PC legend

[irneb]

Another one:
http://news.cnet.com/8301-1001_3-201...language-dies/

We're loosing giants!

[Dave A]

Whatever has become of self-learning AI programs?

I can remember tinkering with them in the '80s - would have thought they'd have come a long way and be a lot more prominent by now.

And is anyone still using Fortran?

[irneb]

I think the way those "self-learning" AI programs worked in those days was more like a growing decision tree (thus the Lisp language was the most effective for such). These days such algorithms are mostly used inside of database engines (data mining). And of course those attempts at making autonomous cars. It's not so much that AI has disappeared ... it's that the definition is a recursively changing definition

AI research is that which computing scientists do not know how to do cost-effectively today.

E.g. in the 50's optical character recognition (OCR) was considered a form of AI. These day's it's nearly ubiquitous.

[Dave A]

Thinking about it, maybe Google search is a modern day example of what would have been seen as an AI fuzzy logic script back then.

Car motor electronic start-up routines also "learn" as they go too, I've heard.

Amazing really. It's so easy to take digital technology today for granted. It's only when you look back at those early days and just how relatively brief its history is that you get to appreciate just how dramatic progress in the digital revolution has been.

[irneb]

Yep. The deeper tragedy behind this is not so much that we're loosing the people who started it all. It seems to be that IT inventions have been petering out since the 50's. The boom in new technology (electronic) since WW2 and then ever less invention rich decades thereafter ... call them a move towards refining instead of designing.

I wonder if it's a similar scenario as happened at the end of the 1800's - start of the 1900's: i.e. when the explorers ran out of places to explore. Is it that we've reached the limits of possibilities with electronics? Do we need to look for new paradigms for future evolution? Or are we simply on a road to stagnation?

[irneb]

Thinking about it, maybe Google search is a modern day example of what would have been seen as an AI fuzzy logic script back then.

Actually strange that you'd mention a search engine. The Yahoo stores was originally written in Lisp (the "original" AI language of choice): http://discuss.fogcreek.com/joelonso...w&ixPost=31402

[Justloadit]

Unknowingly, we on a daily basis use AI in our software. We generate algorithms, whose outcome is based on the changing input variables, it's just become second nature that we do not even realise it. I did an algorithm in the late 90's still running today, at predicting the standing time of a bore hole based on past information and current information, and doing a predictive wait for the next time it runs out of water. Never get the same time with every calculation, as the conditions keep changing, so does the predictive standing time.

[AndyD]

And is anyone still using Fortran?

I actually came across a protein folding predictive modelling client that is written in Fortran fairly recently. I also thought it had gone by the wayside as a language, the last time I saw Fortran was on the UK university mainframes in the very early eighties. Turns out it's still widely used for military and educational facilities number crunching grunt work.

[wynn]

I read somewhere that if you take man's development from the first 'homo sapiens' till today and compared it to the distance from Pretoria to Cape Town city hall to city hall, everything after the discovery of fire will fit into the last few steps up to the entrance, computers would be like the dust on top of the top step.