Mimicking the techniques used by animal trainers to shape the behavior of dogs and horses has proven to be a valuable approach in the development of artificial intelligence. This method was honored on Wednesday with the prestigious A.M. Turing Award, often likened to the Nobel Prize in the tech world. Andrew Barto and Richard Sutton, two trailblazers in the realm of reinforcement learning, have been named this year’s winners for their groundbreaking work.
The innovative research initiated by Barto, aged 76, and Sutton, aged 67, during the late 1970s has been instrumental in the advancement of AI breakthroughs over the past decade. Central to their efforts was the concept of guiding “hedonistic” machines that could continuously adjust their actions based on positive feedback. Their pioneering work in reinforcement learning played a pivotal role in accomplishments such as Google’s computer program triumphing over human players in the game of Go and enhancing tools like ChatGPT and financial trading algorithms.
Despite facing skepticism initially, Barto and Sutton persevered in refining their theories and algorithms at the University of Massachusetts, Amherst. Reflecting on their journey, Barto expressed his satisfaction at seeing the field gaining recognition and relevance. The duo’s pioneering efforts were acknowledged with the annual $1 million award sponsored by Google and presented by the Association for Computing Machinery.
Building upon the foundation laid by early AI thinkers like Alan Turing, Barto and Sutton delved into psychology and neuroscience concepts, particularly those related to reward-based learning. Their seminal work in the 1980s involved a task in a simulated world where they trained an AI to balance a pole on a moving cart. This research laid the groundwork for the extensive use of reinforcement learning techniques in AI applications today.
The impact of their contributions is reflected in the significant advancements made in AI and the substantial investments attracted by the field. Barto and Sutton, both distinguished academics, have influenced a new generation of researchers and played a pivotal role in the current AI renaissance. Their work continues to drive innovation and shape the future of artificial intelligence.
Having worked in AI research for over 14 years, Red characterizes himself as a staunch Luddite, standing firm against the pervasive integration of technology in daily life. In stark contrast, Sutton envisions a future where beings of superior intelligence to current humans will exist—an outlook often associated with the concept of posthumanism.
“People are akin to machines. They possess remarkable capabilities and function in extraordinary ways,” Sutton muses. However, he emphasizes that humans are not the ultimate endpoint in evolution and suggests that there is room for improvement. “This pursuit is inherently intertwined with the realm of artificial intelligence,” Sutton explains. “Our objective is twofold: to gain insight into our own existence and, naturally, to create entities that can surpass our current abilities. Perhaps even evolving into such entities ourselves.”
