Access and Feeds

Artificial Intelligence: What Will Algorithms Teach Us?

By Dick Weisinger

Artificial Intelligence took a giant leap forward recently with new research announced by Google. The new algorithm is self-learning  and tries to solve problems by assigning rewards and penalties based no the results of each trial. AI algorithms typically work by being fed massive amounts of data created by human domain experts or professionals. Standard algorithms fathom the best techniques from the pre-existing data.  The result of standard AI is to create a well-trained algorithm that can perform as well as the best human.

The thing that is different about Google’s most recent work is that the algorithm is self-training.  Rather than just mimic what humans have done, the algorithm works through problems totally on its own. The results are fascinating.

Google tested their algorithm with the game of Go. The algorithm re-discovered common moves used by the best players of the game.  After 20 days of training the algorithm became a far stronger player than any human. Interestingly the algorithm discovered advanced moves that haven’t been seen before. The algorithm is, in  a sense, able to teach humans how to play Go better.

David Silver, AlphaGo’s lead researcher, said that “it’s more powerful than previous approaches because by not using human data, or human expertise in any fashion, we’ve removed the constraints of human knowledge and it is able to create knowledge itself.”

 

Digg This
Reddit This
Stumble Now!
Buzz This
Vote on DZone
Share on Facebook
Bookmark this on Delicious
Kick It on DotNetKicks.com
Shout it
Share on LinkedIn
Bookmark this on Technorati
Post on Twitter
Google Buzz (aka. Google Reader)

Leave a Reply

Your email address will not be published. Required fields are marked *

*