21
$\begingroup$

Are all questions asked on stats and data science SE also on topic here? Or is there some rule such as (on-topic in stats or data science SE implies off-topic here)?

Data science and the stats SE already have a huge overlap (>~80%), I am worried to have a third SE that also significantly overlaps with them.


As a side note, many other SE have an AI tags, e.g.:

$\endgroup$
  • $\begingroup$ discuss.area51.stackexchange.com/questions/24014/… $\endgroup$ – wythagoras Aug 2 '16 at 16:12
  • 1
    $\begingroup$ It's a good question! Given that so much of what's "hot" and "trendy" in the overall rubric of AI these days is probabilistic / machine learning approaches, I wonder how much is left that people are actively talking about, if we declare all of that stuff off-topic. Obviously there are other things to talk about, but are there people out there looking to have those discussions exclusively? $\endgroup$ – mindcrime Aug 2 '16 at 16:31
  • 2
    $\begingroup$ Relatedly, what's the right action if a question is already asked in one of those sites? Link to the answer there in an answer here, mark as duplicate, write a new answer here? $\endgroup$ – Matthew Graves Aug 2 '16 at 16:42
  • $\begingroup$ Related: Is asking about AI algorithm recommendation on-topic? $\endgroup$ – kenorb Aug 4 '16 at 0:42
16
$\begingroup$

No, data science and the implementation of artificial intelligence are off-topic. A community manager explicitly said so in the Area 51 discussions for this site. There have been at least two AI sites on SE before, and they've all failed. We need to bring something new to the table, especially in the private beta stage. Once that's over, we can consider whether we can bring a new viewpoint to such questions.

$\endgroup$
  • 3
    $\begingroup$ Thanks. This would mean we should close the majority of questions that have been asked so far. (which is fine with me). I'll come back once things get stabilized :) $\endgroup$ – Franck Dernoncourt Aug 2 '16 at 16:45
  • 8
    $\begingroup$ "Somebody said so" isn't much justification for deciding what a community driven site is about. Even if that person works for SE. The community ultimately dictates what it's interested in! Personally I'm extremely skeptical that there are enough people who want to talk about nothing but philosophy of AI to keep a site alive. Ruling everything implementation related off-topic is consigning this site to the trash-bin from the get-to, IMO. $\endgroup$ – mindcrime Aug 2 '16 at 16:52
  • 3
    $\begingroup$ @mindcrime Taking advice from a community manager is probably the best plan. The site should be community-driven, but an extra site covering things that already have a good home elsewhere would be community-driven right out of existence. Again, it's very easy to get a site closed in private beta if it's not constructive for the network. $\endgroup$ – Ben N Aug 2 '16 at 16:54
  • 1
    $\begingroup$ Yeah, part of the problem is, they kinda screwed up by not just calling the existing Data Science SE site "Artificial Intelligence". Or, perhaps, by not calling this one "Artificial General Intelligence" or something to make the distinction more obvious. Given that almost everybody considers ML a subset of AI, people are going to come here wanting to talk about ML. Heck, it's already happening. So if you push those people aside.... what's left? $\endgroup$ – mindcrime Aug 2 '16 at 16:56
  • 3
    $\begingroup$ @mindcrime (people who want to talk about nothing but philosophy of AI can already do so on philosophy.stackexchange.com/questions/tagged/… ) $\endgroup$ – Franck Dernoncourt Aug 2 '16 at 17:07
  • 4
    $\begingroup$ So... no reason to talk implementation here, and no reason to talk philosophy. And my one question trying to get some discussion going around the legal side of things has been downvoted already. So what are we going to talk about exactly? :-) $\endgroup$ – mindcrime Aug 2 '16 at 17:10
  • $\begingroup$ @mindcrime Your legal question was downvoted because it asked for off-site resources. Also, there's a difference between implementation and academics - the applied (implementation) stuff is for Data Science, but we can discuss the more theoretical aspects of AI here. $\endgroup$ – Ben N Aug 2 '16 at 17:17
  • $\begingroup$ @mindcrime I didn't see your question but there is law SE anyway :) law.stackexchange.com $\endgroup$ – Franck Dernoncourt Aug 2 '16 at 20:52
6
$\begingroup$

Definitely not. In some minutes we can see lots of questions asking for specific technical solutions about neural networks and genetic algorithms. I agree with Ben that we need to make this site different and start migrating all these questions to other sites, where there is already an answer to most of them.

