New Delhi: A line is making the rounds on X again: half of all entry-level lawyers, consultants and finance professionals will be “completely wiped out” within five years. It’s attributed to Anthropic CEO Dario Amodei, and it’s the kind of sentence built to travel.
One caveat first. The version going viral is a compressed summary, and X’s own AI-generated note under it admits it can make mistakes. Amodei’s original warning dates to 2025, when he told Axios that AI could eliminate up to half of entry-level white-collar jobs and push unemployment to 10 to 20% within one to five years. That’s not quite the same as “wiped out,” and the difference matters. But the core claim is still stark enough to deserve a real conversation.
The ladder problem
The part that sticks with me isn’t the 50% figure. It’s what comes after.
Junior work is where people learn. A first-year associate reads a thousand pages of contracts. A junior analyst builds the ugly spreadsheet. It’s boring, and it’s also how someone becomes a senior lawyer or a partner. If AI does that work, who trains the next generation? Companies need seniors in 2035, and seniors have to come from somewhere. Nobody has a convincing answer yet.
India’s fresher problem
For India, this isn’t abstract. Every year lakhs of graduates head into IT services, consulting, accounting and finance, sectors that have long treated mass fresher hiring as the front door to the middle class. Hiring at the big IT firms has already been uneven, and plenty of graduates in cities like Surat know someone who’s been waiting months for a joining date. It would be unfair to blame all of that on AI. Weak global demand and cost-cutting play their part. But if the entry-level rung gets thinner, India has more to lose than most.
Tasks first, professions later
Still, “AI replaces lawyers” is the wrong picture. What gets automated first are tasks: summarizing documents, drafting first-pass reports, cleaning data, running routine analysis. A profession is a bundle of tasks plus judgment, client trust and accountability, and AI is much better at the first part than the rest. The likelier shape is a smaller junior pool doing more with fewer people, not an empty office.
That’s cold comfort if you’re the person who didn’t get the job in the smaller pool.
The awkward contradiction
Then there’s the obvious point. The company building these systems is also the one warning that they could displace millions of jobs. Cynics say it’s marketing: nothing sells a powerful product like the claim that it’s dangerous to the economy. Others say it’s the opposite, a company being unusually honest about its own risks. Both can be partly true. Reports say Anthropic is leaning on its own AI agents for routine technical work too, though I’d treat that with some care until the company says it plainly.
What I’d say is that a warning from someone with a financial stake isn’t automatically false. It just deserves a second look.
Who’s actually at risk?
The more useful divide might not be profession against profession. It may be people who can direct AI tools against people whose work the tools can simply absorb. Amodei’s advice to young workers, as summarized online, is to master these tools, sharpen critical thinking and build interpersonal skills, and learn to steer the technology instead of racing it. It sounds generic, but it’s not wrong. Someone who can check an AI’s output, spot where it’s quietly wrong and explain the result to a nervous client is more valuable than someone who produces the first draft.
Warning or forecast?
So is the 50% figure a warning or a forecast? Probably neither, strictly. It’s a scenario from someone who believes the technology will keep improving fast. Economists have been wrong about automation in both directions: sometimes jobs vanished faster than expected, and sometimes new ones appeared that nobody had imagined. Nobody knows which way this goes.
What we do know is that the first rung of the ladder is the one under the most pressure. Universities, employers and governments have a few years to work out what replaces it. Graduates, meanwhile, shouldn’t panic. They should learn the tools, ask more of themselves than the machine can deliver, and keep watching how this plays out.
