao proposes that cheaper intelligence could trigger a new wave of voluntary cognitive effort, much like new goods once motivated more voluntary labor.

Use this model when you want to explain why AI might increase effort instead of reducing it. The idea is simple: when a small amount of attention can now unlock much larger results, people have a stronger reason to invest that attention.

For PMs, this is a useful lens for products that help users explore, build, or learn. The value may come from raising the return on attention, not from removing attention altogether.

AI Malthusianism

This is the opposing fear that intelligence supply will grow faster than humans can absorb, leaving large numbers of people economically redundant.

The model is useful because it clarifies what many AI doom stories assume:

• demand for intelligence is limited,
• human adaptation is weak, and
• surplus capability necessarily becomes socially destabilizing.

You do not have to reject this concern to use the framework. For PMs, it is a reminder to test whether a prediction of collapse is really a prediction about fixed demand.

Six demand elasticities

NLW brings in his own framework for how AI can unlock demand. He argues that people often want more than they currently consume, but something blocks that demand.

• Price elasticity: people wanted the outcome, but it cost too much.
• Access elasticity: people wanted the outcome, but could not get it easily.
• Complexity elasticity: people needed the outcome, but the system was too confusing.
• Continuity elasticity: people got occasional help, but would benefit from help all the time.
• Personalization elasticity: people got the generic version, but wanted something tailored.
• Relational elasticity: people wanted the experience to feel more human, meaningful, or trusted.

Use this framework when sizing an AI opportunity. It helps shift discovery from "what can the model do?" to "what friction is suppressing real demand?"

Infinite backlog effect

NLW's "infinite backlog" idea explains why AI can feel both thrilling and overwhelming.

Before AI, people naturally limited what they considered doable because their own time and energy were scarce. With agents and parallelized cognitive labor, far more of the backlog feels actionable. That can create pressure because "done" stops feeling stable, but it can also create excitement because many previously deferred ideas now seem buildable.

Use this model when evaluating internal AI tools. If a team feels busier after automation, that does not automatically mean the tool failed. It may mean the perceived opportunity frontier moved outward.

Decision Principles

Principle: Start with demand

When: a team frames an AI initiative mainly as cost reduction or labor substitution. Why: this episode's central argument is that lower cognitive costs can expand what users want, not just lower the price of existing work.

Principle: Question zero-sum assumptions

When: someone predicts that AI will eliminate value because it automates a current task. Why: that claim often hides an untested assumption that the amount of useful work, curiosity, or consumption is fixed.

Principle: Measure ambition too

When: evaluating whether an AI feature worked. Why: time saved matters, but this episode suggests an equally important signal is whether users attempt more projects, explore more ideas, or engage more deeply because the return on effort improved.

Trade-offs & Nuance

Abundance versus overload

More available intelligence can create more opportunity, but it can also create more pressure.

• The upside is that users can attempt more ambitious work.
• The downside is that the sense of a bounded workload can disappear.
• For PMs, this means AI products may need pacing, prioritization, and workflow controls, not just more raw capability.

Openness versus misuse

The essay leans toward democratizing intelligence, but it also names the fear behind elite resistance.

• Supporters of openness think broad access unlocks human potential.
• Critics worry that low-quality reasoning, misinformation, or reckless use will spread faster.
• For PMs, the practical answer is usually not pure restriction or pure openness. It is wider access paired with guardrails, transparency, and education.

Future gains, present pain

The episode argues that long-run abundance could be highly positive, but NLW does not pretend the transition will be easy.

Teams should expect tension between:

• the strategic case for pushing into AI-enabled opportunity, and
• the operational need to handle displacement, confusion, and trust issues in the near term.

Common Mistakes

Mistake: Treat intelligence as stock

A common failure mode is to think of intelligence as a fixed pool that organizations must capture and defend.

That framing overweights talent hoarding, proprietary control, and expert gatekeeping. It underweights platform effects, user enablement, and new demand. A better question is not "how do we own more intelligence?" but "what becomes possible when more people can command it?"

