The biggest lesson from 70 years of AI research is that general computation methods are ultimately going to win out over more expert-tuned methods. However, this idea has been seen as threatening by some in the field who believe that compute is not the only ingredient to success in AI.