True or False: Is AI Killing Jobs?

In Brief

A new study from Ramp and Revelio Labs found the heaviest AI adopters grew headcount by about 10%, and it has quickly become a favorite data point for the argument that AI is not killing jobs. 
But there's lots of data pointing the other way, including rising unemployment for recent college graduates and 28,000 tech and finance jobs lost per month. Paul Roetzer says one unanswered question makes the optimism meaningless.

What Happened

A new study from the finance company Ramp and Revelio Labs has become highly cited evidence that AI is not destroying jobs. Ramp used its corporate card and bill-pay data, real records of what companies actually spend money on, to identify which of more than 21,000 U.S. firms are paying for AI, then matched that against their headcount. The firms that invested most heavily in AI grew total headcount by about 10% in the two years after adoption, and grew entry-level employees by 12%. Lighter adopters saw no statistically significant change.

The gains showed up gradually, roughly six to 12 months in, and they spanned departments, including engineering, sales, administration, and customer service. The authors acknowledged one catch: The heavy adopters are a selected group, already larger, more technical, faster-growing, and more likely to be venture-backed.

That optimism runs counter to data pointing the other way. A New York Times analysis laid out how, by several measures, AI already appears to be driving up unemployment among recent college graduates and may have already destroyed tens of thousands of jobs. Stanford economist Erik Brynjolfsson used ADP payroll data to find a roughly 16% drop in employment for workers aged 22 to 25 in the most AI-exposed jobs, roles where AI can already handle a large share of the daily tasks, since ChatGPT launched in 2022. And Bloomberg reports that the tech and finance sectors are now shedding around 28,000 jobs a month.

On Episode 224 of The Artificial Intelligence Show, SmarterX founder and CEO Paul Roetzer examined why the feel-good numbers do not settle anything.

The Key Numbers

21,000+ - U.S. firms in Ramp's corporate spending data

10% - Total headcount growth at heavy AI adopters in the two years after adoption, the Ramp study found

12% - Entry-level headcount growth at those same firms, according to the Ramp study

16% - Employment drop for workers aged 22 to 25 in the most AI-exposed jobs since ChatGPT launched in 2022, according to Ramp

28,000 - Tech and finance jobs lost per month, per Bloomberg

Why the Hiring Numbers Don't Settle the Jobs Debate

Polarized data comes with a tell. The study spread fast on X among the jobs-aren't-going-away camp, where posts often open with the phrase "narrative violation." This signals a study contradicts the prevailing story that AI eliminates jobs. "Normally, if I see narrative violation, it's sort of a tip that they're very polarized in one direction, and I don't really put a lot of weight behind the rest of the tweet," Roetzer says.

Fast-growing firms hiring people proves nothing new. "I will just state what seems to be quite obvious to me, which is fast-growing firms hire people. That has always been true," Roetzer says. The question the research leaves unanswered is whether those firms are hiring at the same rate they would have three years ago. A company growing 40% might add 30 employees today where it once would have added 300. "If we don't have the answer to that question, then this research is basically meaningless," he says.

Startup hires don't outpace enterprise layoffs. A venture-backed startup can now reach a billion-dollar valuation with 100 employees instead of 1,000 or 10,000, and the data will register that as job growth.

Meanwhile, a single enterprise with 50,000 employees that cuts 20% of its workforce over the next three years eliminates 10,000 jobs. He asked how many venture-funded startups it would take to offset that, especially since the people laid off will mostly be the ones who never mastered AI, and AI-native startups only hire AI-savvy people. 

A big-company CEO just described exactly how this plays out. Roetzer pointed to Nikesh Arora, CEO of cybersecurity giant Palo Alto Networks, on a recent episode of 20VC with Harry Stebbings. "The challenge right now is 90% of the enterprise employees are not AI savvy," Arora said. "They have to learn."

Arora called it "a Darwinian moment" and described two paths. Some CEOs, such as Coinbase's Brian Armstrong and Block's Jack Dorsey, decimate the organization and rebuild with 30-40% fewer people, because "there is no redemption. I can't train these people."

Palo Alto Networks is on the gradual path. "We've been hiring people only through hackathons now," Arora said, replacing the roughly 2% of employees who leave on their own each month through natural attrition with AI-savvy hires. That is enough to transform 20-25% of the team in 12 months and reach an AI-savvy workforce in three years. As Roetzer described it, that is a farm system for AI-forward employees.

Either path ends in the same place, one Roetzer has been warning about for years.

"If you are not AI savvy, if you're not as we would call AI forward, you don't have a job in three years. I don't know how else to say that other than to be very blunt. The only people that will be working in enterprises in three to five years will be AI forward professionals and leaders."

— Paul Roetzer, founder and CEO of SmarterX, Episode 224 of The Artificial Intelligence Show

SmarterX Take

Both can be true at the same time. AI-native companies and AI-focused roles are real, well-compensated, and growing, and heavy adopters really are adding people. But those gains do not have to outnumber the jobs lost to massive productivity increases at large enterprises, and there is no evidence yet that they do. The Ramp study measures the winners. It says nothing about the displaced.

For leaders, Arora's gradual model is worth studying closely. Roetzer called it the prototypical version of his business AI transformation pillars: give people time, provide training and education, give them the technology and personalized use cases, and let them self-select into the future. That is the human-centered alternative to decimate-and-rebuild, and it still ends with an entirely AI-savvy workforce in three years. 

What to Watch

New job studies and what they really show. Any new jobs study, optimistic or grim, is only useful if it can say whether companies are hiring at the rate they would have before AI. Until researchers answer that, treat every "AI creates jobs" headline and every "narrative violation" post with the same skepticism.

CEO behavior on headcount. Don't watch layoff announcements but the quieter moves: hiring freezes filled through hackathon-style filters, attrition that never gets backfilled with traditional hires, and executives openly setting timelines for an AI-savvy workforce. The gradual model looks gentler, but it transforms a company just as completely.

What 2,100+ Professionals Expect AI to Do to Jobs

Seventy-one percent of professionals believe AI will eliminate more jobs than it creates over the next three years, according to the 2026 State of AI for Business Report. That belief barely varies by seniority, with CEOs and VPs at 73% and specialists at 64%. Yet only 20% say they are somewhat or very concerned about AI's impact on their own role.

That gap, expecting mass disruption but believing it will happen to someone else, is exactly the blind spot Arora's exposes. The full report, based on 2,100+ responses from professionals across roles, functions, and industries, covers where the workforce actually stands on adoption, literacy, and readiness. Read the full report →