78 Million Jobs. Wrong Question.

The World Economic Forum says AI will create 170 million jobs by 2030 and destroy 92 million. A net gain of 78 million. The number has been repeated in every boardroom, op-ed, and ministerial speech from Davos onwards, and it has done a remarkable job of calming executives who were getting twitchy about automation. There’s just one problem. The net figure is a statistical fiction. The displaced workers and the created jobs are not the same people, not in the same places, and not anywhere near the same skillsets. The real story isn’t whether AI makes more jobs than it kills. It’s whether there’s a bridge from one to the other. And right now, there isn’t.

A comforting number

The WEF’s Future of Jobs Report 2025 is the most-cited document in the global AI employment debate, and it’s easy to see why. It is carefully researched, surveys over 1,000 global employers, and lands on a headline number that sounds reassuring. Jobs lost: 92 million. Jobs created: 170 million. A net gain of 78 million. The implied message is that we’ll be fine. The market will sort it out.

Read the fine print and the reassurance evaporates. The 92 million displaced are disproportionately clerical workers, cashiers, bank tellers, administrative assistants, and data entry clerks — roles concentrated in the lower half of the income distribution. The 170 million created are AI engineers, data scientists, cybersecurity specialists, and renewable energy technicians. The two groups share neither salary band, skill profile, nor geography. Telling a 52-year-old cashier in Ohio that “the market” has a new job for her as an AI prompt engineer is not policy. It’s a shrug.

The WEF itself concedes this, if you look for it. Buried in the same report is the real statistic worth paying attention to: 59 out of every 100 workers will need to be reskilled or upskilled by 2030. Eleven of those 100 are unlikely to receive any training at all. That’s roughly 120 million people the Forum itself expects to be left behind. You don’t see that number on the conference slides.

The forum’s track record

It’s worth being honest about who’s doing the forecasting. The WEF is an organization built on the premise that global cooperation and technological progress, thoughtfully managed, will generally produce good outcomes. That is a fine organizing principle. It is not a neutral one. Reviews of its past predictions show a consistent bias toward optimism — the “borderless economy” promised in the early 2000s didn’t materialize, and the confident 2016 prediction of mass autonomous vehicle adoption by 2025 turned out to be… aspirational.

None of this makes the current report wrong. But it should make us read the headline number with a skeptic’s eye. The 78 million figure assumes that reskilling systems scale rapidly, that employers invest at levels they have shown no historical appetite for, and that workers successfully pivot across industries, geographies, and skill ceilings. Each of those assumptions, taken alone, is ambitious. Stacked together, they start to look like a wish list.

The evidence on employer behavior is particularly jarring. 63 percent of employers cite the skills gap as the single biggest barrier to transformation. Yet only a third are investing meaningfully in reskilling. And 41 percent of employers globally say they plan to reduce their workforce where AI can automate tasks — while only 24 percent expect AI to create new jobs at their company. The optimism in the macro forecast is not shared at the micro level, which is where the hiring actually happens.

The bridge problem

Every major technological transition in modern history has eventually produced more work than it destroyed. That is the strong historical argument, and it deserves respect. But “eventually” is doing enormous work in that sentence. The time between a job being destroyed and a new one becoming accessible to the person who lost it is where entire careers, communities, and social contracts break.

The last comparable transition — the manufacturing decline of the late 20th century — offers a sobering case study. Economists now broadly agree that the towns hollowed out by that shift never fully recovered. The aggregate statistics said the economy was fine. The aggregate statistics were right. The people were not. Anyone assuming AI will be different needs to explain what, precisely, will be different this time.

There is one quietly interesting possibility. Emerging markets with large, English-speaking, trainable workforces — the Philippines, India, parts of Southeast Africa — may be better positioned to build reskilling bridges than Western economies, simply because their labor markets are more fluid and their BPO sectors already run at scale. If the West can’t reskill its displaced workers fast enough, the work will move to places that can. That’s not a threat to the future of work. It’s the future of work finding somewhere to happen.

What the number hides

The danger with a number like 78 million is that it discharges the obligation to think. It sounds like good news, so policymakers cite it, employers internalize it, and the hard conversation about who actually catches the 120 million people likely to fall through the gap never happens. The WEF’s own commentary warns that without faster, more accessible reskilling channels, the gap between jobs created and workers able to fill them could become the defining fault line of the AI economy. That’s a phrase worth rereading, because it’s doing the hard work the headline number refuses to do.

The net number isn’t wrong. It’s just answering the wrong question. The question isn’t whether the global economy will have enough jobs by 2030. It will. The question is whether the people who need them will be able to reach them — and whether any institution, public or private, is actually building the infrastructure to get them there. Right now, the honest answer is no. The number on the slide says the bridge is already built. The fine print says we haven’t started construction.

The question for your business

Are you investing in the bridge — or just hoping your people make the jump?

Read more thought leadership articles here:

• • The Great Leveling
  • Robots Need Colleagues Too