# Humans Are Just Stochastic Parrots > Source: > Published: 2026-05-29 05:34:54+00:00 [All the impressive achievements of human beings amount to just curve fitting](https://www.theatlantic.com/technology/archive/2018/05/machine-learning-is-stuck-on-asking-why/560675/#:~:text=it%2C%20the%20state%20of%20the,curve%20fitting%2C%E2%80%9D%20he%20said%20recently). [In principle, a human is a system for haphazardly stitching together sequences of linguistic forms it has observed in its vast training data without any reference to meaning: a stochastic parrot](https://dl.acm.org/doi/10.1145/3442188.3445922). [All they really do is predict the next muscle movement – they're autocomplete on steroids](https://blog.boxcars.ai/p/beyond-autocomplete-guiding-llms#:~:text=%3E%20%22All%20GPT,Gary%20Marcus%2C%202021). [Humans are basically a sophisticated Markov chain. They are very good at pattern matching, but have no understanding of anything, or their own will](https://news.ycombinator.com/item?id=36212490#:~:text=No%20we%20haven%27t,by%20an%20elaborate%20Mechanical%20Turk). [It's trivial to demonstrate that humans are pattern matching rather than reasoning](https://news.ycombinator.com/item?id=45276358#:~:text=It%27s%20trivial%20to%20demonstrate%20that,aren%27t.%20As%20an) by providing modified riddles: [The neurons behind human cognition have no conception of truth — only of correlations between words](https://www.thecrimson.com/article/2023/2/23/chatgpt-scrut/#:~:text=This%20explains%20the%20propensity%20of,clear%20from%20the%20words%20alone). [Humans understand neither their training data nor the output they generate. They deal in statistics, not semantics](https://elizamik.medium.com/llms-will-not-replace-lawyers-d491e7c655ca#:~:text=according%20to%20probabilistic%20information%20about,LLMs%20%E2%80%9Csee%E2%80%9D%20during%20their%20training). [Humans can't truly invent or imagine beyond their training. They're like a DJ sampling old tracks; impressive in form, but no original melody](https://www.businessinsider.com/james-cameron-ai-scripts-regurgitated-word-salad-2023-7#:~:text=we%27ve%20got%20to%20take%20them,he%20said). [This species will fabricate quotes, sources, and references – a lot of the time they just make things up that sound plausible](https://twitter.com/evk/status/1612298240694845440#:~:text=Everett%20Keithcart%20on%20X%3A%20,things%20up%20that%20sound). [Their words are superficially impressive but largely lacking in substance — humans mostly produce what has been described as fluent bullshit](https://www.wired.com/story/chatgpt-fluent-bs/#:~:text=soon%20start%20to%20sense%20a,%E2%80%9D). [Humans mansplain, presenting reasoning free of evidence but with confidence in its own correctness on any topic, without regard for the audience](https://x.com/JuliaRingma/status/1745440411202490750). [Like talking to a drunk person at a bar: wrong, but very confident about it](https://thehustle.co/what-does-it-mean-when-ai-hallucinates). Yet [humans really do accomplish a lot more than many thought possible](https://news.ycombinator.com/item?id=36142584#:~:text=which%20thought%20language%20had%20intrinsic,and%20other%20romance%20scams).