Digital Trends
If you had any doubts about the potential power of artificial intelligence, well, a story just out in Nature should be a wakeup call for us humans. According to the article, Google’s latest foray into AI, a computer system known as the AlphaGo Zero, is now “no longer constrained by the limits of human knowledge,” according to those associated with the program.
What does that mean? Well, you probably remember how an earlier version of the “Alpha Go” computer recently beat the best human Go champion 4 games to 1. Go is a 3,000-year-old Chinese board game considered to be the most complex game known as far as possible positions that can be played. It’s much more complex and random than chess. Scientists figured it would be another 10 years before an AI computer could beat a human player. They were wrong.
Now, AlphaGo Zero, which is actually an even more simplistic design than AlphaGo, just played 100 human Go players – and wiped the floor with all of them. How did it get so good? The original AlphaGo computer learned to play from studying how humans played the game, but AlphaGo Zero used a new scheme called “reinforcement learning:” researchers told the computer what the rules of the game were and made winning a “reward” by earning a point.
The computer then played against itself – completely skipping the human competition/learning step. The result: complete mastery of the game in just two days. But in that time, AlphaGo Zero played millions of matches against itself, learning what moves worked, and according to the Google team, teaching itself moves and strategy no human has used before, thus the 100-to-zip thrashing of the human players it encountered afterwards.
Worried yet? Fascinated? Both? It’s a landmark achievement for AI to be sure, and could point a way forward for, of course, better, faster, and very much smarter AI systems that we’ll all be working with going forward. We’ve got a few links related to this very interesting and somewhat scary development in artificial intelligence programming, so check them out to get the whole story: https://www.digitaltrends.com/cool-tech/alphago-zero/
Haman/Kardon Invoke gets into the smarthome hub game: https://www.digitaltrends.com/wireless-speaker-dock-reviews/harmon-kardon-invoke-review/#/8
Samsung 360 Round VR capture system set to debut: https://www.digitaltrends.com/photography/samsung-360-round-announced-project-beyond/
SUBSCRIBE FOR THE LATEST VIDEOS
http://bit.ly/DTYouTubeSub
VISIT DIGITAL TRENDS
http://www.digitaltrends.com/
DT Daily: https://www.youtube.com/playlist?list=PL8110CBCACD741FEC
DT Originals: https://www.youtube.com/playlist?list=PLEA870D36335F60D2
DT Podcasts: https://www.youtube.com/playlist?list=PLZEIwIHCxaFVemFMYm9Uqixqt7RxRJnhf
FOLLOW US!
https://www.facebook.com/digitaltrends
https://twitter.com/digitaltrends
https://www.instagram.com/digitaltrends
https://plus.google.com/+digitaltrends/posts
Source
Thank you. Have a great weekend DT
When does it become self aware? Lol
AlphaGo Zero did NOT beat 100 top human players. It beat the original AlphaGo program 100-0. This is an oversight.
Sounds like more AI NN Bs. Where is the video for those 100 matches?