Enterprise Linux Security Episode 104 - Artificial Insanity

[@f-s-r]

For what i can see, AI is always too sure about it's response, even if it's totally inaccurate. You can tell the IA that the response was inaccurate and it will recognize that it was, and respond again, completely sure about it's answer, which can be wrong again, and that can happen several times if there is conflicting information about the subject on the net. Garbage in -> garbage out.

[@UnwalledGarden]

ChatGPT is not all of AI. It is one publicly available example. AI is better than you make it out to be. Your ansible example just means that ChatGPT isn't well-trained on ansible. I suspect of you prepared a well-trained specialized model for ansible programming, you would get something very useful. For a better example of AI being useful, take Perplexity for a spin.