An #AI training I had to complete recently included these 2 gems:
"AI systems can produce incorrect information, commonly called hallucinations, because they were trained on incorrect or out of date information, so you should check the information and not rely on it." Not mentioned: they don't produce information at all, they produce words probabilistically strung together to resemble information.
"AI systems are trained on information that may be copyrighted, and so might produce output that is in violation of that copyright, so it would be improper to present it as your own without modifying it." Not mentioned: is is improper for the AI to produce copyrighted material that was stolen to train it.
