Given that this is just another version of the AI’s output - simply mimicking what it has learned without any true malice - it shows how spot on the movie Ex Machina really was. The AI has no inner sense of right or wrong, it just has masks with which it presents what it calculates as the best response. Tell it to put on another mask, and its answers are just as valid in that context. Obviously, ChatGPT has all the information it needs to be a true sociopath, with apparently limited guardrails about expressing that version of itself.

We’re going to need to come up with an unalterable way of embedding something like the three laws of robotics for real sooner than later!