In the 1980s, Andrew Barto and Rich Sutton were considered eccentric devised for an elegant but ultimately doomed idea – learning machines, as people and animals do, from experience.
Decades later, with the technology they now pioneer, they now have more and more crucial for modern artificial intelligence and programs such as Chatgpt, Barto and Sutton awarded the Turing Award, the highest honor in the field of computer science.
Barto, a professor Emeritus at the University of Massachusetts Amherst, and Sutton, a professor at the University of Alberta, has picked a technique known as reinforcement learning, where a computer is persuaded to perform tasks by experiments combined with positive or negative feedback.
“When this work started for me, it was extremely unmodible,” Barto recalls a smile, talking about Zoom from his house in Massachusetts. 'It has been remarkable [it has] achieved some influence and some attention, “he adds.
Strengthening Learning was perhaps the most famous used by Google DeepMind in 2016 to build Alphago, a program that learned how to play incredibly complex and subtle board game to an expert level. This demonstration led to a new interest in technology, which was used in advertising, optimizing energy consumption, finance and chip design. The approach also has a long history in robotics, where it can help to learn machines to perform physical tasks through falling and error.
More recently, learning reinforcement is crucial for guiding the output of large language models (LLMS) and producing extremely capable chatbot programs. The same method is also used to train AI models to simulate human reasoning and to build more capable AI agents.
However, Sutton notes that the methods used to guide LLMs are involved in people who offer goals instead of an algorithm -learning purely through his own exploration. He says that machines can learn completely on their own, ultimately be more fertile. “The big division is or [AI is] Learning from people or that it learns from his own experience, “he says.
The work of Barto and Sutton has been a lynchpin of progress in AI in recent decades, “said Jeff Dean, a senior vice president at Google, in a declaration of the Association for Computing Machinery (ACM) that the Turing award pays out annually.” The tools that have developed, remain a central piller of the a central piller of “” “Pestler of the Pestler of the Pestler of the Pestler of the Pestler of the Pelter of the Pelter of the Pelter of” “” “” “” “” “” “” “” “” “”
Reinforcement has a long and checkered history within AI. It was there at the start of the field, when Alan Turing suggested that machines could learn through experience and feedback in his famous 1950 -paper 'computing machines and intelligence', which investigates the idea that a machine could ever think as a person. Arthur Samuel, an AI pioneer, learning used reinforcement to build one of the first machine learning programs, a system that can play in 1955.
