This is a bit of a weird question.

I am hoping to create an online reference since I have some downtime. I know some about statistics but very little about computer science. As a result, the reference guide I am hoping to create will be very statistics oriented - even though I wish that it could be a reference for someone who wants to start from scratch and work their way to AI.

While I would love to be involved with AI, from what I have read about ML and AI, seems like AI does not involve much statistics. (A lot of statistical theory is based off normal assumption and math, and ML seems to bypass that by not requiring strong assumptions nor analytical results). CS seems to be more relevant.

And so my question is, since my guide will mostly cover statistics, how relevant would it be for someone who wants to get into AI? If it's not relevant, then I guess I'll just make my guide for someone who wants to get into stats/data science, as opposed to someone who wants to be an AI researcher.

I guess another way to phrase my question is, as an AI researcher, when you "google" stuff, wikipedia things, or go to your notes, what subjects are you looking at and what exactly are you googling? Are you getting a refresher on how to code back propagation? Or are you getting a refresher on the pros and cons of L1 vs. L2? Do you ever look at how to implement a boosting tree or NN using a pre-existing package?

Basically, I know that what I can provide will be relevant to HS/college stats and data science students. But what really want to do is create something useful for aspiring/current AI researchers. The former is realistic, the latter is a dream. I want to see if my dream is realistic.


  • $\begingroup$ All I can say is ML research makes significant use of statistics and at a very high level(masters/PhD). What you are describing is Data Enginner and not a researcher. Also as a heads up, this is not the correct forum for your question and will likely be closed. $\endgroup$
    – user9947
    Sep 12 '20 at 2:37
  • $\begingroup$ What topics would be considered "high level"? I think what I can provide is more suited for data science unfortunately. $\endgroup$
    – confused
    Sep 12 '20 at 8:53
  • 1
    $\begingroup$ There is no such thing as "AI researcher" in the sense that different people that do research on AI do research on some specific AI topic, such as reinforcement learning, supervised learning, computer vision, etc., unless you are researching how to develop AGIs: in that case, you probably need to do research on all these topics. So, I think you should take this into account. So, ask a more specific question that will not lead to opinions. Moreover, AI researchers typically use research papers and books to acquire their knowledge, rather than blogs or articles. $\endgroup$
    – nbro
    Sep 12 '20 at 10:42
  • $\begingroup$ Look into refpersys.org ... Is it in your opinion an AI project? $\endgroup$ Sep 13 '20 at 10:23
  • $\begingroup$ Well it seems to be related to AI. Why? $\endgroup$
    – confused
    Sep 14 '20 at 10:05

I think AI researcher mostly google some new papers because now there's just a crazy amount of them published. Sometimes people just forget the new concept which was introduced in a paper they read several months ago and they google that concept. Sometimes I forget some loss functions (and intuition behind them) used in a specific area like computer vision, natural language processing or audio processing. Like dice loss or contrastive losses. So these are more advanced things than pros and cons of L1 and L2 losses and how to code backprop. Usually I find the answers on https://distill.pub/ or https://towardsdatascience.com/. Have a look at those, I think they represent correctly current interests and topics for refreshment in AI research community. From my experience I can say not much statistics is used in contemporary AI research (unless you're doing research in statistical learning theory). Sometimes I google some statistical tests to prove results of my experiments are statistically significant and I think that's it.

  • $\begingroup$ Thanks this is very helpful! Gives me an idea as to what you guys look at. Ok, so seems like my knowledge base is suited for data science as opposed to AI Research. Oh well, maybe one day I can do something for the AI industry. $\endgroup$
    – confused
    Sep 12 '20 at 8:49
  • $\begingroup$ Hopefully creating a stats reference will be useful in some way to people. $\endgroup$
    – confused
    Sep 12 '20 at 8:55

Not the answer you're looking for? Browse other questions tagged or ask your own question.