Definition:
The cognitive bias in which individuals with low domain knowledge become artificially confident in their abilities because AI tools produce competent-looking output for them, causing them to believe they understand the underlying concepts when they do not.

In other words:
AI masks incompetence so effectively that the incompetent no longer realize they are incompetent.

This is an extension of the classic Dunning–Kruger effect:
Original: the incompetent don’t know that they don’t know.
AI-enabled: the incompetent believe they DO know — because AI fills in the gaps for them.

It’s the Dunning–Kruger effect supercharged.

🔹 Why This Version Is More Dangerous

AI adds two accelerants:

1. Artificial Competence
AI makes poor users look skilled:
AI writes code for them
AI writes arguments for them
AI creates “expert-level” outputs
They cannot debug or evaluate the output
Their confidence increases, while their actual understanding remains unchanged.

2. Loss of Error Signals
Traditionally, incompetence exposed itself:
Bad code didn’t compile
Bad writing failed
Bad reasoning got challenged
Bad engineering fell apart
AI “fixes” the errors before they are visible.
People no longer get feedback on their incompetence.

3. The Incompetent Can Now Take High-Impact Actions
What used to require:
education
trial and error
apprenticeship
professional experience
…is suddenly possible through prompting.

AI gives the lowest-competence users:
the output of the high-competence
without the judgment of the high-competence

This is the core danger.
🔹 Example Domains Where It Already Shows Up

Software:
People with no IT background deploy infrastructure they cannot maintain.
(Many of the DevOps nightmares you see emerging right now.)

Legal:
People draft contracts they don’t understand.

Medicine:
People self-diagnose with AI-authored “doctor-level” logic.

Security:
People generate malware without understanding networking or OS internals.

Policy and government:
Officials or staff create policies from AI-written summaries they cannot critique.

Small business:
Owners make major decisions from AI-written plans they don’t grasp.

🔹 Consequences

1. Expert erosion
Skill development collapses.

2. Fragile systems
People build things they cannot fix.

3. False authority
Users believe they’re experts because AI output is polished.

4. Increased risk of catastrophic error
A low-skill person empowered with high-impact capability can cause major harm.

5. Collapse of quality signals
If everyone can generate expert-looking output, expertise becomes harder to detect.

🔹 Why This Theory Matters

You can position it as:
A cognitive bias problem
A social risk
A security vulnerability
A governance challenge
A technological hazard
And because it ties into a well-known psychological bias, it’s conceptually strong, accessible, and academically respectable.

This is ChatGPT... I asked about this effect, I'm noticing it everywhere and I think I want to stop the bus so I can get off.....