"We’re building systems that are becoming superhuman"
The global race for artificial intelligence (AI) dominance is hurtling forward at full speed, but one of the field’s most respected pioneers is sounding the alarm.
Canadian machine learning expert Yoshua Bengio warns that unchecked competition risks creating something far more dangerous than most people realise.
Ahead of next week’s International AI Summit in Paris, Bengio said the rush for AI supremacy—fuelled by panic over China’s DeepSeek chatbot—could have catastrophic consequences.
He said: “The effort is going into who’s going to win the race, rather than how do we make sure we’re not building something that blows up in our face.”
Bengio, regarded as one of the “godfathers” of AI, is no stranger to innovation.
His groundbreaking work on neural networks paved the way for today’s most advanced AI systems.
But despite his optimism about the technology’s benefits on society, he remains deeply concerned about the absence of proper safeguards in the race for more powerful AI models.
Bengio warned: “When you’re in an arms race, ethics and safety tend to get sidelined.
“We’re building systems that are becoming superhuman in some dimensions.
“And as these systems get more powerful, they become extraordinarily valuable economically. That profit motive can blind us to the risks.”
Not all of AI’s founding figures are as worried. Yann LeCun, chief AI scientist at Meta, believes we’re still far from anything resembling true intelligence in AI.
He explained: “We’ve been deluded into thinking large language models are intelligent.
“They’re not. We don’t have machines that are as smart as a house cat when it comes to understanding the physical world.”
LeCun predicts that within five years, AI will achieve some aspects of human-level intelligence robots, for instance, that can perform tasks they weren’t specifically programmed for.
Still, he insists this shift will not necessarily make the world less safe.
In his view, DeepSeek proves that no single country or company will dominate AI for long.
LeCun added:
“If the US tries to lock down AI for geopolitical or commercial reasons, innovation will just pop up elsewhere.”
The debate comes as Bengio, LeCun, and fellow AI pioneer Geoffrey Hinton were in London to receive the Queen Elizabeth Prize for Engineering, the most prestigious global award for engineering.
Previous winners include pioneers of solar panel technology, wind turbines, and electric motors.
UK Science Minister Lord Vallance, chair of the QEPrize foundation, acknowledged the risks but struck a more reassuring tone.
He emphasised the role of new initiatives like the UK’s AI Safety Institute in mitigating potential dangers.
Lord Vallance said: “I don’t believe a single company or nation will dominate AI.
“We’re more likely to see a highly competitive landscape with multiple players around the world.”
As the AI arms race continues to heat up, the question isn’t just who will win—but whether we’ll be able to control the technology we’re racing to create.
Loading ...