A language model uses a special kind of learning to conduct a ‘probability distribution’ over words used to predict the most likely next word in a sentence based on the previous entry. Language models learn from text and can be used for producing original responses (both short and very long), predicting the next word in a text, speech recognition, optical character recognition and handwriting recognition.

That description is mostly meaningless to the average person. When the average person uses it they announce a different description –“Oh, it’s a chatbot! This one is smart.”

We have a visceral response when hearing AI, and movies exploit this response. Let’s talk about AI, our response to it, puppies and a very specific imperfection that has me concerned.

*Note that I am not picking on CHATGPT, but have had the most experience with it. This episode really applies to any iteration of a language learning model—past and future.