In Reply to: AI: Sentient or Unconscious? posted by SagoBob on August 27, 2026 at 08:15:58
As far as LLMs are concerned, there really is nothing else going on. For other systems doing other things, I don't know. All this news about AIs hacking other systems to achieve better scoring on their tasks would suggest some rudimentary reasoning going on, and maybe an extremely limited understanding of the shape of the world, i.e, what is connected to what and how doing something there has an impact on what happens here, which is a start but not where it needs to be before we can talk about a reasoning intelligence.
LLMs alone though, are a very different beast: all they know is strings of characters and how to put them together into an acceptable string. It's a specialized form of AI that leverages the most advanced branch of AI: pattern recognition. (We used to call that Machine Learning back in the day before it got labeled as "AI" because that's a phrase that investors get excited about anything that has to do with something as useless as "Learning".)
Think of it this way: LLMs are the language version of the image generators. No one says a program that generates an image must be sentient. But as soon as they produce a sentence, some primal corner of the human mind goes "it's alive, its alive". A very human response, but doesn't make it true.
The reason I talk about reasoning is you need to get through decision making and reasoning before you can get get to sentient.