Raw LLM Responses

Whilst I think this video is generally good, it uses a lot of terminology for what is commonly called AI (generative chatbots/agents) that has a lot of semantic bias - "knows", "understands", even the concept that these models are "intelligent" is not true. You wouldn't say that wikipedia is really smart because it understands a lot about history, for example. These LLM models aren't intelligent and likely never will be, no matter how far they are pushed. They might be a component of true synthetic intelligence in the future, and certainly can generate answer-like content, but they really don't have any intelligence at all, they are somewhere between posh markov chains and a chinese room experiment.