Khoi P. N. Nguyen

PhD student in NLP at UT Dallas. Interested in maths, programming, and humanity.

2 minute read academia

interview link

  • In math, AI is taking all the low-hanging fruits, like solving the 50 Erdos problems that haven’t been solved before. But the high-hanging fruits will be much harder to get, like the remaining Erdos problems, or the millenial problems.
  • AI helps him do the things that he doesn’t like doing but have to do, like creating figures and writing code. AI makes the paper richer and boarder, but not deeper. But the core of his work (thinking through tough math problems) hasn’t changed much.
  • Advice for young researchers: There will always be a core thing that is much more important and unique to humans that we have been doing. But the nature of the work is changing quickly, a lot more has become possible, things can be learned much faster. The nature of the problems change faster. Be adaptable and curious.
  • On the science of mathematical methods, he said that we don’t know whether our methods of doing math is the optimal one. We may need to experiment with multiple civilizations. At times, when talking about unsolved problems, there is a constant paranoia of “maybe everyone is missing something very simple”.
  • On prime numbers, which is one of his research topics that is more understanding to the public, he said that it has been very productive to think of prime number generator function as a random function, even though no one have been able to prove that it is random yet.
  • Why he blog? To remember things he has learned and not get frustrated for forgetting. It’s something he does for his own, does when he doesn’t want to do other things.
  • On academia: the more senior you are, the more responsibilities you have, like committees.
  • Serendipity in research: He loves meeting new people, or the old rountines of searching for a journal article and seeing interesting things next to it. Life should have high temperature. After covid, everything must be planned ahead, no more serendipity. Maybe we are optimizing for the wrong thing.

My popup thought

AI will help us save lives and improve happiness. But we should also keep the happiness that we already have. For example, we are happy when learning new things. Sometimes AI makes learning so resentful because you can just prompt an LLM to do it instead of spending time learning.

The hard task for AI is to communicate with human to make them understand what it is saying. For example, if I has solved a difficult problem, how does it get a human to sit down and listen to its lecture? The real gain of AI is in enriching a human long-term capability.

Updated:

Comments