Published on September 21, 2024 at 05:30.
Subscribe to our newsletter “La Lettre Cyber”
Every Wednesday, our experts deliver the essential technological news to help you stay on track in an increasingly connected world. A free newsletter with analyses, practical advice and guides.
The tank is almost empty. It contained college entrance exams, complex math problems, and of course the famous Turing test. But week after week, ChatGPT and competing systems manage to pass all these exams, with increasingly high scores. Hence the idea, launched this week, to find other ways to judge the progress of ultra-efficient artificial intelligence (AI) services. This race is accelerating, as calls to regulate this technology have multiplied in recent days.
Earlier this week, the Center for AI Safety (CAIS), which has set itself the mission of minimizing the risks caused by AI, in partnership with the start-up Scale AI, launched a call. “Humanity must maintain a good understanding of the capabilities of AI systems. Existing tests have become too easy and we can no longer properly follow the evolution of AI, or know what it lacks to reach the level of an expert,” according to the CAIS, which thus proposed a competition, called “The last exam of humanity”. Its goal: to create “the most difficult AI test in the world.”
Want to read our full articles?
For CHF 29.- per month, enjoy unlimited access to our articles, without obligation!
I subscribe
Good reasons to subscribe to Le Temps:
- Unlimited access to all content available on the website.
- Unlimited access to all content available on the mobile application
- 5 item sharing package per month
- Consultation of the digital version of the newspaper from 10 p.m. the day before
- Access to supplements and T, the Temps magazine, in e-paper format
- Access to a set of exclusive benefits reserved for subscribers
Already subscribed?
Login
