OpenAI

nitro · Dec 13, 2015

Recommend trader/developers keep an eye out:

Silicon Valley Kingpins Commit $1 Billion to Create Artificial Intelligence Without Profit Motive

"A group of deep-pocketed Silicon Valley investors committed $1 billion to ensure that artificial intelligence serves societal needs rather than commercial interests.

The commitment announced on Friday will fund a non-profit research firm called OpenAI Inc., where researchers will be unencumbered by the pressures of for-profit companies and grant-writing duties of academia.

On its website, the group suggested that companies dedicated to pleasing shareholders should not be allowed to control the future of AI. “Our aim is to build value for everyone rather than shareholders,” the group said..."

http://blogs.wsj.com/digits/2015/12...563142ddb33b4053e007dc1a69c4369&ts=1450021149

Humpy · Dec 13, 2015

There is chasm between limited AI as at present and a machine actually understanding anything and working out a solution.
Could happen next week or 500 years into the future.

nitro · Dec 13, 2015

Humpy said:
There is chasm between limited AI as at present and a machine actually understanding anything and working out a solution.
Could happen next week or 500 years into the future.

Well, Denotational Semantics is the theory behind what you are saying. Here is a good introduction. It is a mystery why math works at all in any field. It is my belief that functions are semantics. So for example, when you do a dot product in a vector space to do neural network weight transform, and then do partial derivatives exploration of the search space through Backpropagation, that has semantic meaning. The tieing of all those things together "tell a story". It is the discovery of how to "tell the story" that computers can't create yet.

In essence, mathematics is pre-canned story telling of tautologous statements, discovered by human beings. It is amazing [to me] that you can take this and create intuition and creativity with it. Computers can check the validity of a statement, but they can't come up with the statement to begin with.

http://homepage.cs.uiowa.edu/~slonnegr/plf/Book/Chapter9.pdf

vanv0029 · Dec 13, 2015

AI is a mirage similar to cold fusion. In particular, machine learning is just normal 20th century searching and feedback system engineering. Here is the video of the 1973 debate on the Lighthill Report that is still relevant. It is the London debate that caused McCarthy and Darpa to eliminate all AI skeptics from the academic research system:
https://www.youtube.com/watch?v=yReDbeY7ZMU

\

There is no AI academic skepticism left except in Copenhagen because of Petrer's Naur's dataology departments. This initiative is the typical Silicon Valley black list/white list system where scientific debate is replaced by PR and advertising revenue funding. AI is needed for social network company revenue growth.

Humpy · Dec 14, 2015

An awful lot of people will be out of a job if AI suddenly happens; so no rush to oblivion imho

syswizard · Dec 14, 2015

AI for trading has proven abysmal at best.....
all it does is curve-fit the past.
Let's all remember LTCM....and all of those PhD's.

Humpy · Dec 14, 2015

syswizard said:
AI for trading has proven abysmal at best.....
all it does is curve-fit the past.
Let's all remember LTCM....and all of those PhD's.

Talking about PHD's one should be impressed I suppose or should we. The only one I knew quite well was a bit of a disappointment. He didn't speak much. Just enough to make oneself talk. I was mistaken if I thought I was going to learn anything. It suddenly dawned on me his forte was memory not extrapolating from knowledge very much to make interesting or useful conclusions. Of course when you look at the educational system the all important exams are largely based on learning lots of facts etc. Useful but not much good for progressing into the future by just quoting the past. Even Einstein was an average pupil at Uni it seems.

jk90029 · Dec 19, 2015

syswizard said:
AI for trading has proven abysmal at best.....
all it does is curve-fit the past.
Let's all remember LTCM....and all of those PhD's.

Since stock market is a lot more efficient than one expect, even PhD does NOT have edge. But PhD has a little advantage, IMHO.

LTCM with novel prize people does NOT have significant edge too.

Jerry030 · Dec 19, 2015

syswizard said:
AI for trading has proven abysmal at best.....
all it does is curve-fit the past.
Let's all remember LTCM....and all of those PhD's.

Are you perhaps expecting those few who have found exceptional success using predictive analytics or AI for trading to sell it as a system for $99.95 or offer a how to do it webinar for $300? It is a very rarified skill set which first requires an understanding of time and causality and no small amount of quantum physics. You don't get there by feeding a triple moving average into an SVM an expect it to spit out winning trades.

Cdntrader · Dec 20, 2015

Humpy said:
There is chasm between limited AI as at present and a machine actually understanding anything and working out a solution.
Could happen next week or 500 years into the future.

Well it will be one thing for the computer to learn how to learn, but becoming self aware and having a consciousness seems like a HUGE stretch to me. Pretty much unknowable unless they discover how "Life" works - or when Skynet takes over.