The effectiveness of general methods that leverage computation is the biggest lesson learned in years of AI research, according to Rich Sutton's bitter lesson. Ilya's deep belief in scaling up neural networks mimics the philosophy of deep learning research and can pay off in certain styles of research.