Ivan Zhao did something a few months back that he had never done before. At Notion, the $11 billion business software developer he founded 13 years ago, he hired a high-schooler.
“We had to ask his parents for permission,” he explained. “This guy has no experience working anywhere, but he is so talented, and he just grew up with YouTube, grew up with language models, so he knows how to access information given the tools in front of him.”
The move was a small example, Zhao argued, of what he referred to as a new “abundance approach” unlocked by powerful artificial intelligence (AI) tools. “We’ve become a lot less picky about your capabilities and years of experience,” he explained. “We are almost doing the reverse of what we were doing before.”
Where once Zhao hired mostly mid-career, mid-level workers, Notion, which has more than 1,000 staff, including in London, now targets either very young or senior operators. The former are high on “agency” (the enthusiasm to try things and embrace new tools); the latter often have high taste (a sense of what works and what doesn’t, refined through years of experience). In the age of AI, these are the attributes, argued Zhao, that have risen to the top. Notion’s 16-year-old hire is part-time as he is still studying, but has been helping build some of the company’s latest product features.
Yet Zhao admitted that his expansive approach to hiring may be an outlier. Many of his clients, which range from start-ups to Fortune 500 companies, have taken a different tack: slash and burn. That is because AI systems are making cheap and plentiful what was once rare and valuable: specialised knowledge and expertise. Anyone with a $20 subscription to Claude or ChatGPT can build a website, pull together a financial analysis, draft a spreadsheet and magic up a marketing plan.
So corporate America has embraced the bots, and begun wielding the axe with abandon. Meta this month is letting go of 8,000 people. Coinbase just axed 700 jobs: 14 per cent of its workforce. Amazon laid off 16,000 staff in January, while Morgan Stanley said it would get rid of 2,500 people in March. Last week Cloudflare, the internet security giant, announced 1,100 redundancies — a fifth of its workforce. All said that AI was a driving force.
Two factors have come together to create the trend. One is the emergence in January of OpenClaw, an AI “agent” that allows machines to act autonomously with little or no direction. All the top AI developers have since rolled out their own agent tools.
The other was the release in February and March of updated models from Anthropic and OpenAI that were dramatically more capable than previous versions. At Anthropic, most of the $380 billion (£290 billion) giant’s code is now written by its own agents.
Viewed from the executive suite, those forces have converged to create what looks like a digital workforce. And in most cases, it is far cheaper than hiring bothersome humans, with their holidays and complaints and tea breaks. Each week, another household name announces a giant round of redundancies — a drumbeat that has generated a paranoia radiating across the economy.
Dan Schulman, chief executive of Verizon, America’s biggest mobile carrier, axed 13,000 people in October. He has predicted that US unemployment could surge to an astonishing 30 per cent. “It’s a very difficult time and everyone knows it is,” he said recently. “Like it or not, we live in the age of AI. I happen to like it.” Schulman has announced a $20 million fund to retrain people for a world where what once made them useful is now much less valued.
Beyond the panic-inducing headlines, the reality is more mixed. Unemployment in America remains low at 4.3 per cent. Outside healthcare, however, which accounts for an inordinate amount of hiring, job growth is anaemic. But rather than mass firings, it appears that more companies are sitting on their hands, quietly withdrawing job postings as they figure out how much work can be offloaded to machines.
A survey of 1,000 business leaders by Resume.org, which helps people write their CVs, found that a fifth of companies had already stopped hiring entry-level workers due to AI, and that nearly half planned to do so by next year. Tobi Lutke, founder of Shopify, the $145 billion ecommerce giant, told his 8,100 employees last month that to hire anyone new, managers must clear a new hurdle: prove that a bot can’t do it. And for those already at the company, he ordered them to get good with AI — or else.
“I don’t think it is feasible to opt out of learning the skill of applying AI in your craft. You are welcome to try, but I want to be honest I cannot see this working out today, and definitely not tomorrow,” he added. “Stagnation is almost certain, and stagnation is slow-motion failure. If you’re not climbing, you’re sliding.”
Citigroup, led by Scottish-born Jane Fraser, is training 175,000 staff in AI, while cutting 20,000 roles by the end of this year. The idea, the bank has said, is to encourage “great prompting versus basic prompting” of AI tools, with the aim of generating “impactful results”. That may prove a boon for the Citi, but will provide little comfort to those who get jettisoned in the process.
There is, however, a counterargument against jobs doomerism. David George, a partner at Andreessen Horowitz, the Silicon Valley venture capital firm and one of the largest private funders of AI start-ups, said predictions of mass unemployment are based on the false premise that there is a fixed amount of work to be done.
In this view, the labour market is, in essence, a zero-sum game where every task taken over by a machine means there is one fewer to do for humans. That belief, however, “defies everything we know about people, markets and economics”, George wrote last week. “If history is any guide, we can expect that technological transformation will enlarge the size of the pie. Every ‘dominant economic sector’ has given way to an even larger successor … which in turn has made the economy only that much larger.”
Farm mechanisation replaced millions of workers — but multiplied productivity and freed those people to do other jobs. Electricity powered a golden age of manufacturing that then required more people to sell goods, finance the purchases and service the machines.
Yann LeCun, Meta’s former AI chief, echoed that sentiment. He said last week that predictions of a permanent underclass were “ridiculously stupid”, adding: “There is nothing qualitatively different between the previous technological revolutions and this one. It’s just another set of tools that makes us more efficient.”
What no one argues, however, is that this is a deeply uncomfortable time. Companies have been presented with a powerful, general purpose technology, and every executive is choosing a slightly different way to deploy it. For some, it is simply a cost-saver. In fact, many of the big companies that have blamed AI for large-scale layoffs were struggling for other reasons, from bad strategy to poor management and over-hiring. AI has emerged as a convenient cover story that can not only mask the true reasons for cuts, but has the added benefit of making it seem like a company is leaning into the hottest new thing since the internet.
For others, however, it is an enabler, a superpower that allows them to try things that they previously thought impossible. Zhao said: “As a business leader, you have a choice. Should you just take on more? Do more? Our engineer productivity at Notion has never been this high. The question for us is: can we just do the things now that we didn’t dare to want to do before?”
It is an enlightened approach, but as one scans the horizon of corporate America in these early days of the AI revolution, it appears not a very common one.