It seems that large language models such as ChatGPT have shown impressive abilities in natural language processing tasks, but they are not designed to solve pure algorithmic or strategic problems like a deep reinforcement learning model, where models like AlphaZero is good at and requires much less data. And yet the human brain is divided into two hemispheres, with the left hemisphere typically associated with logical, analytical thinking and the right hemisphere associated with creative, intuitive thinking. These are simplification of very complex systems, but I wonder whether the future is to combine two kinds of models like our brain does to get a better AI that excels at both the language and logic.