You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
This debate between a human and AI is but one giant step forward, even though the human won. The IBM AI put on a credible debate and at the accelerating rate at which all the technologies (memory, processing, programing) that fuel AI/ML are advancing, this is a critical step for it involves on the fly abstract reasoning. At the rate AI/ML is advancing we should start to see major penetration into everything start to come up to speed in about two years. This will be the next major wave to drive demand for semis/mems/nanotech of all types. I don't make this claim lightly, for I have put a very substantial portion of my personal resources on the line. This is going change everything we touch, including things we don't and have the most dramatic change in the economics and social structure of the world. This will also apply to business, economic and physical warfare, subjects I will cover in later posts.
The interesting question is how long will it be before a human can't beat an AI in a debate, like it has done in chess and go?
I still think AI is pseudo science like cold fusion. There is an interesting article in this week end (Feb. 15-16) Financial Times front section page 5 on a medical researcher who thinks AI does not work. Headline: is:
"Scientist warns against discoveries made with AI"
"Inaccurate findings are being made in some areas of biomedical research"
The problems with deep learning are appearing first in medical research because the human body is so complicated. I think it exists everywhere but is being suppressed. One way to think about it is that all AI programs take the type of distribution (binomial for example) as input. There is no way to know what the right distribution is. That is why scientific research is needed.
Also it is not true computers can beat the best Chess players who now also use computers to analyze positions. It is true that Deep Blue computer plus a group of near world class chess players can defeat one chess master who did not have access to computers.
I trust Micron and they are subject to very severe penalties if they lie on a conference call(many people have gone to prison for long sentences for doing this). They have said they were able to get a system using GAN to program itself to beat chess players in hours instead of days. The audience was both live and on conference calls and had some of the best technical people in the employ of financial institutions and no one called them out. Systems have been beating chess and Go masters for a while, I suggest you do more research. I start my morning off with MIT Technology Review, New Atlas that covers the world's technical journals, Forbes, Bloomberg, Barrons, The Economist, Popular Science, Popular Mechanics and many other technical publications. If you feel the information from these is inaccurate, please list your source material and links. If I do publish speculation or theories I my part, I say so. Most of the time I give links and on this subject I have given well over forty links in forums I have posted. I have also listened to numerous TED talks and conference calls, at least a hundred related to this subject and that's very, very conservative. You can read my past posts and look up the links that have supported my writing. Also, I have put very large financial positions based on my work.
