Is AI Superintelligence an 'Alien Invasion Within 10 Years'? — The 'Next 10 Years' Strategy for Business and Investors as Warned by Hinton
Geoffrey Hinton, a Nobel laureate who laid the foundation for artificial intelligence research, warned in recent interviews that 'AI becoming superintelligent is like an alien invasion arriving on Earth within 10 years' (CNN, BBC, The Guardian).
Hinton is the person who led the development of deep learning as a co-founder of Google Brain. However, he left Google in 2023 and has publicly expressed concerns about the 'risk of AI running out of control'.
His warning is not merely a science fiction metaphor. It suggests the possibility that the moment AI surpasses human intelligence—the technological singularity—could become reality sooner than expected.
1. Competition-obsessed companies and the hollowing out of 'safety'
1-1. Structural problems that downplay safety
While Hinton acknowledges that some companies like Anthropic and Google DeepMind prioritize safety, he points out that 'many companies prioritize commercial competition and do not take existential risk seriously'.
He states the following:
'They say, "Don't worry because our scientists are in control," but in reality, they are obsessed with the race for hegemony. What we should be more concerned about is whether humanity can survive.'
He also criticizes OpenAI, stating, 'Although they started with safety as their mission, they are losing their responsible stance, and excellent safety researchers are leaving one after another.'
1-2. The 'baby and mother' model: Coexistence through dependence, not domination
Regarding the relationship between AI and humans, Hinton rejects the existing 'CEO and secretary' model.
'The idea that humans are the boss and AI is the subordinate is wrong. There are almost no examples in the natural world where a less intelligent entity can control a more intelligent one.'
The alternative model he presents is the 'relationship where a baby moves its mother'.
In the process of evolution, mothers are instinctively designed to protect their babies. Similarly, he argues that for humanity to coexist with AI, we need 'a design of dependence and trust' rather than 'domination'.
2. The US-China AI race and the crisis of basic research
2-1. The reversal of education and research
Hinton stated, 'The US is still slightly ahead of China, but the gap is closing rapidly.'
He criticized the structure where China is cultivating anoverwhelmingly large educated populationin science, mathematics, and engineering fields, while the US relies on immigrant talent.
Furthermore, he warned, 'If cuts to basic research investment and attacks on universities continue, US innovation power will be depleted in 10 to 20 years.'
3. The economic impact brought by superintelligence
3-1. The reality behind the '$1 trillion AI investment'
According to Hinton, the funds invested by global tech companies in AI-related areas over the past year have reacheda cumulative total of $1 trillion.However, he views the fact that this investment is directed toward 'replacing humans' rather than 'job creation' as a problem.
'The surest way for companies to make a profit is to replace workers with cheap AI.'
Amazon's personnel cuts are one example of this trend. Hinton states, 'Unlike the Industrial Revolution, this change is highly unlikely to generate enough new job categories.'
4. Is 'coexistence between humanity and AI' possible?
4-1. A Problem of 'Social Structure,' Not 'Nuclear Weapons'
Hinton says he has 'mixed feelings' about having led the development of AI.
'Unlike nuclear weapons, AI brings great good in areas like medicine and education. However, the problem lies not in the technology, but in the "social structure."'
If profits are concentrated among a few tech elites, inequality and unemployment will explode—that is the future Hinton fears most.
Summary: Is a 'Chernobyl-level Warning' Necessary?
In closing, Hinton remarked with irony, 'Humanity may not seriously invest in safety measures until a "Chernobyl-level event" occurs where AI runs wild and control is lost.'
Right now, the speed of AI development is at a stage as if an "alien intelligence" were descending upon Earth.
When that happens, can humanity find a path to "coexistence" rather than "domination"? This question will determine the fate of civilization over the next decade.