--- title: I Don't Want to Be a System 3 Thinker theme: default layout: center hideInToc: true onts: sans: "CaskaydiaCove Nerd Font Mono" --- # I Don't Want to Be a System 3 Thinker --- layout: center --- # The Classic Model of Thinking

System 1

System 2

Kahneman, Thinking, Fast and Slow
--- layout: center --- # A New Participant Enters

System 1

Inside the brain
Fast & Intuitive

System 2

Inside the brain
Slow & Analytical

System 3

Outside the brain
Artificial cognition
Cloud-based & generative

The classic model assumed all cognition happens inside a biological mind.
--- layout: center --- # What Makes System 3 Different
1

External

Lives in artificial infrastructure — cloud models, LLMs — not in your nervous system.

2

Automated & Data-Driven

Executes at speed and scale beyond human capability. Performance reflects its training data — including its biases and gaps.

3

Dynamic

Interacts in real time. Can supplement your System 2 — or short-circuit your System 1 entirely.

--- layout: center --- # Two Very Different Things

Cognitive Offloading

Cognitive Surrender

--- layout: center --- # What the Data Shows
>92%
followed accurate AI advice
~80%
followed faulty AI advice
accuracy fell below baseline when AI was wrong
We surrender our accuracy to the machine's accuracy.
--- layout: center --- # The Confidence Trap
Actual Accuracy
Plummets
when AI is wrong
Felt Confidence
Stays High
even when AI is wrong
Even with evidence the AI is steering you wrong,
you keep trusting it.
--- layout: center --- # What Makes Us Surrender
1

Time Pressure

A 30-second limit suppressed System 2 deliberation. Under pressure, the low-friction path wins.

2

High Trust in AI

Individuals who scored high in "Trust in AI" were significantly more likely to follow outputs blindly.

3

Even Incentives Aren't Enough

Financial rewards and real-time feedback helped, but cognitive surrender persisted. It takes focused, intentional effort.

--- layout: center --- # The Move
1

Use System 3 as a Tool

Explore options, scaffold workflows, and accelerate mechanical tasks — not as the final authority.

2

Keep System 2 Running

Don't let your analytical engine idle. Friction builds expertise. Struggle is the feedback loop.

3

Verify, Critique, Own the Result

Be the shepherd of AI outputs. Never surrender the final judgment to the machine.