1
$\begingroup$

I wrote a terrible question on the SE Meta. (Albeit, I enjoy a significant amount of rep from it because it has about equal up and down votes. The community is split.) But in the process, I learned a lot about asking questions that could be understood.

The purpose was to consider ways SE's own AI might look for things to improve user experience, both helping write good posts and being polite in comments.

I'm new to AI.SE. So, is it okay to ask a Question about whether or how certain factors could be considered in AI algorithms for field-specific uses?

$\endgroup$

1 Answer 1

3
$\begingroup$

In my opinion, in principle, yes that could be on-topic. However, I would try to keep the following things in mind:

  1. The "whether" version of the question would likely not be great, if the expectation is that it's just going to receive a "yes" or a "no" as answer without any elaboration. That's not the type of question that we would want on the StackExchange network in general. So it'd have to lean more towards "how could we apply AI for this?", or even "what would we have to be careful of / what could go wrong if we tried to apply AI for this?".

  2. Make sure that the question is actually about the AI perspective. From your example, if the question is just about whether or not there would exist a connection at all between the grammatical structure of a sentence and its "rudeness", that would probably more so be a question about linguistics (or even culture more broadly), rather than AI. But if we assume that such patterns do indeed exist, a question about how AI could (reliably) pick up on that (or any other form of structure/patterns in text) would be more likely to be on-topic.

  3. Make sure that the question is not overly broad. Especially for people who are not already experts at AI, I think it can be very difficult to not make it either overly broad, or too "small". Like I mentioned in (1), the "whether" question with just a yes/no answer would not be interesting. But a pure "how can AI be used for this problem?" question would be too broad. Valid answers to that could be getting into essentially all AI techniques that have been developed for all kinds of Natural Language Processing tasks over the past several decades, but they could also get into huge practical answers about the required data collection, or they could more into the required software engineering to put the idea into practice... several different directions, and all very broad. We prefer specific, precise questions that can be answered in specific and targeted ways.

$\endgroup$
1
  • 1
    $\begingroup$ This is very helpful because I thought "whether" was more open-ended to Answers going in any direction, but actually "whether" easily implies that answers ought be binary. That distinction really helped! $\endgroup$
    – Jesse
    Commented Mar 20, 2022 at 4:15

You must log in to answer this question.

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