We recently asked AI to rank which jobs would be taken by AI first. Now, Anthropic, the company behind Claude is trying to answer the question we are all asking: what happens if AI keeps getting better?
The company's new economic model explores what happens the more AI grows, specifically, how it could affect jobs, wages and the U.S. economy through 2030. Some of the results are striking.
Under one set of assumptions, the economy grows significantly while wages for "knowledge workers" barely move. The phrase was first coined by management theorist Peter Drucker in 1959 to describe a growing shift in the global economy away from traditional manual labor and toward intellectual and information-based work. A knowledge worker is essentially a professional whose main capital is their expertise, critical thinking and specialized knowledge.
When those assumptions were pushed further, the divide becomes even more dramatic, with the economy growing 32.4% larger than it otherwise would have while knowledge-worker wages fall more than 10%.
To be clear, Anthropic isn't predicting either outcome. The company is just trying to show how factors including AI adoption, automation and workers' ability to change careers could determine who actually benefits if AI delivers the productivity boom its developers have been promising.
Anthropic modeled three different AI futures
We've also covered the future of AI and what the world will look like in 2035. Anthropic's model lets researchers adjust assumptions including how capable AI becomes, how widely businesses adopt it, how autonomously it can work, how much more productive it makes people and how easily displaced workers can find new occupations.
The company then uses three sets of assumptions to illustrate modest, substantial and extreme AI scenarios.
The modest version looks more like previous technological shifts. AI improves productivity and the economy grows, but the effects remain largely within historical norms.
Things look considerably different in the substantial scenario, where AI becomes capable of performing half of all knowledge-work tasks by 2030
Things look considerably different in the substantial scenario, where AI becomes capable of performing half of all knowledge-work tasks by 2030 and can perform most of those tasks autonomously.
That doesn't mean AI is doing half of all knowledge work. Businesses don't adopt the technology everywhere, so people still perform the majority of knowledge-work tasks across the economy.
Even with that limitation, Anthropic's model has economic growth running at roughly twice its normal rate.
The economy could boom without your paycheck doing the same
In Anthropic's substantial scenario, U.S. GDP reaches $36.3 trillion by 2030, 8.3% higher than it would have been without AI.
Knowledge-worker wages, meanwhile, are essentially flat. Part of the reason is that Anthropic doesn't treat a job as one thing that AI either can or can't replace. Instead, it breaks occupations into individual tasks, some of which AI can augment, some it can automate and others that still require a person.
Consider nursing, an example Anthropic highlights in the model. AI might be able to draft discharge instructions, help monitor patients or organize a care schedule. Other administrative tasks could eventually be automated altogether.
But an AI chatbot can't bathe a patient. Therefore, demand for people performing some types of digital work can fall while the productivity created by AI generates more demand elsewhere.
Anthropic uses construction as another example. If AI dramatically reduces the time required to design and permit infrastructure, more projects could become viable. Those projects still need people to build them, potentially increasing demand and wages for construction workers.
Coders could end up becoming electricians
The calculation changes when AI can automate enough of an occupation that employers simply don't need as many people doing it.
Anthropic uses coders and call-center workers as examples of people who might eventually move into less AI-exposed occupations such as nursing or electrical work.
How easily people can make those transitions turns out to be one of the most important variables in Anthropic's model. If AI eliminates work faster than displaced employees can move elsewhere, unemployment rises, which is also one of the reasons the more extreme results should be reviewed cautiously.
What happens when Anthropic turns the assumptions way up?
Anthropic's extreme scenario is a stress test of what could happen if several aggressive assumptions about AI all come true.
In this version, AI becomes more productive than humans at the vast majority of knowledge-work tasks, performs nearly all of them autonomously and doesn't create enough new types of knowledge work to replace what disappears.
Anthropic says reaching that point would likely require recursively self-improving AI alongside extraordinarily fast adoption. Under those assumptions, the numbers become enormous. By 2030, U.S. GDP reaches $44.4 trillion, making the economy 32.4% larger than it would have been without AI.
Knowledge workers don't experience a corresponding windfall. Their wages fall by more than 10%, while unemployment rises beyond levels normally associated with recessions. This scenario reveals a lot about the relationship between economic growth and the people generating it.
Most Americans aren't expecting the extreme version
As mentioned in the scenario site, Anthropic also asked 10,980 Americans in August what they expect AI to be capable of, how quickly they think businesses will adopt it and how difficult it will be for displaced workers to find other jobs.
When those answers are fed into the model, the typical respondent lands much closer to Anthropic's substantial scenario than its extreme one.
Those expectations produce an economy roughly 10% larger by 2030 than it would have been without AI, with unemployment around 5%. Around 10% of respondents gave answers consistent with Anthropic's extreme scenario.
Even that middle ground would represent a significant change in how millions of people work.
AI capable of performing half of all knowledge-work tasks by 2030 could change which careers companies are hiring for, which skills command higher wages and whether some of today's most sought-after white-collar jobs remain as valuable as they are now.
What this means for your job
There are plenty of reasons to be cautious about taking any of these numbers literally.
Anthropic's model doesn't account for government responses, business cycles, financial-market disruptions, changes in consumer demand or catastrophic AI risks. It also doesn't model a future in which highly capable robots can perform much of today's physical work, which could dramatically change the outlook for occupations that appear relatively protected here.
And again, these are scenarios, not Anthropic's predictions for 2030. What they do show is that the future of work may be more complicated than the familiar question of whether AI will take your job.
AI could make some workers far more productive, reduce demand for others and create new demand in occupations that have little to do with sitting behind a computer. At the same time, the economy could grow much faster without those gains showing up proportionally in workers' paychecks.
Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds. Subscribe to Tom's Guide on YouTube and follow us on TikTok.
More from Tom's Guide
- AI is turning ordinary photos of kids into sexual abuse imagery — here's how to protect your family
- Meta just launched an AI that can actually run your life online — here's what Muse can do
- Gemini just got a 'floating bubble' multitasking upgrade that's a game changer on Android phones — here's how to use it
