Descriptions:
Professor Yoshua Bengio — Turing Award winner, the most-cited scientist on Google Scholar with over one million citations, and one of the three researchers universally credited as Godfathers of AI — offers a rare combination of deep technical authority and urgent safety advocacy in this Diary of a CEO conversation. Bengio spent four decades building the foundational techniques behind modern deep learning, yet it was the arrival of ChatGPT in late 2022 that prompted him to go public with concerns he now describes as potentially civilization-defining.
Bengio explains that before large language models demonstrated genuine language comprehension, most researchers — himself included — believed human-level linguistic understanding was decades away. ChatGPT’s emergence collapsed that timeline and, in his view, activated a risk Alan Turing had identified in 1950: once machines understand language, their intelligence becomes a genuine threat to human primacy. He maps specific near-term risk categories under the CBRN framework — chemical, biological, radiological, and nuclear weapons — noting that AI is already lowering the expertise threshold required to design these threats, and discusses how AI-powered persuasion is beginning to measurably undermine democratic discourse.
The conversation also examines the economics driving a robotics boom: cheap AI software has removed the primary cost barrier to physical automation, a shift Bengio observed firsthand at a San Francisco accelerator. He raises the specific security concern of misaligned AI hacking into a global fleet of humanoid robots — a scenario Elon Musk’s own projections of 10 billion humanoid units make increasingly non-theoretical. Bengio closes with conditional optimism: he believes technical solutions to AI alignment are achievable, but only if safety is treated as the binding constraint on development from this point forward.
📺 Source: The Diary Of A CEO
🏷️ Format: Interview
