The only obstacle is AI's 'own conservatism'
A leading scientist warns that AI could surpass human intelligence by 2027 — decades earlier than previously predicted. Mathematician and futurist Ben Goertzel said that advanced AI could create its own artificial intelligence. The scientist predicts that “artificial superintelligence” will be smarter than the entire human civilization.
Computer scientist and CEO who popularized the term "artificial general intelligence" (AGI), believes that AI is on the verge of an exponential «intelligence explosion.»
As the Daily Mail writes, Ph.D. mathematician and futurist Ben Goertzel made this prediction while wrapping up this month's Artificial General Intelligence Summit: «It seems likely that we will be able to achieve human-level AI within, say, the next three to eight years.» ;.
“Once general artificial intelligence reaches human levels,” added Goertzel, who is sometimes called the “father of AGI,” “in a few years you will be able to have completely superhuman AGI.”
Although the futurist admitted that he «may be wrong,» he went on to predict that the only obstacle to runaway, cutting-edge AI — far more advanced than its human creators — is there will be “own conservatism” a bot that advises caution.
Goertzel made his predictions during closing remarks at the 2024 Benefit AI Summit and Unconference, sponsored in part by his own firm, SingularityNET, of which he is CEO.
"There are known unknowns and likely unknowns unknowns,” Goertzel acknowledged during a speech at an event held this year in Panama. – No one has ever created human-level artificial general intelligence (AGI); no one has a clear idea of when we will achieve this.»
But unless the computing power required, in Goertzel's words, «a quantum computer with a million qubits or something like that,» the exponential growth of AI seemed inevitable to him.
"Personally, I think that once you reach human level AGI, in a few years you can get completely superhuman AGI,” he said.
In recent years, Goertzel has been exploring a concept he calls «artificial superintelligence.» (ASI), which he defines as artificial intelligence that is so advanced that it matches all the brain capabilities and computing power of human civilization.
Goertzel listed «three lines of converging evidence» that he said support his thesis.
First, he cited updated work by longtime Google employee, futurist and computer scientist Ray Kurzweil, who developed a predictive model suggesting that AGI will become achievable in 2029.
< p>Kurzweil's idea, which will be detailed in his forthcoming book The Singularity Is Coming, relied on data confirming the exponential nature of technological growth in other technology sectors to help guide his analysis.
Goertzel then cited all known recent improvements made to so-called large language models (LLMs) over the past few years, which, he noted, «have opened up the potential of artificial intelligence to many in the world.»
Finally, the computer scientist, wearing his signature leopard-print hat, turned to his own infrastructure research aimed at connecting different types of artificial intelligence infrastructure, which he calls «OpenCog Hyperon.»
The new infrastructure will bring together more mature artificial intelligence and new forms of artificial intelligence that can target other areas of cognitive thinking besides language, be it mathematics, physics or philosophy, to help create a more comprehensive true AGI.
The work Goertzel "OpenCog Hyperon" has received support and interest from other AI researchers, including Berkeley Artificial Intelligence Research (BAIR), which published a paper last month that he co-authored with Databricks CTO Matej Zaharia and others.
< p>This is not the first potentially dangerous or certainly bold prediction about AI made by Goertzel in recent years, the Daily Mail notes.
In May 2023, the futurist said AI could replace 80% of human jobs «in the next few years.»
«Virtually any job that involves paperwork,» he said at a web summit in Rio de Janeiro the same month, “should be automated.”
Goertzel added that he doesn’t see anything negative about it, arguing that it will allow people to “find something better to do.” than to work for a living.»
The self-proclaimed panpsychist, who said he believes even «a coffee cup has its own level of consciousness,» has suggested that researchers look into creating a «soft superintelligence.»
< p>Gertzel also proposed an artificial intelligence-based cryptocurrency rating agency capable of identifying fraudulent tokens and coins.
But the futuristic computer scientist is perhaps best known for his work on the robot Sophia, the first robot to gain legal citizenship.