6

Proposal: Artificial Intelligence & Robotics

Related question: Closed AI project

Presently there's a lot of "What if ?" as well as "What are the books" or "Where do I start?" type questions. I don't see a lot of questions discussing statistical models or anything specifically pertaining to AI work as much as there is people with a pedestrian interest in AI saying "Yeah, that'll be cool."

These questions aren't very good arch-types for questions and will very likely lead to the proposal being closed for the same reasons our last AI proposal died.

I know AI is really cool, but we need to ask good archetypes of questions that can express many problems. We presently have highly-ranked questions like "What are some good books to learn from?" We don't have lots of these questions on successful sites, we have a handful for reference sake but this is not a typical question that will come up on the site.

tl;dr: Focus on the archtypes of real problems you're facing, not the neat stuff that only represents 3% of the questions.

3

I'm also rather concerned about the lack of good questions hence the announcment that we needed more to look at.

The various proposals for discipines in robotics, machine learning and AI in general appear to be peering over the edge of a precipice and we really need somebody at SE to come up with a solution for everybody.

I had asked Peter Norvig and Sebastian Thrun at Stanford if they could help us champion this but understandably they are very busy and I have not had any reply.

If anybody can help please let us know.

One way forwad would be to have this AI, all of the Robotics proposals and Machine Learning (betamoo's) proposal combined into one united community.

3

It's odd that Thrun and Norvig can attract interest from 160,000 people and this proposal has already died once. Clearly there is an audience out there but we aren't finding it. Have any of the proposals that we are talking about clumping together attracted a significant amount of support or are they all hanging fire like this one?

I just finished ai-class and will continue pursuing some aspect of AI. So while my questions won't be "fanboy" level they won't be practitioner level either. Should I just go away and come back later when I'm more accomplished?

  • No, stick around. See, people who are actively doing something, or solving actual AI problems (entry or high level, just like on Stack Overflow) is a good thing, and we need that. People asking "What if robots are evil and things become Irobot also what books?" lead us to dead-ends. Suggest questions to problems you're facing with your code or models/methods/whatever, just keep it tangible problems that are obvious archtypes of something that we'd like to see a lot of. – Incognito Dec 23 '11 at 2:35
  • To build on what @Incognito has said, technical questions are definately winners. I hate to admit it but I suspect Lego Mindstorms is likely to induct many users with good Q&A style questions in the early stages. This is no bad thing as the tinkerers of todays youth are the engineers of tomorrow. – Moog Dec 23 '11 at 22:52
  • Actually, I don't find it odd that Thrun and Norvig can attract 160,000 people but the questions fail to reach practitioner level. After all, the AI-CLASS.COM was full of fanboy level of comprehension and when discussions on implementation were actually made, I noted there were only a few dozen participants. I actually got voted down on Aiqus when I talked about Python nuts and bolts with people implementing A* and so on. The reason the general programming Stack Overflow worked was there were so many millions of practitioners. – Godeke Jan 6 '12 at 14:56
  • @Godeke - "The reason the general programming Stack Overflow worked was there were so many millions of practitioners." - What would you estimate to be the total pool of practitioners related to this proposal? Tens of thousands? More? Less? – robrambusch Jan 7 '12 at 1:39
  • I don't have industry numbers (I checked both IEEE and ACM and they don't break out AI related fields clearly). One complexity is that many of the people who are doing Machine Learning fall in the Statistics and Analysis side. Based on my own sphere of connections though, I can tell you that I know a lot of programmers and none of them do AI as "practitioners". As a back of the envelope guess, I would say that AI related programmers are 1/50th the quantity of general purpose programmers. (Looking at job postings, I find that ratio to be reasonable.) Interest is high, but the practice is small. – Godeke Jan 11 '12 at 17:05

You must log in to answer this question.

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