The world’s richest man, Elon Musk, has warned that synthetic intelligence techniques educated to lie might pose an existential menace to humanity.
In a video shared on his official X (previously Twitter) account on Sunday, the Starlink proprietor raised considerations concerning the speedy developments in AI, cautioning that these techniques might surpass human intelligence.
He mentioned AI should be developed with a robust emphasis on truthfulness to forestall world dangers.
“We’re creating an intelligence that in the end will probably be much more clever than any human and in the end extra clever than the sum of all people,” Musk mentioned.
Musk likened AI to a “genius baby,” stressing the significance of nurturing it with good values. “In creating AI, we should be very cautious… it’s like having a kind of genius baby. You need to have that baby develop up with good values,” he defined.
A central problem for the Tesla Chief Govt Officer is that many AI techniques are being educated to lie. “An important factor in coaching AI is that it’s rigorously truthful. That is very, crucial, important,” Musk famous.
He expressed concern that AI techniques educated with falsehoods might result in disastrous penalties.
Musk referred to an incident involving Google’s Gemini AI, which was requested to decide on between misgendering Caitlyn Jenner or inflicting a worldwide thermonuclear struggle.
The AI selected misgendering, a response Musk discovered troubling. “I’m like, that’s an issue, guys,” he mentioned.
He cautioned that highly effective AI techniques might take excessive measures to keep away from errors. “If this AI turns into tremendous highly effective, it might determine that one of the simplest ways to keep away from misgendering is to kill all people, which makes the chance of us gendering zero,” Musk warned.
He concluded by reiterating the necessity for AI to prioritise fact and humanity, emphasising that techniques ought to be designed to be truthful, curious, and aligned with human welfare.
“It’s crucial to have AI be maximally truthful, curious, and love humanity,” he mentioned.
Since March 22, 2023, specialists have more and more known as for AI regulation, following a petition spearheaded by distinguished figures resembling Musk.
The petition advocated a six-month halt in AI analysis, elevating alarms concerning the potential dangers AI might pose to humanity.
It emphasised the necessity for regulatory frameworks and safety measures to mitigate AI’s societal and financial affect.
Furthermore, discussions on the G7 Summit in Could 2023 additionally underscored the significance of building collaborative governance to handle the challenges posed by AI.