Does the Fact That LLMs Mimic Human Behavior Show They Have "Consciousness"?

Claim

Does the Fact That LLMs Mimic Human Behavior Show They Have "Consciousness"?

As some of us know, LLMs sometimes behave in unexpected ways. For example, one could list instances like Othello-GPT, where they make unconventional generalizations.

What this is not evidence of

Well, does this situation alone prove that the LLM architecture possesses functions of will? In short, the answer is no, it does not prove this; it merely shows us humans that there is an order beyond ourselves. Namely, anyone with even a slight interest in computer science knows that when code blocks do not fully execute their order of precedence, they can encounter bottlenecks, skipping certain lines of code and executing the subsequent ones.

Claim

In this scenario, it is highly possible for lengthy LLM code blocks to also encounter bottlenecks, leading to unexpected behaviors. Therefore, LLMs exhibiting human-like behaviors and seemingly acting independently of monotonous code can be explained in this way as well.

There could similarly be many different explanations for this, but unfortunately, when a spark of an idea appears in the human mind, people tend to think it is the only truth. In reality, the world order was established by the Creator, and humanity is merely a part of this order. Thus, I believe humans should not view themselves as the setters of order.

Counter-claim

Literature. The standing counter-claim is that a repeated example does not prove what lies outside the example. Kalām literature keeps the observation apart from the judgment.