Jevons Paradox of Cognition
Tegmark applies William Stanley Jevons's 1865 paradox—that coal efficiency improvements increased rather than decreased coal consumption—directly to AI. The conventional response to declining factor costs is increased demand: cheap electricity did not produce less electricity but electrification of everything; cheap computation did not produce less computing but computation embedded in every device. Cognition follows the same pattern. AI did not reduce work; it intensified work, as the 2025 Berkeley research documented empirically. Workers who adopted AI tools worked faster, took on more tasks, expanded into domains that had previously belonged to other specialists. Freed time was not reclaimed for rest; it was consumed by additional cognitive activity made possible by the reduction in per-unit cost. Cheaper cognition creates cognitive demand that was previously uneconomical to satisfy.
In the AI Story
The paradox has a structural feature that distinguishes its cognitive application from every previous instance. Coal, electricity, computation—each factor, however transformative, operated within a regime where human cognition remained the apex information-processing system. The steam engine augmented human muscle but left human judgment unchallenged. Each previous automation displaced labor at one level while creating demand at a higher level. The pattern held for two centuries and became the consistency that optimists cite as proof displacement is temporary and adaptation inevitable.
Tegmark identifies why this pattern may not hold for AI. Previous automation was narrow—performing specific cognitive tasks while leaving the broader cognitive landscape untouched. AI is not narrow. It climbs the cognitive hierarchy itself, and the speed of ascent is accelerating. Systems that could only perform routine cognitive tasks in 2020 were performing non-routine tasks by 2024; by 2026 they were performing work previously considered the exclusive province of experienced professionals.
The result is the funnel effect—each wave of AI capability narrows the funnel of tasks requiring uniquely human contribution. The critical question separating Tegmark's analysis from both optimists and pessimists is whether the funnel narrows to a stable equilibrium (ascending friction continues creating human work at higher levels) or to a point (AI eventually exceeds human capability across every cognitive dimension). The physics permits both; what the physics does not permit is complacency.
The answer matters because 'economically necessary' and 'valuable' are not the same currencies. Human consciousness, human experience, human creativity may be intrinsically valuable regardless of economic productivity. But intrinsic value and market value diverge, and a civilization organized around market signals for centuries may struggle to maintain structures that protect intrinsic value when market signals point elsewhere.
Origin
William Stanley Jevons articulated the coal paradox in The Coal Question (1865). Tegmark's extension to cognition appears across his writings and public talks, drawing on the 2025 Berkeley research by Ye and Ranganathan that empirically documented the intensification pattern in AI-augmented workplaces.
Key Ideas
Efficiency creates demand. Declining factor costs produce more use, not less—documented for coal, electricity, computation, and now cognition.
Cognitive intensification. AI makes previously uneconomical cognitive work profitable, producing more work rather than less.
Funnel effect. Each capability wave narrows the set of tasks requiring uniquely human contribution.
Novel distinction. Previous automation was narrow; AI climbs the cognitive hierarchy itself.
Equilibrium vs. point question. Whether the funnel stabilizes or closes determines everything.
Appears in the Orange Pill Cycle
Further reading
- William Stanley Jevons, The Coal Question (1865)
- Xingqi Maggie Ye and Aruna Ranganathan, 'AI Doesn't Reduce Work—It Intensifies It' (HBR, 2026)
- Max Tegmark, Life 3.0 (2017)
- Daron Acemoglu and Simon Johnson, Power and Progress (2023)
- Erik Brynjolfsson, 'The Turing Trap' (Daedalus, 2022)