Source : INDIA TODAY NEWS
Anthropic CEO Dario Amodei admits that AI is being developed at an exponential speed and, if left unchecked, it will wipe out all of humanity as early as the end of this decade. While he agrees with a former company researcher’s warning that artificial intelligence could eventually kill all humans, he does make it clear that but says he does not believe it will definitely happen. Instead, Amodei wants the AI industry to slow down so humans have enough time to make the technology safer.
advertisement
Amodei’s recent comments came during an exclusive interview with CNN’s Anderson Cooper, following the resignation of former Anthropic researcher Jacob Coxon. Coxon had said that people building AI genuinely believe the technology could “kill us all” by the end of the decade and accused the industry of moving too quickly.
When Amodei was asked about his views on Coxon’s warning, he said he agrees with it, but not with everything. “I agree with Jacob much more than I disagree with him,” he replied.
Amodei, however, pushed back on the idea of putting a simple percentage on the probability of AI wiping out humanity. ““Instead of saying, it’s a 10% chance, that sounds like it’s a roll of the dice.” He explains that he prefers to look at the different paths AI development could take. According to him, if the industry takes the right path, the chances of something going wrong could be very low. But if it takes the wrong paths, those chances could become much higher.
“If we take the right path, then the chance of something going wrong is very low. If we take the wrong paths, then the chance of something going wrong could be even higher than that.” Amodei added.
This is also the thinking behind his recent call to “pace the frontier” of AI development. In a 3,800-word essay, Amodei argued that AI capabilities have been advancing “drastically faster” in recent months, partly because AI systems are increasingly capable of helping build better AI systems themselves, a process known as recursive self-improvement.
Amodei warned that if this continues unchecked, AI development could eventually move faster than humans’ ability to understand and control increasingly powerful systems.
What does Amodei mean by slowing down?
But here, “slowing down” does not mean stopping AI development altogether. Amodei wants companies to give safety research, testing and oversight more time to catch up with rapidly improving capabilities.
“We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain,” he said while sharing the essay on X.
His proposal includes independent third-party evaluators being embedded inside AI companies with access similar to employees, allowing them to assess models, training processes and safety practices.
“If slowing down bought us even an extra year or two before models reach critical levels of capability, and we used that time to advance alignment, we could greatly reduce the risk that something goes seriously wrong,” he wrote in the blog post.
And many top AI leaders are agreeing with Amodei’s calls. OpenAI CEO Sam Altman said the company would also commit to independent evaluators with employee-like access, while Elon Musk responded to Amodei’s proposal simply by saying, “Dario is right.” on X.
– Ends
SOURCE :- TIMES OF INDIA
