8
$\begingroup$

I came to ask a basic fundamental question about AI: In a neural network when inputting nerve input to sense a 2D environment, how do you differentiate two types of objects so the neural network can treat them differently?

That's a solid, fundamental, extremely important AI question. It isnt based on writing code, it's about fundamental neural network structure. I was down-voted and told that's off topic. So I look at the on topic scope here:

  • social issues in a world where artificial intelligence is common,
  • conceptual aspects of AI, or
  • human factors in AI development

Let's address these one by one

social issues in a world where artificial intelligence is common

That's already covered 100% by Worldbuilding Stack Exchange, people have no need to come here for those questions when they could get a response way faster and from a larger and more active community than here.

human factors in AI development

What does that even mean? Seriously, that line means nothing to anyone and should be revised / clarified.

conceptual aspects of AI

This makes sense, but it certainly shouldn't exclude fundamental aspects of AI. In its current state, this site's defined scope makes it useless for anyone who's an expert on AI: all of whom will be interested in creating AI, and will therefore be interested in asking and answering fundamental questions about topics such as structuring and designing AI, which can be asked and answered without involving any code.

To reiterate, no AI experts are going to be drawn to this current scope, it's essentially only useful for the world building audience, which already has a popular SE site. If this is going to be called AI SE, it needs to be a place attractive to actual AI experts in the field, not just science fiction enthusiasts speculating about challenges of a world with AI. Questions about fundamental AI design needs to be on-topic. Not programming questions. Just structural, fundamental design questions.

| |
$\endgroup$
  • 1
    $\begingroup$ Your question ai.stackexchange.com/questions/3329/… falls under the "conceptual aspects of AI". $\endgroup$ – nbro May 17 '17 at 15:02
  • 2
    $\begingroup$ @nbro great, well if "conceptual aspects of AI" covers fundamental AI design questions then I guess my primary argument is null, however, there does seem to be a problem here with that number of questions being at -1 score or lower... 13 of 15 on the front page yesterday, that doesn't bode well. $\endgroup$ – J.Todd May 17 '17 at 15:26
  • 3
    $\begingroup$ We don't need another duplicate of stats.stackexchange.com $\endgroup$ – Franck Dernoncourt May 28 '17 at 20:34
  • $\begingroup$ We don't need a duplicate, but there will be cross-fertilization. I've been active for a long time on SO and still routinely flag to migrate questions that belong on serverfault, stats, datascience, codereview, .... There has to be tolerance for both people confused about the mission of the site as well as questions that can be described as lying at the boundaries. $\endgroup$ – Shawn Mehan Jun 11 '17 at 18:24
8
$\begingroup$

You are correct; conceptual aspects of AI are on-topic and your question does indeed qualify. I hit Leave Open on it in the review queue, so it should survive. People have somewhat different ideas of what the scope is, and especially what the scope should be, so there will be some spurious scope-related admonishments.

The help center's on-topic page is also subject to revision, and I am always happy to adjust it if it needs clarification. Allow me to expound a bit on the current text:

  • Social issues: while Worldbuilding does indeed explore hypothetical worlds, this site requires that answers to these questions have basis in reality. Sci-fi writing is not acceptable here.
  • Human factors: this line was my attempt to describe questions about humans' role in creating or guiding AIs. It was originally inspired by one interesting question about displaying an AI's configuration/state for human inspection (which I can't find at the moment, sorry). I'll think about how best to express this.
  • Conceptual aspects: while non-mathematical concepts are definitely on topic, more concrete implementation issues are already better handled by Cross Validated or Data Science; diffusing those questions across more Stack Exchange sites would add more confusion and duplication.

One thing that isn't captured currently is the academic/humanities arena, as set forth for us back when the site was being considered for private beta. Those questions are definitely also on-topic.

I think our current scope is unique and interesting, though you are right: we could use more experts. Specific proposals for policy changes or wordsmithing are welcome!

| |
$\endgroup$
  • 1
    $\begingroup$ I think this subject matter is so new, industrially speaking, that it's like the early days of programming - you don't have a huge saturation of experts in the general population yet. Those who are experts are highly sought after and likely very busy. That's going to be a long running problem for the next few years while programmers start to take that leap to understanding AI. However, that being said, 13 out of 15 questions being down-voted to -1 or lower makes it clear that there's a problem: Either someone is serial down-voting, someone doesnt understand the scope or disagrees, or $\endgroup$ – J.Todd May 18 '17 at 20:39
  • $\begingroup$ (cont) or the scope is too narrow. If it's one of the former two possibilities, that's going to discourage and drive away potential members at a time when the community desperately needs to grow. If the scope is too narrow, that's a whole different beast to tackle. I'm new here, I can't yet say which one it is. Your guess will be a lot better than mine. $\endgroup$ – J.Todd May 18 '17 at 20:40
  • 2
    $\begingroup$ @Viziionary With a site this small, it's definitely possible for one or a handful of people to greatly affect the front page. If you like, you can help implement our scope by voting up good, on-topic questions. $\endgroup$ – Ben N May 18 '17 at 22:21

You must log in to answer this question.

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