Can a computer think like a human? Or is it merely an optical illusion that deceives us?
Today, the world's largest tech companies are creating a stir by claiming that artificial intelligence is truly capable of "thinking" something new. However, many do not know that the true beginning of this latest revolution was not in a state-of-the-art laboratory in Silicon Valley, but among a group of video gamers on the notorious 4chan website in 2020.
These young people, engaged in a computer game called 'AI Dungeon' which ran on artificial intelligence, not only made its characters do various things but also asked them to solve mathematical problems. Unexpectedly, even the then-flawed AI systems were able to provide step-by-step explanations when given a math problem, delivering answers in a manner appropriate for the game's characters.
Today, this accidental discovery is widely discussed in the tech world as 'Chain of Thought'. Giant companies like Google and OpenAI highlight this as a technological revolution, claiming their latest AI models "think" before solving a problem. Although institutions like Anthropic try to give these systems a scientific appearance by likening them to a human brain, the early 4chan gamers were more realistic. They clarified the truth in a special note in 2024, stating that none of these AI systems truly think, and they are merely machines that guess the next word based on the data they have been given.
Even researchers at Apple have recently confirmed that this thinking ability is merely an illusion. They point out that when an additional piece of information, irrelevant to a mathematical problem, is added to an AI system that successfully solves it, its success rate drops by about sixty-five percent. True human intelligence does not get misled in such a way. On the other hand, Zach Robertson, a Stanford University student who discovered this same method concurrently with the gamers, also states that it is a clever trick and not true intelligence beyond that.
No matter how much companies worth billions of dollars boast that artificial intelligence has begun to think like humans, the reality is that it is still just a highly successful magic trick that uses our own words to amaze us.