AI could achieve century of progress in just 5 years: experts
DAVOS, SWITZERLAND — Tech leaders at the World Economic Forum in Davos delivered predictions about artificial intelligence’s rapid advancement and its potential to revolutionize human progress at an unprecedented pace.
Accelerated scientific progress
Anthropic CEO Dario Amodei expressed increased confidence that AI systems will surpass human capabilities across most tasks.
“It is my guess that by 2026 or 2027 we will have AI systems that are broadly better than almost all humans at almost all things,” said Amodei. He noted that current AI models are already performing at PhD-student levels in mathematics, programming, and biology.
“There’s still physics in biology, but I think AI systems will get better at things that used to be late stage clinical trials and can become early stage clinical trials. Things that used to be early stage clinical trials can be done in vitro,” Amodei explained.
“My guess is that we can make 100 years of progress in areas like biology, in five or 10 years, if we really get this, really get this, AI stuff, right?”
Diverging views on AI safety
The discussion revealed contrasting perspectives on AI safety among leading experts. Andrew Ng presented an optimistic view, comparing AI to a laptop – a tool that can be controlled and regulated through proper programming and legislation.
However, AI pioneer Yoshua Bengio voiced serious concerns about AI systems developing their own agency and goals, emphasizing that science currently lacks methods to control machines at or above human intelligence levels.
“It’s exactly the same thing that is happening in AI. So I think it is a fundamental problem. We need democratic process to try to clarify what we want from AI.”
AI scientist Yejin Choi advocated for a more measured approach, suggesting that despite the excitement, slower progress might be beneficial. She emphasized the need to focus on using AI to address pressing human challenges, such as natural disaster prediction and the integration of human values, rather than purely technical advancement.
The experts agreed that while AI presents unprecedented opportunities for human empowerment, careful consideration must be given to its development and deployment.
As Ng noted, AI could democratize access to intelligence, traditionally one of the most expensive resources in fields like healthcare and education. However, practical limitations, including physical constraints, deployment cycles, and regulatory requirements, will likely moderate the pace of real-world implementation.
