Musk, Bengio and others have called for a six-month hiatus from artificial intelligence research

Just over a thousand people, including researcher Joshua Bengio and billionaire Elon Musk, are attracting all artificial intelligence labs: put your technologies on hold for the next six months, rather than the GPT language engine. 4, “Time to take back control of these unpredictable black boxes”.

“AI research and development must be refocused to make powerful, cutting-edge systems more accurate, more secure, more transparent, more reliable,” it reads. Published letter This Wednesday morning with the support of the Future of Life Institute and Mila Quebec. Since its creation in 2014, the Boston-based association has aimed to reduce the risks that threaten humanity, more specifically the risks posed by new technologies.

“Current AI systems are as competent as humans in some tasks,” the published letter said, among others, in particular calling for an end to unrestrained nuclear weapons. “You have to ask yourself: Should we allow the machines to fill us with propaganda and lies? Should we automate all jobs? Should we produce non-human spirits that will eventually surpass and replace us? »

There is no silent answer. A company called Future of Life launched a double appeal. Labs working on developing artificial intelligence systems similar to or more powerful than the GPT-4 just released by California lab OpenAI are invited to take a six-month hiatus during which they can create a series of protocols and rules to make future progress more accountable. And transparent.

“Summer of AI”

Governments, on their part, are invited to work on an AI governance system. This would include, at a minimum, a regulatory authority to oversee the development of technology, a mechanism to monitor the use of highly powerful computing and massive data storage systems, and a system to identify the source of systems that distinguish legitimate applications. From simulations and audit and verification tools.

The researchers who signed the letter believe that adequate public funding and research institutes specifically specializing in finding solutions to economic and political problems should also be considered.

“Humanity will have a prosperous future through AI,” the letter concludes. “We’ve succeeded in creating powerful AIs, it’s time to enjoy the “Summer of AI”, during which we reap the benefits, we invent systems for the benefit of many, and give society the opportunity to adapt.”

Montreal researcher Yoshua Bengio is realistic: publishing the open letter remains a step in the right direction, but a small one. “It will take more sustained efforts,” he said at a press conference Wednesday afternoon.

Pending GPT-5

Trudeau, founder of the Mila Quebec Institute, emphasized the importance of the artificial intelligence and data legislation contained in the government’s Bill C-27. It is moving in the right direction, he said. “One thing I like about this project is that it separates the principles set out in the law from the regulations, which can be quickly adopted when problems arise. »

Bill C-27 would appoint a commissioner for artificial intelligence and coordinate with other federal agencies and departments — such as the CRTC — to ensure the law is respected.

“This is important because it takes a long time to pass a law and it depends on the elected officials in place,” continues Yoshua Bengio. “At the same time authorities can react more quickly. »

Speed ​​of adaptation is critical given the speed at which technology is changing. The Future for Life Institute cites the GPT-4 language comprehension and generation engine developed by California company OpenAI as proof that things are changing fast. “He could pass the Turing test, which has defined the level of human intelligence for decades,” says Joshua Bengio.

Already, it seems that OpenAI could release GPT-5, a new evolution of its technology, before the end of the current year. The race to produce the most intelligent AI doesn’t seem to be stopping itself…

Watch in video