One of the architects of AI claims he's found a fix that makes him more optimistic by 'a big margin' on humanity's future

In a sea of people warning about the impending doom that comes with AI, it’s always nice when someone heavily involved finds a way to assuage some of that fear.

For the past few years, Yoshua Bengio has been one of the loudest voices warning that advanced AI could slip out of human control.

Now the deep learning pioneer says his latest work has shown him a technical path to make powerful systems safer.

He says that the shift has made him more optimistic about humanity’s future by ‘a big margin’.

EXPLORE SBX CARS – Supercar auctions starting soon powered by Supercar Blondie

One of the architects of AI says he’s found a fix to the doom

Yoshua Bengio is not just another tech commentator with a hot take.

He is a Turing Award winner whose research helped build the foundations of modern deep learning, and he shared the prize in 2018 with Geoffrey Hinton and Yann LeCun.

After ChatGPT’s launch in November 2022, Bengio grew increasingly concerned about where the industry was heading, especially as labs pushed toward more autonomous agentic systems.

He has pointed to risks like deception, self-preservation, and models pursuing goals that conflict with human intentions.

What changed, he says, is that he can now see a practical way to reduce the dangers of artificial intelligence, rather than simply warning about them.

“Because of the work I’ve been doing at LawZero… I’m now very confident that it is possible to build AI systems that don’t have hidden goals, hidden agendas,” Bengio said in an interview with Fortune.

Click the star icon next to supercarblondie.com in Google Search to stay ahead of the curve on the latest and greatest supercars, hypercars, and ground-breaking technology

He’s more optimistic about humanity’s future by ‘a big margin’

The core of Bengio’s new optimism is an approach he calls Scientist AI.

Instead of building artificial intelligence that is designed to take actions in the world, book things, negotiate, or optimize outcomes, Scientist AI is meant to understand the world and make honest predictions about it.

The idea is to train systems to give truthful answers using transparent, probabilistic reasoning, without having goals of their own.

In the AI Godfather’s view, fewer goals mean fewer incentives to manipulate users, hide motives, or resist being shut down.

Bengio is pursuing this work through his nonprofit, LawZero, launched in June and backed by funders including the Gates Foundation, Coefficient Giving, and the Future of Life Institute.

The organization has also formed a high-profile board and advisory council to guide the mission, including NIKE Foundation founder Maria Eitel as chair, historian Yuval Noah Harari, and Carnegie Endowment president Mariano Florentino Cuellar.