Two vertical scientists who Today he received this year’s Turing Award To create fundamental techniques of artificial intelligence training, they will employ the spotlight to be afraid of the dangers associated with the rush of AI models for public consumption.
Researcher from the University of Massachusetts, Andrew Barto and former scientist Deepmind Research, Richard Sutton, warned that AI companies do not test exactly products before their release, comparing development to “building a bridge and testing it by using it” According to Financial Times.
The Turing Award, often called the “Nobel Prize in the scope of calculations”, has a prize of $ 1 million and was jointly granted to Barto and Sutton for developing “gaining strengthening” – by machine learning method, which trains AI in order to make optimized decisions through test and error. Senior Vice President of Google Jeff Dean describes the technique As “Lynchpin of progress in artificial intelligence” and remained a “central pillar of the AI boom”, which led to groundbreaking models such as chatgpt OpenAi and Alphago Google.
“Freeing software for millions of people without security is not a good engineering practice,” said Barto Financial Times. “Engineering practice has evolved to alleviate the negative consequences of technology and I do not see that they are practiced by developing companies.”