Embed Notice

As we probed deeper on A' terms & concepts that we knew were not in the training set (because they are proprietary), ChatGPT essentially gleefully threw itself off a semantic cliff with no casual user indication.

It never once said "I don't know".

In one hysterically funny example, it created an entire procedural definition of an A' term drawn essentially from pieces of procedural descriptions from covering domain A.

Again, completely credible presentation!

But rotflmao level wrong.