Why would we want to ask them again? (apart from rush for reputation)

$\endgroup$
  • $\begingroup$ "Why would we want to ask them again?" -> maybe Datascience SE could answer that one… $\endgroup$ – Franck Dernoncourt Aug 2 '16 at 17:10
5
$\begingroup$

Yes

I am sorry to be the one who posts Yes, but as we are in the beta, I want to be straight forward.

In addition to that, AI is also on-topic in the CS site. I was the one who raised this in the definition phase.

So, a lot of topic which this site aims to cover are already covered in the existing sites.

$\endgroup$
4
$\begingroup$

@RobertCartaino suggested in this post that:

"programming" and "implementation problems" be explicitly listed as outside the scope of this site

in order to direct the authors to sites which were explicitly created to handle these "technical" issues.

This site failed already two times, because people didn't ask the right questions and most of them were already covered by somewhere else (e.g. Stack Overflow, Statistics, Data Science, and similar applied sites).

Basically:

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

a more-comprehensive site which included the development of AI, machine learning, statistical tools, big data, NLP, data mining, etc,

so:

No, machine learning as far as implementation goes is not on topic for this site.

and:

if this site were to simply start reiterating the implementation/tools questions that are already covered elsewhere, this site will not likely make it out of private beta.


On the other hand:

Everything in the proposal is considered when evaluating whether the site would likely be viable. If the proposal looks good across the board, that is the "compelling case"

$\endgroup$
4
$\begingroup$

I believe questions asked at Stats.SE about artificial intelligence should be on-topic here as well, because:

You may suggest they may lacking of AI experts there, so lets move there. However not all AI experts are using or are interested in statistics models with AI.

For example I'm no where near as statistician, I've no idea about cross-validation aka rotation estimation models, but I may use and implement practical AI algorithms.

Therefore I think our site has already its own distinct and unique scope in comparison to Stats.SE, because it is about pure Artificial Intelligence and beyond.

You can still asks about AI at Stats.SE, but it should be focused to statistical learning. To support that, check this post:

Question on AI including a comparison with statistical learning would be pretty clearly on topic here.

They were accepting even without that, but I think most likely because people didn't have the right place to ask. If they've asked, didn't have much attention (maybe AI experts aren't interested in statistical models).

If you've question about theoretical AI, you can consider asking at: CSTheory.SE (not active either).

We can only hope that after 6 years of previous failures, we're able to break some ice this time.

We've one-time final opportunity to not have AI spread across the whole network:

Stats.SE, CSTheory.SE, CogSci.SE, Philosophy.SE, Worldbuilding.SE, SO.SE, CS.SE, HSM.SE, Robotics.SE, GameDev.SE, gosh where else, with no real AI experts in one place.

So basically the goal of this site is as pointed by @lejlot:

To bring people from this one particular field, which exists in between all above in one place. I see the reason behind it - as now questions regarding AI are scattered across these sites and get very little attention from actual experts, who also visit just a subset of these. Additionally - on each site these questions are tagged in a different way, so it is impossible to track them. Unification (new site) would make all of it much easier (and in fact - possible for the first time).

If this going to fail this time, people still will have to have 10-20 different accounts to ask the right questions on the right sites (which is very inconvenient). This would be very sad.


To summary, some say:

statistical learning is not the path to AI (Artificial Intelligence)

but it's open to debate.

$\endgroup$
3
$\begingroup$

I'm going to say "yes". That doesn't mean we need to solicit those kinds of questions, but if / when they show up, I think we should just handle them "organically" if you will. That is, up/down vote them, answer them, comment on them, etc., exactly as we would anything else. I don't see any point in us taking on the effort of cross-checking with other sites and migrating questions, etc. IF the SE infrastructure makes it super easy to do some in some cases, then sure, fine, I guess. But I oppose having ai.se mods waste their energy and time dealing with pedantic quibbling over which site is "most" appropriate for a question.

$\endgroup$

You must log in to answer this question.

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