Everyone Wants To Be a Reinforcement Learning Startup

These days, artificial intelligence developers, investors and founders are all obsessed with “reinforcement learning,” a method for improving AI by rewarding it for accomplishing certain goals and penalizing it for other behaviors.

Investors tell me that they’re getting pitched on dozens of RL-related startups per week, ranging from those building simulations of real apps for models to learn how to use to those aiming to help businesses specialize models on their performance metrics. 

Existing startups are pivoting into this area as well. One example is LMArena, a site that pits AI models against each other and lets users vote on which response they like the best. Originally funded through a combination of grants and donations, LMArena raised a $100 million “seed” round in May from Andreessen Horowitz and the investment arm of the University of California.