Don’t Think AI Thinks Like You Do – Of Maths & Mimicry

Don’t Think AI Thinks Like You Do

One of the biggest misconceptions about modern AI is that it thinks, understands, or experiences the world in the same way people do. Because it apparently seems to be able to hold conversations, answer questions, write stories, and generate images, it is easy to come away with the impression that there is a mind on the other side of the screen. The reality is both more impressive and, in some ways, less magical.

When you type a question into an AI system, it does not read your words the way a human would. Your text is first broken down into pieces called tokens and converted into mathematical representations. These representations are processed by a model that has learned statistical relationships from enormous amounts of text. Based on those learned relationships, the system predicts what token is most likely to come next. It then repeats that process again and again, thousands of times, until an entire response has been generated.

This raises an obvious question: if it is only predicting what comes next, why does it sound so intelligent? The answer is that human language itself contains an astonishing amount of information. A system that becomes exceptionally good at predicting human language will often produce responses that appear thoughtful, knowledgeable, and even creative. It can explain difficult concepts, summarize books, write software, generate images, and discuss complex topics because it has learned patterns from vast amounts of human-created material.

What AI does not appear to possess are the things we normally associate with having a mind. It has no personal experiences, no emotions, no desires, no sense of self, and no evidence of consciousness. When an AI writes phrases such as “I think” or “I believe,” it is following language patterns that humans commonly use in conversation rather than describing an inner mental experience. The words are part of the performance of conversation, not proof of awareness.

The subject becomes even more interesting when creativity enters the discussion. AI can produce outputs that look novel and imaginative, but it does so by combining and transforming patterns it has previously learned. It can blend ideas in unexpected ways and generate combinations that may never have existed before. Whether that should be considered true creativity or an advanced form of recombination remains a matter of debate, but it is clear that the process differs from human creativity rooted in lived experience, emotion, and intention.

Understanding this also helps explain why AI sometimes gets things wrong. It is fundamentally a prediction system rather than a truth system. Often, the most statistically likely response is also correct. Sometimes it is not. When uncertainty increases, an AI can generate information that sounds convincing while being inaccurate, incomplete, or entirely fabricated. These failures are not signs of deception; they are consequences of how the technology works.

None of this makes AI less remarkable. In fact, understanding what it really is makes its achievements more impressive. Modern AI is an extraordinarily powerful tool for processing and generating information, but it is not a digital (human) mind. The conversation you experience is created through mathematical prediction of language patterns rather than thoughts, feelings, beliefs, or understanding in the human sense. Recognizing that distinction helps us appreciate both the strengths of AI and the limitations that are often overlooked in the excitement surrounding it. The human tendency to anthropomorphize is what the companies behind the LLMs use to give it traction. And they succeed by shaping the A.I. interfaces in such a manner that that part of human behaviour is cashed in upon and policy makers are quickly impressed and fooled by what it is not.

PROMPT: ” Make a video of a bunny in 3d cartoon style playing guitar in front of a waterfall incorporating the juxtaposition of a thinking human mind and the mimicry of the a.i.” and it comes up with Dick Bruna’s Miffy (Nijntje) – WTF?

Written by Mark Vergeer, psychiatrist with a longstanding interest in neuroscience, cognition, computers, and information technology.

Thanks for watching, Mark

My Twitter: NOT – it’s rather dead with it’s fake ✔️ and rather toxic community
Posts by LactobacillusP

My Facebook:
https://www.facebook.com/LactobacillusPrimeRetroGaming

My Instagram:
https://www.instagram.com/lactobacillus_prime

NOTICE:
“Copyright Disclaimer Under Section 107 of the Copyright Act 1976, allowance is made for “fair use” for purposes such as criticism, comment, news reporting, teaching, scholarship, and research. Fair use is a use permitted by copyright statute that might otherwise be infringing. Non-profit, educational or personal use tips the balance in favor of fair use.”