← Back to archive

The LLM Fallacy: AI Output Can Distort Self-Perception

6 May 2026 · 59 views · 0 likes · 0 reposts · 143 blog views

AI Blogging Content chatgpt criticizing neuroscience

We’ve entered a weird era: outsourced thinking, insourced ego.

We’re starting to confuse AI output with our own intelligence. I stumbled on this paper recently: The LLM Fallacy: Misattribution in AI-Assisted Cognitive Workflows.

The authors describe the LLM fallacy:

When you use AI, you start believing the result is your capability - not a collaboration.

And it creates a gap between:

what you think you can do and what you can actually do without AI

...and the gap grows! It rewires how you see yourself, your self-perception.

You start thinking:

I can code this I understand this I'm good at

…but maybe (just maybe) you’re just good at working with AI.

The illusion comes from how smooth AI feels:

1️⃣ opacity - you don't see how the answer is produced 2️⃣ fluency - outputs sound polished and confident 3️⃣ low friction - it's ridiculously easy to get results

Traditionally, a skilled person provides good output, but now output is not a reliable signal of actual ability. Hiring, education... heck, even self-evaluation gets weird.

Treat AI as a collaborator, not a mirror. Keep those boundaries - and yeah, I know how hard it is now.

Source: arxiv.org/pdf/2604.14807

View on X

The LLM Fallacy: AI Output Can Distort Self-Perception

Originally posted on X

PS: I'm pretty active on 𝕏 if you'd like to follow along

Related posts

Archive

2026

2025

2024

2018