Anthropomorphization of AI is a big problem. If we are to use these AI effectively as tools people must remind themselves these are just simple models that buil

Anthropomorphization of AI is a big problem. If we are to use these AI effectively as tools people must remind themselves these are just simple models that build a text response based on probabilities and not some intelligence putting together its own thoughts. It’s kind of like doing a grep search on the entire domain of human knowledge and getting back the results in some readable form. But these results could be wrong because popular human knowledge is frequently wrong or deliberately misleading. Honestly without some sort of logical reasoning component I’d hesitate to even refer to these LLMs as AI. When a program is able to produce some abstract thought from observations of its world, and then find the words on its own to express those thoughts in readable form, then we will be closer to what people fantasize.