Top scientists warn that AI could surpass human intelligence by 2027 – decades earlier than previously predicted

The computer scientist and CEO who popularized the term “artificial general intelligence” (AGI) believes AI is on the verge of an exponential “intelligence explosion.”

PhD mathematician and futurist Ben Goertzel made the prediction as he closed a summit on AGI this month: “It seems very likely that we will be able to achieve human-level AGI within, say, the next three to eight years.”

“Once you achieve human-level AGI,” added Goertzel, also known as the “father of AGI,” added “you could achieve radically superhuman AGI within a few years.”

While the futurist admitted he “could be wrong,” he went on to predict that the only obstacle to runaway, ultra-advanced AI — far more advanced than its human creators — would be the bot’s “own conservatism” would encourage caution.

Goertzel made his predictions in Panama City, Panama during his closing speech for the Beneficial AI Summit 2024, which is sponsored in part by his company SingularityNET

Goertzel made his predictions during his closing remarks last week at the ‘2024 Beneficial AI Summit and Unconference’, sponsored in part by his own company SingularityNET, where he is CEO.

“There are known unknowns and probably unknown unknowns,” Goertzel acknowledged during his talk at this year’s event held in Panama City, Panama.

“No one has created human-level artificial general intelligence (AGI) yet; no one knows exactly when we will get there.”

Goertzel has been researching what he calls artificial superintelligence (ASI) – which he defines as an AI so advanced that it can rival all the brainpower and computing power of human civilization combined.

But unless the processing power, in Goertzel’s words, requiring a “quantum computer with a million qubits or something,” an exponential escalation of AI seemed inevitable to him.

“My own view is that once you get to human-level AGI, you can get radically superhuman AGI within a few years,” he said.

In recent years, Goertzel has been researching a concept that he ‘artificial superintelligence (ASI) – which he defines as an AI so advanced that it rivals all the brain and computing power of human civilization.

Goertzel listed “three lines of converging evidence” that, he said, support his thesis.

First, he cited the updated work of Google’s futurist and computer scientist Ray Kurzweil, who has developed a predictive model that suggests AGI will be feasible by 2029.

To support his analysis, Kurzweil’s idea, which will be covered in new detail in his forthcoming book “The Singularity is Nearer,” drew on data documenting the exponential nature of technological growth within other technology sectors.

Goertzel then cited all the known recent improvements made to so-called large language models (LLMs) in recent years, which he said have “made much of the world aware of the potential of AI.”

Goertzel is perhaps best known for his work on Sophia the Robot, the first robot ever granted legal citizenship. Goertzel (right) visited Sophia (left) as part of SingularityNET’s push to create a new online space for the exchange of AI algorithms

Finally, the computer scientist, wearing his signature leopard print hat, turned to his own infrastructure research, designed to combine different types of AI infrastructure, which he calls “OpenCog Hyperon.”

The new infrastructure would combine more mature AI such as LLMs and new forms of AI that could focus on areas of cognitive reasoning other than language, be it mathematics, physics or philosophy, to help create a more complete true AGI.

Goertzel’s ‘OpenCog Hyperon.’ has received support and interest from others in the AI ​​space, including Berkeley Artificial Intelligence Research (BAIR) which hosted an article he co-wrote with Databricks CTO Matei Zaharia and others last month.

This isn’t the first potentially bleak or undoubtedly bold prediction about AI that Goertzel has made in recent years.

In May 2023, the futurist said AI has the potential to replace 80 percent of human jobs “in the coming years.”

“Virtually every job that involves paperwork,” he said that month at the Web Summit in Rio de Janeiro, “should be automatable.”

Goertzel added that he did not see this as a negative thing, claiming that it would allow people to “find better things to do with their lives than work for a living.”

That same month he also told the site as much Futurism: “I took drugs with an AI, if by that we mean I took drugs and then interacted with an AI.”

The ‘psychedelic’ practice, part of his work on ‘algorithmic music composition’ from the 1990s, is just one of many eccentric episodes from Goertzel’s history.

The self-described panpsychist, who has said he believes even “a coffee cup has its own level of consciousness,” has suggested that researchers aim to create a “benign superintelligence.’

Goertzel has also made a proposal AI-based cryptocurrency rating agency able to identify scam tokens and coins.

But perhaps the futuristic computer scientist is best known for his work on Sophia the Robot, the first robot ever granted legal citizenship.