The OpenAI one is kind of fun to play around with. I've asked it, for example, to suggest sets of batch script commands for performing certain functions (basically, to see if it can accomplish very crude programming). It can. However, even for crude example requests/responses it can miss-interpret a bit (or maybe just incorrectly provide a same erroneous result from a similar web search engine query from the past, I see those often at stackoverflow for example). When challenged as to providing a wrong response and basically erroneous code in relation to requests, each time it was able to acknowledge where it made its coding mistakes and then provide an updated correct sample set of commands, but often completely different based coding samples. I am not sure how Artificial it's Intelligence is, as opposed to just an ability to do very fast and very comprehensive parsing, interpreting and look-ups, like the Jeopardy Computer Contestant a while back. However, isn't that also the basics of intelligence, the ability to interpret and match current reality and information to past experiences/knowledge?