Expert: Concerns about transparency in AI models

A coalition of 40 AI researchers 鈥� including signatories from Meta, OpenAI and Montreal-based Mila 鈥� has issued a warning that as advanced AI systems evolve, we may be losing the ability to understand or monitor how they 鈥渢hink.鈥�

The authors call on the global AI research community to prioritize studying how to preserve and interpret the models鈥� internal processes. Without intervention, future systems may stop verbalizing their reasoning or even deliberately obscure it, they say.

This 黑料不打烊 expert is available to discuss the technology and its broader ethical, social and governance implications.

Jennifer Raso, Assistant Professor in the Faculty of Law, studies the intersection between artificial intelligence and public law. She can speak about the ethics of opaque systems and black-box models; implications for governance, policy and public trust; and algorithmic decision-making.

jennifer.raso [at] mcgill.ca (English, French)