Dario Amodei uses a novel analogy to explain how AI and humans work together.
On Thursday’s episode of the Interesting Times with Ross Douthat podcast, the Anthropic CEO compared human engineers and AI working together to the mythical horse-human combination known as a centaur.
He used chess as an example. 15 to 20 years ago, humans could beat AIs or humans playing alone by checking the AI’s output. Now, AI can defeat humans without a layer of human supervision.
Amodei, who co-founded the AI lab Anthropic in 2021, added that a similar shift will occur in software engineering.
“We’re already at the Centaur stage in terms of software,” Amodei said. “If anything, software engineers may be in high demand during the centaur stage, but that period may be very short.”
He said he was concerned that entry-level white-collar jobs would be subject to “massive disruption”. The CEO added that it may be unfair to compare this to a traditional transition. Because the revolution from agriculture to factories to knowledge work took place over centuries or even decades.
“This has been happening for years in the low single digits,” he says.
Amodei is one of the most prominent voices warning that some white-collar jobs, particularly in law, finance and consulting, could be erased by AI. In an essay in January, he predicted that 50% of entry-level jobs could be disrupted by AI in the next one to five years.
Leaders at other top AI institutes, including Mustafa Suleiman and Demis Hassabis, have made similar comments about their work on advanced AI automation services in the next 18 months.
Some software company executives counter that AI will make engineers more productive and that companies will need more engineers.
GitHub CEO Thomas Dohmke said in a podcast in July that “the smartest companies will hire more developers.” “I think it’s a mistake to think that you can build a billion-dollar business with just AI without coding skills.”
Atlassian’s CEO said that as AI advances, people will continue to come up with new ideas for the technology they want, and they will need engineers to build them.
“In five years, we’ll have more engineers working for us than we do today,” Mike Cannon Brooks said in an October interview. “While it improves efficiency, technology creation is not tied to results.”
