At the Financial Times AI Future Summit in London recently, leading names in the field of artificial intelligence from Geoffrey Hinton, Yoshua Bengio, Fei-Fei Li, Yann LeCun to Jensen Huang... expressed a clear division in their views on the question: " can AI reach intelligence equal to humans, and if so, when?"
Geoffrey Hinton (University of Toronto, Canada), known as the "father of AI", believes that could happen in less than 20 years.
Mr. Geoffrey Hinton visited the prospect of the future when people debate with a machine, and it always wins.
Hinton recalled his 40-year journey of research on the back-to-back spread algorithm, which is the foundation for deep learning today, and said that the current boom comes from the ability to eventually calculate and data to catch up with ideas.
In contrast, Yoshua Bengio, one of three people along with Hinton and LeCun to receive the Queen Queen Queen Elizabeth Award for Engineering 2025, is more cautious.
Mr. Bengio commented that in the next 5 years, AI can reach the capacity of an average employee, but at the same time emphasized that any prediction about this process still has many uncertainties.
We should be objective and avoid exaggerated statements about AI, Bengio said.
Meanwhile, Yann LeCun - Vice President and Director of AI Science at Meta - was the most skeptical. He compared AI to not reaching the intelligence level of a " cat" and machines reaching the intelligence level of humans will not happen in a sudden moment, but a process of gradualolution over many years.
According to Yann LeCun, progress will take place gradually and the problem is not only investment or data, but the fundamental scientific problem of how to understand the world and learn like humans.
Jensen Huang - CEO of NVIDIA - again considers the question "Is AI as smart as humans?" to be meaningless in practice.
According to him, more importantly, AI is solving real problems, from labor productivity to industrial operations.
We are not building AI to replace humans, but to enhance their capabilities, said Jensen Huang, expressing a similar view to NVIDIA scientist Bill Dally.
On the other hand, Stanfords Fei-Fei Li believes that AI and human intelligence should be viewed as two different types of intelligence.
Machines can identify tens of thousands of objects or translate hundreds of languages, but still cannot pass basic spatial intelligence tests, said Fei-Fei Li.
According to Li, human intelligence is not only language but also the ability to perceive, create and sympathize, something that machines cannot achieve.
When asked whether AI is a sustainable revolution or a temporary bubble, Mr. Jensen Huang affirmed: Unlike the dotcom era, this time all GPUs are used. AI has truly operated the world.
However, Yann LeCun warned that the bubble could coexist in the expectation that the large language model will soon reach human intelligence.
Although each technology expert has a different view of the future of AI in the process of developing into super intelligence, all agree that AI is creating a civilized turning point. And AI is truly reshaping the way people live, work and understand themselves.
