What will happen with questions when AI research advances?

Question

I have been thinking about the "shelf life" of the questions & answers here, and have the following observations:

1. Artificial Intelligence is a rapidly changing, very active research area. I think there are questions open... I mean without current answer, to say it coarsely. I can imagine, that some answers will turn out to be out-of-date or be outperformed many times. It is possible that in one month or one year we get a very different answer, because (a) Some people are researching actively and discovered something amazing, or (b) New users come to the site (and knew of a better answer).

2. AI.SX is definitely different from other sites in stackexchage, because the questions are not like quickies. It is not like I need to solve this urgent issue now, how do I do it?. Many questions have different answers, which often complement themselves. Also, from the comments on an answer (or question), it can be edited to include new points and will be better.

3. The former point is much more noticeable, since this site is about science and not technology. The topic about specific algorithms or techniques has been discussed here on several meta questions. A side-effect is that the questions tend to be (in my opinion) broader. I personally think that is OK, and wish for a certain discussion rather than the answer.

Seeing all that, I think that many questions could be left open for... Well, like forever. Because many are active questions, which cannot be solved like in other sites of the network:

Question → Answer → hasaccepted:yes

Perhaps that could lead to more answers in community wiki, to which one comes (next month) after reading some new things or hearing another conference?

Or we just get new answers to questions with an accepted answer and switch (the checkmark) if the new is better?

What do you think will happen?

Answer 1

Or we just get new answers to questions with an accepted answer and switch (the checkmark) if the new is better?

Yes. That's exactly what we'll do. It makes no sense to do it any other way.

If we have community-wiki answers to everything, than that complicates the reputation system, also - not enough people will reach new privilege levels.

As the field grows and changes, so too the site - we'll have new questions, and new answers to old questions.

Answer 2

This is something to consider even on very technical sites like Stack Overflow. New developments (e.g. new language features) allow new and better solutions to problems. That's yet another reason why questions should allow new answers even after one is accepted. The accept mark indicates that the answer is the best for the question poster at the time. In some cases, the question poster vanishes, never to be seen again on the site. Fortunately, we have something else to measure answer usefulness:

Votes. Posts accept votes forever (in most cases), and new answers (among other events) push the question back onto the site front page so it can be examined anew. Community members should definitely read new posts and vote on their quality. In an ideal world, better answers would always overtake old decent answers in score, but that doesn't always happen. If you see a really awesome answer going unnoticed, you might consider placing a bounty!

As Mithrandir mentioned, community wiki isn't ideal for this scenario, since it has the undesired effect of disabling reputation changes. Newer users should add new takes on the issue via new answers (or possibly comments, if the changes are tiny).