3 Laws for GenAI

Today, I was going through my old notes on GenAI while preparing for an upcoming keynote presentation. While pondering on the capabilities and dangers of GenAI I got remined of Asimov's 3 laws of robotics and felt that the time had come to frame 3 laws of GenAI in a similar manner. After giving this some thought, I have come up with these hopefully adequate 3 laws of GenAI.

1. Thou shalt not steal (nor aid to steal) - GenAI is trained on existing data and essentially recombines and generates something new from what it already knows. If the data used is stolen or taken without the explicit permission of the creators, it leads to copyright issues, plagiarism, and exposure of personal data to danger.
2. Thou shalt not spread untruths - GenAI has a horrible habit of hallucination and can easily produce deepfakes, which are already causing havoc, and this is only the beginning.
3. Thou shalt be consistent - GenAI is “creative” in the way that it does not give exactly the same answer for the same question every time. While this is to be expected, the meaning of the response must be consistent every time, unless there is new information to warrant a change.