The chief executive of one of the world’s leading artificial intelligence labs is warning that the technology could cause a dramatic spike in unemployment in the very near future. He says policymakers and corporate leaders aren’t ready for it.
“AI is starting to get better than humans at almost all intellectual tasks, and we’re going to collectively, as a society, grapple with it,” Anthropic CEO Dario Amodei told CNN’s Anderson Cooper in an interview on Thursday. “AI is going to get better at what everyone does, including what I do, including what other CEOs do.”
Amodei believes the AI tools that Anthropic and other companies are racing to build could eliminate half of entry-level, white-collar jobs and spike unemployment to as much as 20% in the next one to five years, he told Axios on Wednesday. That could mean the US unemployment rate growing fivefold in just a few years; the last time it neared that rate was briefly at the height of the Covid-19 pandemic.
It’s not the first dire warning about how rapidly advancing AI could upend the economy in the coming years. Academics and economists have also cautioned that AI could replace some jobs or tasks in the coming years, with varying degrees of seriousness. Earlier this year, a World Economic Forum survey showed that 41% of employers plan to downsize their workforce because of AI automation by 2030.
But Amodei’s prediction is notable because it’s coming from one of the industry’s top leaders and because of the scale of disruption it foretells. It also comes as Anthropic is now selling AI technology on the promise that it can work nearly the length of a typical human workday.
The historical narrative about how tech advancement works is that technology would automate lower-paying, lower-skilled jobs, and the displaced human workers can be trained to take more lucrative positions. However, if Amodei is correct, AI could wipe out more specialized white-collar roles that may have required years of expensive training and education — and those workers may not be so easily retrained for equal or higher-paying jobs.
Amodei suggested that lawmakers may even need to consider levying a tax on AI companies.
“If AI creates huge total wealth, a lot of that will, by default, go to the AI companies and less to ordinary people,” he said. “So, you know, it’s definitely not in my economic interest to say that, but I think this is something we should consider and I think it shouldn’t be a partisan thing.”
Researchers and economists have forecast that professionals from paralegals and payroll clerks to financial advisers and coders could see their jobs dramatically change – if not eliminated entirely – in the coming years thanks to AI. Meta CEO Mark Zuckerberg said last month that he expects AI to write half the company’s code within the next year; Microsoft CEO Satya Nadella said as much as 30% of his company’s code is currently being written by AI.
Amodei told CNN that Anthropic tracks how many people say they use its AI models to augment human jobs versus to entirely automate human jobs. Currently, he said, it’s about 60% of people using AI for augmentation and 40% for automation, but that the latter is growing.
Last week, the company released a new AI model that it says can work independently for almost seven hours in a row, taking on more complex tasks with less human oversight.
Amodei says most people don’t realize just how quickly AI is advancing, but he advises “ordinary citizens” to “learn to use AI.”
“People have adapted to past technological changes,” Amodei said. “But everyone I’ve talked to has said this technological change looks different, it looks faster, it looks harder to adapt to, it’s broader. The pace of progress keeps catching people off guard.”
Estimates about just how quickly AI models are improving vary widely. And some skeptics have predicted that as big AI companies run out of high-quality, publicly available data to train their models on, after having already gobbled up much of the internet, the rate of change in the industry may slow.
Some who study the technology also say it’s more likely that AI will automate certain tasks, rather than entire jobs, giving human workers more time to do complex tasks that computers aren’t good at yet.
But regardless of where they fall on the prediction scale, most experts agree that it is time for the world to start planning for the economic impacts of AI.
“People sometimes comfort themselves (by) saying, ‘Oh, but the economy always creates new jobs,’” University of Virginia business and economics professor Anton Korinek said in an email. “That’s true historically, but unlike in the past, intelligent machines will be able to do the new jobs as well, and probably learn them faster than us humans.”
Amodei said he also believes that AI will have positive impacts, such as curing disease. “I wouldn’t be building this technology if I didn’t think that it could make the world better,” he said.
For the CEO, making this warning now could serve, in some ways, to boost his reputation as a responsible leader in the space. The top AI labs are competing not only to have the most powerful models, but also be perceived as the most trustworthy stewards of the tech transformation, amid growing questions from lawmakers and the public about the technology’s efficacy and implications.
“Amodei’s message is not just about warning the public. It’s part truth-telling, part reputation management, part market positioning, and part policy influence,” tech futurist and Futuremade CEO Tracey Follows told CNN in an email. “If he makes the claim that this will cause 20% unemployment over the next five years, and no-one stops or impedes the ongoing development of this model … then Anthropic cannot be to blame in the future — they warned people.”
Amodei told Cooper that he’s “raising the alarm” because other AI leaders “haven’t as much and I think someone needs to say it and to be clear.”
“I don’t think we can stop this bus,” Amodei said. “From the position that I’m in, I can maybe hope to do a little to steer the technology in a direction where we become aware of the harms, we address the harms, and we’re still able to achieve the benefits.”
Read the full article here