« first day (1201 days earlier)      last day (1922 days later) » 

John Doucette
John Doucette
02:09
I think I'm on the same page as @DuttaA about the topic issue @nbro. I find the breadth of topics covered here to be about right as it stands now. I routinely visit the close queue, and I vote to close questions that I think are low quality, or off topic. But I also think that some questions you flag as off topic are fine, and I leave them open. I think that some of your views regarding what should be off topic are good, but that others do not align with my vision for the stack.
I also find that while some posts are just too low quality to be saved, there's' a big muddy middle where users don't quite know how to phrase what they are trying to ask. The biggest issue for me is that we have very few experts on the site, so the signal / noise ratio is often poor. However, limiting the scope of the site to be less scientific won't attract more experts, it'll attract fewer.
Experts in AI are likely to have multiple degrees in computer science, including a PhD In AI. Indeed, many of our most knowledgeable users either have this background, or are current PhD students. These people are best equipped to answer specific questions about the field, but would be interested in the site because of its scientific value. I am somewhat frustrated by the site's conflation of scientific with technical, and the eagerness to declare scientific discourse off topic.
I think if the site is going to exit beta, it will need to have more expert users, and it will need to embrace the scientific side of AI. To do this, it will need to come to grips with the fact that AI as a scientific field is not data science, and that technical content should be on topic here, as long as it is about AI research and not AI implementation.
nbro
nbro
02:30
Maybe I am just investing my time in the wrong SE site.
I believe that people will agree more with your ideas and suggestions (because my vision for this site is a little too strict), so I think you should take the lead of the site, given that I think you're one of the most knowledgeable users here and you like this site more than I really like it. More importantly, you're willing to take the lead!
John Doucette
John Doucette
I'm unsure whether that is true. I appreciate your perspective on things, and I am truly impressed by the effort you are willing to invest in this site.
I do think the breadth of the field (and thus, of the site to some degree) is problematic.
I think that if we had 10x as many domain experts at each activity level, this would be less of a problem
but making that happen seems like it could be complex.
nbro
nbro
Yes, I completely agree with you. We need more expertise.
DuttaA
DuttaA
I have been on the site for last 2 years and experts are always coming and going. It seems our stack fails to provide the stimulation for an expert to stay.
John Doucette
John Doucette
Something I may try is raising awareness of the site at some of the large AI research conferences, like AAAI. But researchers are busy folks, and it may be difficult to recruit from that pool. I'm only active here because my current job doesn't involve enough teaching and mentoring for my taste, so this is a compelling hobby, but I think most of my colleagues who share that itch work in academia and have an opportunity to teach.
The expert stimulation issue is, I think, related to the rate at which we receive interesting questions (note: not just questions). I think this has actually improved over the time I have been here (about a 16 months). However, it is still not very quick. Sometimes good questions exist, but were never upvoted, and never answered.

I think my favorite questions recently were:
https://ai.stackexchange.com/questions/16296/what-is-the-feminist-perspective-on-artificial-intelligence/16299#16299

https://ai.stackexchange.com/questions/15802/distance-between-pointers-in-stochastic-universal-sampli…
nbro
nbro
02:49
@DuttaA I think the site does not attract many experts because experts are probably solving more difficult problems (and do not have time for anything else) and because the quality of the questions and answers is not very high. I almost never used this site as a reference. I often just use Stats SE and, of course, research papers and books.
John Doucette
John Doucette
@nbro Yeah, I agree. The last two questions in my message above are examples of the kinds of questions I wish we got every day, and that could actually make the site useful as an academic reference.
But I'm unsure how to get more questions like that
nbro
nbro
There are some interesting questions indeed, even though the 1st one isn't yet clear to me. The question in the title is "What is the feminist perspective on artificial intelligence?" (which is a little too broad, because there are many feminists, who may have different opinions), which seems to be slightly different than "Has academic feminism developed theories about the Cyberpunk universe and especially to the problem of artificial life, or about other areas of artificial intelligence?".
However, your answer points out some interesting issues I had already heard of. For example, you say "For example, Siri is a compliant servant, and was gendered (by default) as a woman by its creators. Proponents of this school might say that this reflects and perpetuates cultural trends that devalue women or portray them as passive or subservient."
DuttaA
DuttaA
03:30
@nbro andrej karpathy has a medium blog while yoshua bengio actively participate in quora. I wish someone made them aware our stack exists.
 
9 hours later…
Manuel Rodriguez
Manuel Rodriguez
12:34
@JohnDoucette quote “But researchers are busy folks, and it may be difficult to recruit from that pool.”, – Instead of trying to improve something, the first step is describe the current situation. Today, SE.AI has around 7 new questions per day, which are generated and answered by approximately 150 users. In contrast, the university ivory tower is producing 165 papers about Artificial Intelligence each day which is generated by approximately 300000 experts in that field.
(The number about the ivory tower are based on the assumption that 3% of all the papers and all the academic writers are focussed on Artificial Intelligence).

« first day (1201 days earlier)      last day (1922 days later) »