Science problems with vast combinatorial search bases, like protein folding, can be effectively tackled using AI models such as AlphaFold. By leveraging large datasets and solved examples, AI can learn complex patterns and provide valuable insights. Similar shape problems in science, like drug compound discovery or material design, offer opportunities for AI to optimize solutions by searching only a fraction of the vast solution space based on model predictions. Utilizing AI tools for these types of problems can lead to significant advancements in various scientific fields.