Define when AI is assistant, collaborator, simulator, critic, or replacement risk.

Core idea:
AI should not always play the same cognitive role.

Sometimes AI should simply assist.
Sometimes it should collaborate.
Sometimes it should simulate possibilities.
Sometimes it should criticize human reasoning.
And sometimes, if used poorly, it becomes a replacement risk.

The question is not only:

“Can AI help?”

The deeper question is:

“What role should AI play in this thinking process?”

1. The Five Roles of AI in Human Thinking

Assistant
↓
Collaborator
↓
Simulator
↓
Critic
↓
Replacement Risk

The first four roles can strengthen human cognition.
The fifth role weakens it.

The boundary depends on one thing:

Does the human remain the source of judgment, direction, and responsibility?

2. The Model

AI Role	Core Function	Best Used When	Human Role	Cognitive Benefit	Risk
Assistant	Helps with execution	The task is clear and low-risk	Direct the task	Saves time and reduces mechanical effort	Over-reliance on convenience
Collaborator	Develops ideas with the human	The problem requires exploration	Co-reason and decide	Expands reasoning and perspective	Human may defer too early
Simulator	Models scenarios, arguments, or futures	The decision has uncertainty	Compare possibilities	Improves foresight and systems thinking	False confidence in simulated outcomes
Critic	Challenges assumptions and weak reasoning	The human needs intellectual resistance	Defend, revise, or reject	Strengthens judgment and clarity	Excessive skepticism or confusion
Replacement Risk	Takes over thinking and deciding	The human stops questioning	Passive acceptance	Short-term ease	Agency loss and responsibility drift

3. Role 1 — AI as Assistant

AI as assistant is useful when the task is already defined.

It helps with:

summarizing
formatting
translating
organizing notes
drafting simple outputs
retrieving basic information

Ở đây AI không cần shape the thinking.
Nó chỉ giúp execute.

Human decides the direction
↓
AI helps perform the task

This is healthy when the human already understands the goal.

But it becomes dangerous when assistance quietly becomes dependence.

4. Role 2 — AI as Collaborator

AI as collaborator helps develop thinking with the human.

It can:

propose alternatives
expand arguments
compare viewpoints
identify implications
help structure ideas

This is the ideal mode for deep work.

Human brings purpose
↓
AI expands possibilities
↓
Human judges and integrates

Trong mode này, AI is not just a tool.
It becomes a thinking partner.

But the human must remain the final integrator.

5. Role 3 — AI as Simulator

AI as simulator helps humans think through possible worlds.

It can simulate:

future scenarios
second-order consequences
stakeholder reactions
counterarguments
decision pathways
institutional outcomes

This is powerful for strategic thinking.

Question
↓
Scenario A
Scenario B
Scenario C
↓
Human evaluates trade-offs

The value of simulation is not prediction certainty.

The value is expanded imagination under structured constraints.

The danger is treating simulation as truth.

6. Role 4 — AI as Critic

AI as critic creates productive cognitive friction.

It can ask:

What assumption is hidden here?
What evidence is weak?
What alternative explanation exists?
What consequence is being ignored?
Where is the reasoning too convenient?

This mode is essential for preserving judgment.

Human claim
↓
AI challenge
↓
Human revision
↓
Stronger judgment

AI as critic should not merely disagree.
It should improve the quality of reasoning.

This turns AI from an answer machine into an intellectual resistance system.

7. Role 5 — AI as Replacement Risk

Replacement risk appears when AI no longer supports thinking but substitutes for it.

This happens when:

the human stops forming questions
the human accepts outputs without evaluation
AI makes decisions by default
convenience replaces inquiry
responsibility becomes unclear

AI answers
↓
Human accepts
↓
Judgment weakens
↓
Responsibility drifts
↓
Agency declines

This is not co-thinking.

This is cognitive outsourcing.

The danger is subtle: the human still appears involved, but no longer exercises real agency.

The Framework

Healthy Human–AI Co-Thinking
↓
AI assists execution
AI collaborates on reasoning
AI simulates possibilities
AI critiques assumptions
↓
Human judges
Human chooses
Human remains responsible

The goal is not to use AI less.

The goal is to use AI in the right cognitive role.

Role Selection Table

Thinking Situation	Best AI Role	Why
Clear task, low ambiguity	Assistant	Execution matters more than exploration
New idea development	Collaborator	Multiple perspectives improve thinking
Strategic uncertainty	Simulator	Scenarios reveal consequences
Weak or biased reasoning	Critic	Challenge strengthens judgment
High-stakes moral decision	Support only, not replacement	Responsibility must remain human
User wants instant answer without thinking	Replacement Risk	Convenience may bypass cognition
Learning a difficult concept	Collaborator + Critic	Explanation plus resistance builds understanding
Evaluating a major decision	Simulator + Critic	Possibilities and objections both matter

The Human Control Principle

AI should increase cognitive capacity without transferring authorship.

AI may generate
AI may challenge
AI may simulate
AI may organize

But the human must:
frame
judge
choose
own consequences

This is the boundary between augmentation and replacement.

The Co-Thinking Ladder

Level	Human State	AI Role	Outcome
Level 1	Human gives instructions	Assistant	Faster execution
Level 2	Human explores ideas	Collaborator	Broader reasoning
Level 3	Human tests possibilities	Simulator	Better foresight
Level 4	Human examines assumptions	Critic	Stronger judgment
Level 5	Human stops judging	Replacement Risk	Agency decline

The best systems help users move upward through the first four levels, while preventing collapse into level five.

Evolara Principle

AI should think with humans, not instead of humans.

A good AI system does not merely answer.

It helps humans:

ask better questions
examine assumptions
compare possibilities
strengthen judgment
remain responsible

Human–AI co-thinking is not about replacing the mind.

It is about rebuilding the conditions under which the human mind stays active.

Design Implications

A Human–AI Co-Thinking system should allow users to choose modes:

Mode	Purpose
Assist Mode	Help me execute
Collaborate Mode	Think with me
Simulate Mode	Explore possible outcomes
Critic Mode	Challenge my reasoning
Reflection Mode	Help me form my own judgment

The system should also detect replacement risk.

Warning Sign	System Response
User asks for decision without context	Ask clarifying questions
User accepts output too quickly	Show assumptions and trade-offs
User avoids judgment	Prompt the user to form a view
Topic is high-stakes	Emphasize human responsibility
Reasoning is shallow	Switch to Critic or Reflection Mode

Research Questions

What AI roles best preserve human judgment during complex decision-making?
When does AI collaboration become cognitive outsourcing?
How can AI systems detect and prevent replacement risk?

Knowledge Graph Connection

Concept: Human–AI Co-Thinking
Framework: Assistant → Collaborator → Simulator → Critic → Replacement Risk
Research Question: Can AI function as a thinking partner without weakening human agency?