| |||||||||||||
Computing failsComputing failsPosted Oct 17, 2012 19:50 UTC (Wed) by man_ls (subscriber, #15091)In reply to: Computing fails by Cyberax Parent article: Plasma Active Three released I don't think a self-driving car would qualify as "Artificial Intelligence" as it was understood 20 or 30 years ago. Sure, there are very clever algorithms involved, but there is no learning involved by the part of the machine. Do you think the robot will adapt to driving in e.g. snowy roads unless the algorithms have been programmed to deal with it? It is just that we (meaning Google engineers) have got much better at developing adaptive algorithms. Compare with the Turing test: impersonating a real human in a conversation. I don't think Wolfram Alpha qualifies, even though I have had many conversations more boring than a single search in the Alpha engine.
(Log in to post comments)
Computing fails Posted Oct 17, 2012 20:14 UTC (Wed) by corbet (editor, #1) [Link] I expect a self-driving car would have indeed been seen as AI, just like chess playing was. Remember one of the central tenets of AI, as I learned in my AI courses so many years ago: once we get something working and understand how to solve the problem, it's not AI anymore...
Computing fails Posted Oct 17, 2012 21:15 UTC (Wed) by dgm (subscriber, #49227) [Link]
> Do you think the robot will adapt to driving in e.g. snowy roads unless the algorithms have been programmed to deal with it?
Yes.
Google's self driving cars are programmed with a mix of clever algorithms, classical AI algorithms and new machine learning techniques. All put together leads to some surprising results on the tests.
I had the opportunity to take Thrun and Norvig's course on AI (www.ai-class.com), and I can say that the field has advanced more in the latest 5 years alone than in the previous 30. It's well worth the effort:
|
|||||||||||||