At the end of February, I attended IASEAI 2026 in Paris, a conference focused on Safe and Ethical AI. During the conference, I realized the importance of protecting ourselves as AI continues to advance. Aligning AI with human values and developing safe AI agents is a challenging problem.

One reason this problem is particularly difficult is that there is no clear consensus among researchers about the right direction. Professor Yoshua Bengio introduced the concept of Scientist AI, a non-agentic AI designed to counter autonomous AI agents. The paper "Superintelligent Agents Pose Catastrophic Risks: Can Scientist AI Offer a Safer Path?"  is something all of us should read. (https://arxiv.org/abs/2502.1565)