Awakening the AGI God
- By Lachlan Colquhoun
- October 09, 2023
SoftBank chief executive Masayoshi Son is used to making big calls; sometimes, they haven't worked out for him.
It’s a while ago now, but during the dot.com crash of 2000, Son famously lost USD70 billion of his USD78 billion fortune in one day.
Many of his investments have failed, and his ranking as a billionaire is mainly due to one sole investment in Chinese e-commerce giant Alibaba.
So, when Son makes a big claim, as he did last week, that artificial general intelligence (AGI) will surpass human intelligence within a decade, some people will see him as a visionary. In contrast, some might see him as a fool.
Never one given to understatement, Son believes that AGI will surpass human intelligence in a decade and that AI will be ten times smarter than humans in that time.
He’s likely to put his money behind his convictions, with reports of a USD1 billion deal with OpenAI and iPhone designer Jony Ive to create the “iPhone of AI.”
“I want to be on the side of progress,” he said, revealing that he uses ChatGPT daily and that the chatbot has told him his ideas are “feasible and wonderful.”
AI has already exceeded human intelligence in some areas, he said and went even further to claim that "Artificial Super Intelligence" would be realized within 20 years and surpass human intelligence by a factor of 10,000.
"It would be sad to be on the side that gets left behind, like an old, shuttered high street," he said and called on Japanese companies to "wake up" to his vision of a future led by AI.
Doubters, he said, were “hallucinators” and “goldfish”, an opinion he reinforced in his presentation with a goldfish image on a slide.
“It is wrong to say that AI cannot be smarter than humans as it is created by humans,” said Son.
“AI is now self-learning, self-training, and self-inferencing, just like human beings.”
“The youth of Japan, let’s wake up.”
The race is on
Regardless of anyone's opinion of Masayoshi Son, the AI arms race is heating up. OpenAI rival Anthropic is preparing to raise USD2 billion in new funding. At the same time, Amazon has announced it will invest USD4 billion in the company.
Softbank also has skin in the game, with its US investment Arm Holdings raising USD4.87 billion last month in the biggest U.S. IPO since e-truck maker Rivian Automotive raised USD13.7 billion in 2021.
The market, however, doesn’t appear to be as giddy as Masayoshi Son on the company’s prospects, with the shares treading water and falling below the IPO price as analysts have a good look at the outlook for the U.K. chipmaker.
“It is wrong to say that AI cannot be smarter than humans as it is created by humans.”
Even Arm chief executive Rene Haas, appointed to the Softbank board in April, is less bullish than his boss.
While Son believes that AGI will be smarter than humans in a decade, Haas said it would come in his lifetime. He is 62, so perhaps in another 30 years.
Son has some support from video game Doom developer John Carmack, whose AGI startup recently made some big hires and is working towards 2030 as a possible "prototype to show signs of life."
"Nobody has a line of sight on a solution to this today, [but] we feel there is not that much left to do," Carmack told a press conference in September.
Those remaining problems, he said, don't require entirely new thinking or architectures.
It is also a view supported by Max Tegmark, the scientist involved in the open letter published earlier this year calling for a pause in large AI experiments.
“A lot of people here think that we’re going to get God-like artificial general intelligence in maybe three years. Some think maybe two years.”
Science fiction
In looking at the hype, it is worth distinguishing between AGI and other types of AI, many of which are currently being used at commercial scale and more driven by machine learning.
The promise—or threat—of AGI remains hypothetical in 2023.
The idea that AGI can be developed to accomplish any intellectual task performed by a human or an animal remains the stuff of science fiction.
"Uncontrollable artificial general intelligence is science fiction and not reality," according to William Dally, the chief scientist at the AI chipmaker Nvidia, recently at a U.S. Senate hearing.
“Humans will always decide how much decision-making power to cede to AI models.”
What is clear is that there's a big difference between AGI and an AI that can analyze hundreds of thousands of data points to create a climate change model, or one which can make real-time buying suggestions to customers as they interact with a brand.
While the latter is business innovation to drive efficiency and growth and is a reality now, AGI is something else again.
For centuries we have had the term “deus ex machina”—or “a god from a machine”.
Up until now, it has been a literary expression. Still, Masayoshi Son tells us this is less than ten years away.
It sounds amazing, but technology is moving faster than humanity's ability to understand the ethical and social implications of what this might mean.
The god in the machine that our smartest people are working to create may not necessarily be a benevolent one, even if Masayoshi Son is proven correct.
Lachlan Colquhoun is the Australia and New Zealand correspondent for CDOTrends and the NextGenConnectivity editor. He remains fascinated with how businesses reinvent themselves through digital technology to solve existing issues and change their business models. You can reach him at [email protected].
Image credit: iStockphoto/Artystarty