I have a naïve (and possibly due to ignorance) question for AI.

1.      AI is not my field and I have stopped doing active research and barely upkeep in my field for almost two decades

2.      My question is “How well does LLM answer well-designed multiple choice questions. This is because answering multiple choice question correctly requires understanding and reasoning over different and multiple data. It is not “The Proof is Out There!. You just need to find it!” (one good example of multiple choice exam is given at the end of my own best selling textbook co-authored with A.E. Bryson. But unfortunately the answer to the exam is also given in the book. Thus this cannot serve as a test.)

3.      My one personal experience is not encouraging. But I am not in this field and don’t know enough about AI

4.      Nor does My 40 minutes search online using Google yielded anything definitive

5.      My current guess is “NOT YET!”

 Can someone enlighten me?