AI leaders say machines now match or surpass humans in key cognitive tasks

Leading artificial intelligence experts argue that machines are performing at a human level and, in some cases, surpassing humans in several cognitive domains. This is according to top tech executives attending the 2025 Future of AI Summit in London.

Jensen Huang (Nvidia), Yann LeCun (Meta), Geoffrey Hinton, Yoshua Bengio, Fei-Fei Li, and Bill Dally were among the winners of this year’s Queen Elizabeth Prize for Engineering this week. While accepting their reward, they said machines have already reached human-like levels of intelligence in specific domains.

Huang remarked, “For the first time, AI is intelligence that augments people, it addresses labour, it does work. We have enough general intelligence to translate the technology into an enormous amount of society-useful applications in the coming years; we are doing it today.”

This development comes as HSBC Holdings Plc’s chief executive, Georges Elhedery, cautioned about the surge in corporate spending on artificial-intelligence infrastructure. He said that the scale of investment is outpacing the ability of firms to generate meaningful profits from it in the near term.

Bengio anticipates that machines will perform human tasks in nearly all disciplines

The question of whether and when AI will break through to deliver “artificial general intelligence,” which would be capable of human-level performance, has become one of the defining disputes within the field.

Corporate discourse on AGI increased by 53% in early 2025 relative to 2024, a promising sign of increased focus on its future implications. Some of the most zealous voices on the topic expect AGI within two years; skeptics predict it will take generations.

According to some of the field’s most influential figures, AGI will unfold gradually, not all at once. For starters, Meta’s LeCun commented, “It is not going to be an event because the capabilities are going to expand progressively in various domains.”

Huang also suggested that they had already crossed that threshold, but emphasized that the debate itself is largely academic.

However, these leading AI figures are divided on the question of whether AI will ultimately surpass human ability in all fields.

According to Fei-Fei Li, CEO of World Lab, in certain respects, AI systems have already surmounted human abilities. She provided examples, such as image recognition, where machines can distinguish thousands of objects, and translation, where they can handle around 100 languages.

Geoffrey Hinton, who earned the Nobel Prize in Physics last year with U.S. researcher John Hopfield for their influential contributions to machine learning, remarked, “How long before you have a debate with a machine, and it will always win?” adding that this may be achieved in the next two decades.

Canadian AI pioneer Yoshua Bengio said he believes that, in principle, machines will one day be capable of performing almost anything people can.

However, he admits that they are far from this point. He cautioned against overconfidence in the future of AI, urging neutral observation, as there is a large spectrum of potential outcomes possible for these systems.

King Charles cautioned against “bad actors” in the AI industry

When speaking to the recipients of the awards, King Charles mentioned the problematic presence of bad actors and the rapid emergence of various technologies. Prof. Li even told the press that her conversation with the King focused on ensuring that the technology was used for good, while also considering the risks associated with it. 

She said that AI is a powerful tool, but, like any powerful technology, it has both positive and negative aspects. Prof. Li noted that AI could bring significant benefits to healthcare, education, and manufacturing.

“However, it presents growing dangers, including unemployment and various impacts across industries, as well as the potential rise of deepfakes or decision-making being handed over to machines on a large scale,” she added.

Sharpen your strategy with mentorship daily ideas - 30 days free access to our trading program