Anthropic CEO: "AI could surpass human intelligence by 2027"
What arguments does the entrepreneur give?
What arguments does the entrepreneur give?
Dario Amodei, CEO of the startup Anthropic, who developed the AI model Claude, believes that artificial intelligence could surpass human intelligence by 2027, writes The Wall Street Journal.
“I’m relatively confident that over the next 2-3 years we’ll see models that will appear in the workplace, that will be used by consumers, and that will be human assistants — but will gradually become better than us at almost everything,” says the entrepreneur.
Early in his career, he believed that AI could outperform “all humans at almost everything.” Amodei had previously predicted that this could happen in the mid-2020s, although he had serious doubts about that. However, in the past 3-6 months, his uncertainty has diminished significantly.
The Anthropic CEO emphasized that AI that outperforms humans will have both positive and negative consequences. In the short term, which could be 1 to 3 years, he calls for the use of workload comparisons.
“Even if a machine does 90% of your work, the other 10% becomes super-loaded. You spend all your time doing that 10%, but you do 10 times more than the 10% you did before, and you do 10 times more because the other 90% is being used by AI,” Dario Amodei concluded.



