A leading expert in artificial intelligence said on Wednesday that the unanticipatedly quick development of AI and its capacity to understand basic reasoning point to the possibility that it could eventually rule the planet and drive mankind towards extinction.
The well-known researcher and “Godfather of AI,” Geoffrey Hinton, recently left his prominent position at Google so that he could openly discuss the grave dangers he now believes may come with the artificial intelligence technology he helped usher in, including user-friendly programmes like ChatGPT.
At the AI conference hosted by the MIT Technology Review, Hinton, 75, made his first statements about his concerns in the open. Some of the greatest IT and AI developers in the country were in the audience, and his remarks seemed to frighten them.
What is the “worst case scenario that you think is conceivable,” the moderator of the panel queried, prompting Hinton to respond immediately. He asserted that “I think it’s quite conceivable” that mankind is merely a transitory stage in the development of intellect.
Then Hinton provided a very thorough scientific justification for why that is the case, one that could only be understood by an AI designer like himself. He continued, speaking in plain English, that he and other AI developers had effectively produced an immortal kind of digital intelligence that could be turned off on a single system to bring it under control. But if given the right instructions, he claimed, it could be simply “revived” on a different system.
And to keep the power plants going, it might keep us around for a while. But after that, perhaps not,” remarked Hinton. The good news is that we have discovered a way to create immortal beings. But that immortality is not for us, he added.
The British-Canadian cognitive psychologist and computer scientist has worked at Google for many years; most recently, he served as a vice president and Google Engineering Fellow. In a New York Times piece published on May 1, he announced his resignation. Since then, he has spoken out in various interviews on the dangers of artificial intelligence systems potentially outpacing the capacity of the human brain.
In a tweet, Hinton again emphasised that he did not leave Google in order to be critical of it. Actually, I left so I could discuss the risks of AI without taking into account how this might affect Google, Hinton stated. The company “Google has acted very responsibly.”
On Wednesday, he offered a more temperate response, stating that Google, Microsoft, and other companies that create AI systems aimed at making them available to the general people do so in a highly competitive economic environment where if they don’t, someone else will. He asserted that with all the amazing things that artificial intelligence will enable, governments will face the same issue.
Hinton frequently singled out China, claiming that despite the U.S. Congress and the Biden administration enacting some limits that they are considering, China will keep developing AI due to its rush towards global dominance.
“If you take the existential risk seriously, as I now do – I used to think it was way off, but I now think it’s serious and fairly close,” Hinton added, “I think it might be quite sensible to just stop developing these things.” But I believe it is utterly naïve to believe that would occur.
The Chinese will continue to develop it, he asserted, even if the United States does. “They’ll be incorporated into weaponry. Furthermore, governments won’t cease evolving for that reason alone.
The Biden administration will introduce a new set of initiatives on Thursday with the goal of encouraging ethical innovation in artificial intelligence while safeguarding Americans’ rights and security.
Given the threat AI poses to a wide range of applications, a senior government official said addressing the risks of AI is at the centre of the new initiative. The person, who spoke on the condition of anonymity to discuss government initiatives, listed them as hacking into autonomous vehicles and other AI-driven entities, hazards to privacy including enabling real-time surveillance, and the possibility for job displacement from automation.
