A core goal of the private beta is to generate high-quality content that will attract experts. We are also given the opportunity to invite experts by email to the private beta. My question is simple: exactly what kind of experts are we trying to attract?

According to this question , data science and the implementation of artificial intelligence are off-topic. The problem is that we don't want to become a duplicate of Stats or Data Science SE. The question links to this answer on Area 51 which says that this site is for questions in the "academic humanities arena". This seems to suggest that we want experts in academic humanities.

However, most experts in the field of artificial intelligence are experts of implementation. They are applied mathematicians and computer scientists who are trying to make artificial intelligence a reality. The recent advances in artificial intelligence, like Alpha Go, have been the result of breakthroughs in implementation.

If this site is about humanities-style questions about Artificial Intelligence, then what appeal does it have to the type of people who created Alpha Go, who are primarily computer scientists and mathematicians? I'm not convinced they have special expertise about the ramifications of Artifical Intelligence on human society, politics, law, etc.

Perhaps we need to redefine what this site is about. I think a place to look for inspiration is Math SE and MathOverflow. One is about mathematics at any level, while the other is a site for research level mathematicians. Maybe Artificial Intelligence SE should be to Data Science SE and Stats SE what MathOverflow is to Math SE. That is, it should be a site about tackling research level AI problems with the tools of data science and statistics.

This means that we'll have to seriously elevate the quality of our questions and answers to attract real AI experts. But at least we'll have experts to attract.

  • 6
    $\begingroup$ I wholeheartedly agree with you! As an AI implementation person, I never felt comfortable posting to Data Science or Statistics, as their common use cases (e.g. exploratory data analysis) are quite far removed from those of AI. A expert-level site allowing for AI specific theoretical and implementational issues is definitely what I thought I was signing up for. Disallowing all implementation questions doesn't leave much outside of pop science and Bostrom's books. $\endgroup$
    – zergylord
    Commented Aug 2, 2016 at 22:34
  • 2
    $\begingroup$ That's also what I thought when I saw Artificial Intelligence. It didn't really cross my mind that I was signing up for a humanities-oriented site. $\endgroup$
    – bpachev
    Commented Aug 3, 2016 at 0:29
  • $\begingroup$ "I'm not convinced they have special expertise about the ramifications of Artifical Intelligence on human society, politics, law, etc." They do, however, have experience with the implementation of AI, meaning they can understand its strengths, weaknesses, and quirks. That's useful knowledge to have if someone wants to predict AI's impact on human society, politics, etc. $\endgroup$ Commented Aug 3, 2016 at 1:49
  • $\begingroup$ @zergylord I never had any issue with stats SE: what kind of of implementation questions don't fit there (aside from pure programming or hardware questions)? $\endgroup$ Commented Aug 3, 2016 at 23:27
  • $\begingroup$ @FranckDernoncourt Maybe something like deep reinforcement learning? $\endgroup$
    – bpachev
    Commented Aug 5, 2016 at 20:15
  • $\begingroup$ @bpachev on-topic on CV $\endgroup$ Commented Aug 5, 2016 at 20:32
  • $\begingroup$ @FranckDernoncourt That's true, but I think that zergylord's point is that most CV users can't help with those type of questions, and so it would be good to have an expert-level site. Like Math SE and MathOverflow. $\endgroup$
    – bpachev
    Commented Aug 5, 2016 at 20:49
  • $\begingroup$ @bpachev I see. There are already some strong statisticians and ML folks in CV though. $\endgroup$ Commented Aug 5, 2016 at 20:57

3 Answers 3


First there is a need to distinguish modeling from implementation. They are not exactly the same, although strongly related. This was a very difficult lesson to learn among mathematicians and early programmers, notably in the 70s (mathematical proofs can demand a lot of non-trivial programming work to make them "computable", as in runnable on a computer).

As for Machine Learning (by far the most active AI category), modeling belongs to Data Science SE---perhaps the one thing that most people agree on. Implementation should be out of there, as the issues and focus differ (but again, they are related).

Now, should implementation issues be in AI SE, or StackOverflow? The recurring example is TensorFlow, who's home page states that questions should go to StackOverflow. And we should respect that...

But we should keep in mind that the TensorFlow team will choose SO, because it is the largest community, and because the team has something else to do rather than experimenting with hardly visible communities. Well, size matters. We may think that if AI SE becomes big enough on the implementation side, the TensorFlow team (and other major frameworks) may move actually.

In fact, I think now that implementation questions would benefit from a dedicated site (my view has evolved since the Area 51 definition phase). I have replied and tried to reply to several SO questions related to ML tools, and I think some are out of place compared to other questions. For example, some TensorFlow questions are not really programming questions, and not really framework questions. I mean, there is background knowledge on graph construction and execution, as well as background knowledge about statistics and probabilities that are really necessary to make meaningful contributions.

This is not to say that all questions are out of place on SO. Some are really framework issues or (Python) programming issues, and they are good there.

Based on this opinion, I think the site should be interested in implementation experts, whether they work on ML or Expert Systems (or both?).

See also some threads on Area 51 like this one and this one.

  • $\begingroup$ Related: Should AI programming questions be on-topic? $\endgroup$
    – kenorb
    Commented Aug 11, 2016 at 21:37
  • $\begingroup$ I tend to think exactly the opposite of this answer: AI surely handles the modeling and development of the algorithms, and DS handles issues of implementation and deployment. As mentioned in Ben N.'s answer, the Data science site views itself as "applied" and I think this is consistent with the views of AI researchers as well. $\endgroup$ Commented Sep 2, 2018 at 11:26

First up, when I posted my answer to the question you reference, I was just passing along the information given to us by Robert Cartaino. I'm not wedded to that opinion.

I think all the scientists working on AI would be helpful here even though we're not working on implementation. This is what the original Area 51 Discussion post said (excerpted):

Data Science is an applied site for all the programmers/statisticians/mathematicians who are trying to make this stuff work.

There's some leeway there. Specifically, technical questions seem to be OK, as long as they're not super in depth about the math or programming. There are also "why" questions (as opposed to "how") that are very interesting and educational. I like this question a lot. Scientists are welcome.

We don't have to limit ourselves to the philosophy and practical effects of AI, though they're in scope. Philosophers are welcome too.

  • 1
    $\begingroup$ As a note, Robert Cartaino's opinion was not any kind of consensus, back when it appeared. Let's keep in mind there is quite some division, which is probably good to find the best balance... $\endgroup$ Commented Aug 3, 2016 at 1:10
  • 2
    $\begingroup$ @EricPlaton Indeed, and I think we can broaden our scope a little after the private beta stage. Right now, we need to show that we bring something new to the SE network. $\endgroup$
    – Ben N
    Commented Aug 3, 2016 at 2:05
  • $\begingroup$ Of course the site would benefit from AI researchers. However, this site isn't going to attract them. Why would they decide to contribute a significant amount of time to something that will give them zero to little help? $\endgroup$
    – bpachev
    Commented Aug 5, 2016 at 20:25

For FSM's sake, not this again. Please, no... stop with the "let's attract experts" verbiage. I mean, don't get me wrong.. of course we want experts, but we don't want only experts and we don't want to anoint "experts" with some special degree of relevance. This is a HUGE part of what made it so hard to have a successful ai.se before... we chased away the good, in pursuit of the perfect.

  • 3
    $\begingroup$ Good point. The point of my question is that the only experts in AI are experts in subject matter that is off-topic for this site, therefore there are no experts to attract.. $\endgroup$
    – bpachev
    Commented Aug 3, 2016 at 23:28

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .