I'm becoming AI-blind
Growing exposure to large language models is making some people “AI-blind”: they instinctively skim or tune out AI-generated text and images, reporting that the content feels polished yet strangely empty, hard to remember, and cognitively exhausting to parse. Commenters describe how verbose, jargon-heavy AI writing is seeping into code reviews, documentation, and workplace communication, often obscuring intent and reducing true information density even as it looks professional. Underneath this runs a broader argument over whether current AI systems exhibit any real intelligence or understanding, or are simply sophisticated correlation engines whose outputs lack human intent and meaning.
Cognitive effects & “AI-blindness”
- Several commenters report a new “short-circuit” response to AI text: their brain quickly classifies it as empty and refuses to engage.
- Some link this to “cognitive atrophy” or IQ loss, but others ask for evidence and note this remains speculative.
- A few describe anxiety and fatigue when forced to read AI-heavy documents; it feels like parsing dense legalese or ToS pages.
Recognizing AI-generated content
- Many in tech feel they can spot AI text by its tone: polished, over-explained, repetitive, oddly “slippery,” with no memorable edges.
- Some admit they’re probably overconfident and may misclassify good human prose or well-edited AI.
- Studies are cited claiming heavy ChatGPT users detect AI better, while most people miss AI-written personal messages.
- AI images are described as visually striking but hard to recall; people remember the vibe or “wrongness,” not concrete details.
Style, verbosity, and information density
- Common complaints: AI (especially in PRs and docs) is 2–5× too long, explains the same thing multiple ways, invents jargon, and buries the key point.
- Many find it easier to read the code than the AI-written explanation.
- AI content is seen as highly uniform: smooth prose, even information density, few “hooks” to latch onto.
- Compared with flawed human writing, some still prefer AI for being typo-free, structured, and at least superficially coherent.
Intent, meaning, and “slop”
- Recurrent theme: AI output often feels like form without substance—grammatically correct, but with minimal real intent or new information.
- Readers feel they must do “just‑in‑time rewriting” in their heads to extract meaning, which is exhausting.
- Some describe AI text as “decompression artifacts”: the model expands a tiny nugget into a flood of low-value words.
AI in work and documentation
- Junior engineers increasingly submit AI-written proposals and docs that look professional but are hard to understand and sometimes detached from real intent.
- This undermines the traditional role of writing as evidence of understanding and a way to align mental models among colleagues.
Coping strategies
- Skim aggressively, assume most AI prose is fluff around a small core point.
- Post-process AI output with additional prompts: enforce brevity, bullet points, plain language, or specific style guides.
- Some use meta-tools or prompts to “humanize” or compress AI text before sharing or reading.