The Importance of Quality in AI Generating Data

So AI models are as good as the data that it's fed on. One way to enhance the data is just quantity, just add more or hire people to write you demonstration data and then more thumbs up thumbs down for other RLH apps. But I do think that there's so much more that we can explore there as well. Some of our research is sort of investigating AI generated data. And usually AI generated data is not very good. We explored AI generated data that goes through collection of different AI based calling over generating and then criticizing another AI's generation. It's almost like reinforcement learning exploration combined with reward network filtering out some of the bad explorations.