Sign In

Communications of the ACM

ACM TechNews

Why Talking Is So Much Tougher Than Math

View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
Geoffery Hinton

University of Toronto artificial intelligence professor Geoffery Hinton

Photo courtesy of University of Toronto

University of Toronto artificial intelligence (AI) professor Geoffrey Hinton believes that in order for AI to be useful, scientists must focus on finding ways to enable computers to learn.

In a recent interview, Hinton discussed AI's future, including the technology's future opportunities, and diffuses fears about the dangers of AI. He notes that AI can be based on logic or biology. Although the logical side of AI has dominated the field since its inception in the 1970s, the biological approach has recently been gaining support.

In the long run, researchers want to be able to make things that are as smart and adaptable as people, Hinton says. He says that humans are generally poor at the abstract symbolic thinking involved in arithmetic or playing chess, but computers mastered these tasks very quickly. However, it has been much harder to develop AI systems that can master vision and speech recognition. Hinton notes that companies such as Google, Amazon, and Netflix have already used basic AI functions in their services.

From Toronto Globe & Mail
View Full Article

Abstracts Copyright © 2011 Information Inc. External Link, Bethesda, Maryland, USA 


No entries found