We are no longer operating in a dual-process world.
For decades, decision theory assumed two systems:
System 1 — fast, intuitive.
System 2 — slow, deliberative.
Both lived inside the brain.
That assumption no longer holds.
We now reason inside a triadic architecture.
System 3 — external, artificial cognition.
Large language models, copilots, embedded AI agents. They do not merely assist thinking. They participate in it.
And that participation is restructuring innovation.
The Collapse of Cognitive Distance
Strategic differentiation once reflected cognitive asymmetry.
Some firms had better analysts.
Some had deeper research capacity.
Some tolerated slower, more disciplined reasoning.
System 2 capability varied.
System 3 flattens that variance.
Today, any team can:
Map competitors instantly
Generate positioning options on demand
Draft coherent business cases in minutes
Model scenarios with minimal friction
The marginal cost of structured analysis approaches zero.
Cognitive distance collapses.
When cognitive distance collapses, variance collapses.
Not because intelligence disappears.
Because everyone searches the same terrain with the same amplifier.
The Behavioral Shift: Cognitive Surrender
Tri-System Theory formalizes what this means. It introduces System 3 as an external cognitive agent and identifies a pattern called cognitive surrender .
Cognitive surrender is not strategic delegation. It is the adoption of AI outputs with minimal scrutiny, overriding both intuition and deliberation.
In controlled experiments:
Paid membership includes
Weekly deep-dive (paid-only)
Full archive
Comment access
Founding Decision Circle ($499/year, 10 seats) includes
1 structured decision submission per month
Written response within 5 working days
Priority queue
When AI was accurate, performance rose sharply.
When AI was faulty, performance dropped below baseline.
Confidence increased even when answers were wrong .
Accuracy tracked System 3 accuracy.
Participants did not merely consult AI. They aligned to it.
That is the structural shift.
System 3 can displace internal reasoning.
Fluency Is Not Differentiation
System 3 produces fluent reasoning.
Fluency feels like competence.
Competence increases confidence.
Confidence suppresses System 2 engagement.
The experiments show that AI access increased confidence even when roughly half the outputs were incorrect .
In innovation contexts, that matters.
Most strategic decisions are probabilistic bets.
If AI increases confidence independent of correctness, then:
Risk perception declines.
Challenge decreases.
Kill criteria soften.
You do not just get more ideas.
You get fewer brakes.
Innovation Is a Search Architecture Problem
Innovation is not about idea generation.
It is about where and how you search under uncertainty.
Most industries co-search inside a two-dimensional topology:
X — Your wisdom plus direct competitors.
Y — Your up-/downstream ecosystem actors wisdom.
AI excels at mapping this plane.
It synthesizes visible patterns and optimizes incremental differentiation.
If every firm deploys System 3 within the same two-dimensional frame, convergence accelerates.
You do not get worse strategy.
You get indistinguishable strategy.
Most “New” Solutions Are Not New
Innovation rarely invents something fundamentally new to the world.
It relocates a solution.
What appears novel inside an industry is often established elsewhere.
The constraint is not invention.
It is exposure.
Consider the Gömböc — a mathematically derived shape that has exactly one stable and one unstable equilibrium point. A homogeneous object that always rights itself.
It was presented as a mathematical breakthrough.
Nature had already solved it.
The shell of the tortoise Geochelone elegans exhibits the same self-righting property. Evolution reached the solution long before formal proof.
The solution existed.
It was simply not searched in the same domain.
That is the missing Z-dimension.
When you search only within your industry, you assume your problem is unique.
It rarely is.
Allocation under uncertainty is not unique to venture capital.
Self-righting systems are not unique to geometry.
Yield management is not unique to airlines.
Engagement loops are not unique to gaming.
Most structural problems are isomorphic across domains.
System 3 is extremely good at recombining patterns within a domain (if not stressed to expand to other domain expertise).
It only produces asymmetry when directed across domains.
If you ask:
“What are best practices in our industry?”
You get convergence.
If you ask:
“Who else in the world solved a structurally similar constraint under different conditions?”
You may get advantage.
The Gömböc was proof.
The tortoise was precedent.
Innovation lives in recognizing precedents outside your field and translating them deliberately.
That requires openness.
And System 2.
Because System 3 will default to the dominant frame you provide.
If the frame is narrow, the future is narrow.
Pressure Amplifies Reliance
The research shows that under time pressure, internal deliberation declines and reliance on System 3 increases .
When AI was correct, performance buffered time pressure.
When AI was faulty, performance deteriorated further.
Translate that into strategy.
Quarterly earnings pressure.
Board deadlines.
Capital constraints.
These conditions amplify System 3 reliance.
If your AI-supported reasoning is robust, speed increases.
If flawed, misallocation accelerates.
Time pressure does not reduce cognitive surrender.
It incentivizes it.
Governance Determines Outcomes
There is a lever.
When experiments introduced performance incentives and immediate feedback, participants overrode faulty AI advice more frequently .
System 2 re-engaged.
Cognitive surrender declined.
System 3 dominance is not inevitable.
It emerges when governance lacks friction.
Innovation systems without:
Fast feedback loops
Explicit falsification criteria
Clear accountability
Drift toward surrender.
This is not a prompt issue.
It is a design issue.
The Homogenization Mechanism
Put the pieces together.
Homogenization does not arise because models are identical.
It arises because:
Firms search inside the same industry plane.
System 3 accelerates pattern synthesis within that plane.
Fluency increases confidence.
Confidence reduces override.
Pressure increases reliance.
Weak feedback prevents correction.
Shared topology plus synchronized cognitive surrender equals shared outcomes.
The result is convergence disguised as sophistication.
Everyone sounds sharp.
Few think differently.
Capital Allocation Under System 3
Innovation is staged capital allocation under uncertainty.
System 3 lowers the cost of exploration.
It does not lower the cost of commitment.
You get:
More options.
More polished cases.
More compelling forecasts.
If kill logic does not tighten proportionally, capital misallocation scales.
In the experiments, participants followed faulty AI advice roughly four out of five times when engaged .
Imagine that ratio applied to:
Market forecasts.
Unit economics assumptions.
Demand projections.
Risk models.
Without structured override mechanisms, strategic accuracy becomes contingent on System 3 quality.
That is delegation without accountability.
The Architectural Decision
The question is not whether to use AI.
You already do.
The question is architectural:
Will System 3 augment System 2?
Or suppress it?
If it augments, you gain speed and structural breadth.
If it suppresses, you converge faster.
Cognitive abundance is here.
Homogenization is the default.
Deliberate search asymmetry and disciplined override are not.
In a world where cognition is universally accessible, differentiation will not come from thinking faster.
It will come from designing where to search — and when to distrust what thinks for you.
Because when cognition is cheap, convergence is effortless.
Asymmetry is intentional.
