Ten fold

2:43 PM

AI site started...

It seems it is a complete junk...

For instance:

Q: How can I avoid over-fitting when training neural networks?

A: Use "validation" set!

And it gets 5 upvotes...

3 hours later…

Wayne

5:58 PM

Wow, lonely in here... was going to ask a question about DLM's, which I'm learning on my own, but I guess not...

Do it, do it! This chat is now in a process of broad revival! (-;

3 hours later…

Gavin Simpson

8:45 PM

@mbq Fail!

9:35 PM

@mbq I agree...I looked at it too. Somewhat unsurprising since the main general success in AI is in machine learning; everything else is more academic. But most programmers are "interested".

Ha. Case in point:

Q: What are Markov models?

I hear about Markov models and Hidden Markov Models in AI all the time, especially in data mining and NLP. What are they? And why do we use them?

Here's a wild idea: do a small bit of homework before asking a question...

A: Too many general, subjective, and lightweight questions

Star-worth idea.

The best thing is that Noldorin is voting to close random question... same as on physics (-;

I wrote my diagnosis about that place on its meta; even someone upvoted

I think the real problem is that there is no such thing as "experienced AI people". There are: people who want to do AI because it is cool and still believe such thing exists people who tried it and found themselves doing soft computing, machine learning, steering theory or similar avid science...

@mbq You're tempting me with links I can't access!

I didn't sign up for the beta

Right, here is the copy:

I think the real problem is that there is no such thing as "experienced AI people". There are:
- people who want to do AI because it is cool and still believe such thing exists
- people who tried it and found themselves doing soft computing, machine learning, steering theory or similar avid sciences build on the remnants of the original AI movement. Yet those people will assure everybody around and write in a grant proposal / lecture description that they are doing AI
- neuro-cognito-philosophers and cats' brains extractors

To change the topic of conversation, I've noticed that in CV the questions have a lot more meat to them. There are very few - "how do I plot a scatterplot in R"

Most of the questions on the front page are non-trivial

9:51 PM

Certainly. There is also more of them, and the overall traffic increases.

@csgillespie I know. This really is a success...

I think that we can take some credit for it, for really controlling how the community was started.

@Shane I'm all for taking credit ;)

Ay!

I'll be quite satisfied to pass the baton over to @chl and co. :)

@mbq Are you going to moderate on physics?

Probably, but certainly my heart will stay with CV

9:57 PM

I really like the the tex SE. I don't really contribute anything other than page views, but it's very handy for hints and tips.

Yep. I'm glad to see these other exchanges taking off. You never know when they will be useful, even when they're not specifically in your interest area.

@csgillespie Sharpie and Vivi left CV for TeX; remember them?