Mistake: Ignore latent demand

Many AI roadmaps assume today's product usage reflects all meaningful demand. This episode argues the opposite.

Users often do not ask for something because it is too expensive, too hard, too slow, or too specialized today. When AI removes those constraints, demand can appear suddenly. PMs who only extrapolate from current usage can miss the real upside.

Mistake: Equate tasks with jobs

The episode pushes back on the idea that automating a task tells you the fate of an entire role.

A job is a bundle of tasks, goals, rights, incentives, and evolving expectations. When AI changes one part of that bundle, people and institutions often reorganize around new work rather than disappearing outright. PMs should be careful not to model user behavior as if workflow change equals role extinction.

Practical Application

Map unlocked demand

For the next AI initiative, run a quick demand review before writing requirements.

• Ask where AI changes price, access, or complexity enough to unlock new usage.
• Ask where always-on support, personalization, or more human-feeling interaction could expand engagement.
• Write down the new behavior you expect to appear, not just the old task you expect to get faster.

Reframe AI product bets

Take one current roadmap item and rewrite it in two ways:

• Version one: "How does AI reduce time or headcount here?"
• Version two: "What new outcome could a non-expert now achieve here?"

If the second version is much more interesting, the opportunity is probably demand-side, not just efficiency-side.

Track ambition signals

Add metrics that show whether users are trying more ambitious things because AI lowered the barrier.

Examples include:

• first-time use of advanced workflows,
• number of projects started per user,
• breadth of use cases attempted,
• frequency of follow-on actions after an AI response.

These signals are often more revealing than a simple "minutes saved" survey.

Design for non-experts

If abundant intelligence matters most when more people can consume it without producing it themselves, then product design should reflect that.

• Reduce setup and jargon.
• Explain outputs clearly.
• Make correction easy.
• Build trust cues for users who are capable and motivated, but not already specialists.

Plan transition support

Pair every internal AI rollout with explicit change support.

• Define which tasks are changing.
• Define which higher-value work becomes newly possible.
• Train teams on how success will be measured after the shift.

This helps avoid the fear that AI is only a subtraction story.

Questions to Consider

• Where is your team assuming current user demand is the ceiling, even though cheaper intelligence could make entirely new use cases worth pursuing?
• Which AI initiatives in your roadmap mainly optimize the supply of intelligence, such as model quality or expert productivity, while ignoring who newly becomes a meaningful consumer of intelligence?
• If an AI feature removed most of the complexity from a hard workflow, what adjacent desires would that unlock for users: more frequency, more personalization, more depth, or entirely new goals?
• What metrics in your product would show rising user ambition, not just rising automation?

Bottom Line

This episode argues that the biggest effect of AI may be expanding what people want and attempt, not simply replacing the work they already do. For PMs, the strategic mistake is to treat demand as fixed. If intelligence gets dramatically cheaper, the most important product question becomes what new behaviors, markets, and ambitions that abundance unlocks.

Resources Mentioned

Concepts to Explore

Lump of labor fallacy

This is the belief that there is a fixed amount of work to be done, so if machines do more, humans must do less. The episode challenges that assumption directly. PMs should understand it because many bad AI product debates quietly depend on it.

Demand-side innovation

Demand-side innovation asks how a technology changes what customers want and what becomes economically viable, not just how efficiently something gets produced. That is the most important product lens in this episode, and it is a useful counterweight to feature lists focused only on automation.

People to Follow

Venkatesh Rao

Venkatesh Rao is the writer of "Cognitive Mercantilism" and the main intellectual source for this episode. He is worth following if you want original frameworks that connect technology, economics, and social change.

Joyce Appleby

Joyce Appleby is the historian whose book anchors the episode's economic analogy. Her work is relevant here because it explains how changing production conditions can reshape demand, labor, and social assumptions over long periods.