The inventors of reinforcement learning, a foundational technique in artificial intelligence, were awarded the Turing Award yesterday, the computer science equivalent of a Nobel Prize. North American academic researchers Richard Sutton and Andrew Barto outlined the mathematical framework for the technology in a 1998 book, which laid the groundwork for popular chatbots like ChatGPT. The pair will share a $1M prize.
Reinforcement learning was first theorized by the award’s namesake, Alan Turing, in the 1950s, when he argued that sufficient computing power would allow machines to learn through experience like humans and animals. Barto and Sutton began modeling this technique mathematically in the 1980s, seeking ways to incentivize computer systems to seek out rewards and avoid failures. The technique laid the groundwork for the recent AI revolution rooted in large language models, which are often trained via reinforcement learning with human feedback.
Take a crash course in reinforcement learning here.
Comments