Boston Dynamics founder Marc Raibert says reinforcement learning is helping his creations gain more independence.

Claude 3.7, the latest model from Anthropic, can be instructed to engage in a specific amount of reasoning to solve hard problems.