The advent of artificial general intelligence (AGI) is drawing nearer, leaving policymakers with less time to prepare. This was stated by Anthropic CEO Dario Amodei during a panel discussion at the World Economic Forum in Davos.
The entrepreneur warned that the rapid development of AI might outpace the ability of labor markets and social institutions to adapt to changes.
Amodei believes changes will occur within “a few years, not decades.” He reaffirmed his earlier prediction that AGI will emerge by 2026 or 2027.
“I don’t think it will take long. It’s hard for me to imagine it taking more time,” Amodei emphasized.
The driving force behind accelerated AI development is self-learning, where models automate their own creation. The Anthropic CEO noted that in his company, AI already fulfills the traditional role of a software engineer.
“I have developers who don’t write code. They let the neural network do it, and then they edit. Perhaps in six months or a year, the LLM will do most, if not all, of the work,” Amodei stated.
In his view, progress is limited only by chip supplies and training cycles.
Google DeepMind CEO Demis Hassabis holds more reserved views.
“Significant progress has been made in some areas. In mathematics or programming, it’s easy to imagine how AI can be automated because their results are verifiable. However, in natural sciences, it’s more complex. You can’t know if a chemical compound or a physics forecast is correct,” the expert stated.
Hassabis added that current LLMs are still unable to generate original questions, theories, or hypotheses.
“This is the highest level of scientific creativity, and it’s unclear if we’ll have such systems,” he noted.
The head of DeepMind suggests that AGI might appear by 2030 with a 50% probability.
Consequences
Both executives reached a grim consensus regarding the economic implications. They believe that “white-collar” jobs are at risk.
Amodei opines that up to half of entry-level professional positions could disappear within a decade.
Hassabis warned that even the most pessimistic economists might underestimate the speed of transition.
“Five to ten years is not a lot of time,” he said.
Some analysts argue that changes will manifest not in direct job replacement but in the restructuring of professional activities.
“We should stop asking if AI will replace our jobs and start asking how it will degrade them,” stated Human Voice Media CEO Bob Hutchins.
The impending shift changes the human role from “creator” to “checker,” the expert added.
“This deprives professionals of the ability to make their own decisions and breaks meaningful professional work into unskilled, low-paid tasks,” he noted.
Back in August 2024, British trade unions warned of the risk of millions of job cuts due to artificial intelligence.