Anthropic CEO Sees AI-Powered Advances Doubling Human Lifespans

Anthropic CEO Dario Amodei said Thursday (Jan. 23) that accelerated advances in artificial intelligence (AI), particularly in biology, can lead to a doubling of human lifespans in as little as five to 10 years “if we really get this AI stuff right.”

During a panel at the World Economic Forum in Davos, Amodei called this the “grand vision.” He explained that if AI today can shrink a century’s worth of work in biology to five to 10 years, and if one believes it would take 100 years to double the average length of human life, then “a doubling of the human lifespan is not at all crazy, and if AI is able to accelerate that we may be able to get that in five to 10 years.”

Amodei also said that Anthropic is working on a “virtual collaborator,” an AI agent capable of doing higher-level tasks in the workplace such as opening Google Docs, using the Slack messaging channel, and interacting with workers. A manager will only need to check in with this AI agent “once in a while,” similar to what management does with human employees.

Amodei said that in the last three to six months, he has become “more confident” that “we really are heading towards AI systems that are better than almost all humans at almost all tasks.” He has seen some AI models starting to be as good as Ph.D. students in math, programming and biology.

Amodei believes that by 2026 or 2027, “we will have AI systems that are broadly better almost all humans at almost all things.” Anthropic was founded by former OpenAI engineers and backed principally by Amazon and Google. Anthropic differentiates itself from OpenAI, its chief rival, with its focus on constitutional AI that seeks to make the AI helpful and harmless.

However, with all these smart and capable AI agents coming, “do we immediately solve all of the world’s problems?” Amodei asked rhetorically. The implied answer is no. He said the two main hurdles to AI advancing rapidly, which one could currently see in the struggles to develop a fully self-driving car, is the physical world and bureaucracies from human institutions.

Another example Amodei gave was in the realm of drug development. While AI can come up with solutions for drugs, they still need to go through clinical trials and get regulatory approval.

Pfizer CEO Albert Bourla, who was in the same panel, said AI is shrinking drug discovery timelines from years to months. But several steps are needed before these advances can directly benefit the public. For example, AI can help identify promising targets, but there are subsequent processes, such as the need to pair an antibody to a specific protein commonly found in cancer cells, and then arm it with a “warhead” to destroy only cancer cells, not healthy ones.

Uber CEO Dara Khosrowshahi acknowledged that for autonomous vehicles, the physical world remains a challenge that needs to be solved. For example, a defaced stop sign can still confuse a self-driving car, something that would not befuddle teen drivers. Khosrowshahi, however, remains bullish on autonomy. He sees electric, AI-driven vehicles as the future of transportation. He believes one day all vehicles on the road would coordinate with one another and alleviate congestion on the streets.

Khosrowshahi also pointed to another challenge for autonomy: Lack of trust from the public. He said that it’s easier for people to accept mistakes from other human beings but the AI needs to be 10 times better than humans for it to be trusted. Thus, while he believes self-driving cars are safer than human-driven vehicles, even a handful of accidents erodes public trust.

As for bureaucratic red tape, Mark Rutte, secretary general of NATO, acknowledged that his institution can be “too slow” to innovate. Often, NATO would like to see a technology solution work nearly perfectly before rolling it out. But Rutte said that is a mistake: “Speed is of the essence.”  

Salesforce CEO Marc Benioff, who was on the same panel as Amodei, observed that the crop of CEOs in Davos could be the last ones managing all-human workers, since AI agents are coming to the workplace. Fairly soon, managers will be working with human and digital workers. Salesforce itself offers Agentforce, which lets clients build and customize AI agents.

Google President and CIO Ruth Porat, who was another panelist, said Google is already looking ahead to things like quantum computing. Last December, Google unveiled Willow, a quantum chip that can perform a standard benchmark computation in under five minutes that it said would have taken one of today’s fastest supercomputers 10 septillion, or 10 followed by 24 zeros, years to solve. “We’re already investing in where the world will be next,” Porat said.

Looking ahead, Amodei said he is concerned about AI’s impact on political systems. Will it empower more democracy or more autocracy? “This is one of my deepest worries,” Amodei confessed. “Is AI stabilizing for autocracies and destabilizing for democracies?”

He wondered what would happen if China deploys 10 million super-smart AI agents. “What could they do in terms of a surveillance state? The power of dictatorships has traditionally been constrained by the need to have humans who carry out the will of the dictator. That has limited how terrible a dictatorship can be.

“There is a chilling possibility that AI could remove some of those limits and make possible something like a ‘1984’ (the novel about a dystopian society) or darker on the international stage,” Amodei warned. “This is very serious, and I’m not sure it’s going to go right.”

Google’s Porat said that’s why it’s important for the West to stay ahead in AI development. This requires a pro-innovation regulatory environment for AI. Currently, she said the U.S. is ahead in AI models and chips advancements by about a year. But this could change. “We need to be bold and responsible, but we need to make sure we have the ability to maintain the lead.”

Porat also said U.S. companies must be engaged with foreign countries in their AI development. She said when she talks with heads of state, they all express interest in enabling AI in their countries. “But they’re also very clear that in the absence of us being there … they will partner elsewhere.”

The post Anthropic CEO Sees AI-Powered Advances Doubling Human Lifespans appeared first on PYMNTS.com.