CNRS researcher describes GPT-5.6 agent shift in math

A mathematics researcher reflects on the profound shift brought by advanced LLM agents in mathematical problem-solving, arguing that while these tools threaten traditional workflows and prompt ethical questions about authorship, they can be effectively harnessed through careful toolization and human direction. The author examines the psychological and professional implications of outsourcing cognitive tasks to frontier models like GPT-5.6 and newer AI systems. Open article →