IBM says 40% of workers must reskill for AI

The fear of generative AI taking everyone's job might be slightly off the mark. According to IBM, the real shift is much simpler: the work itself is staying, but nearly half the people doing it will have to reinvent what they do.

In a recent study drawing from surveys of 3,000 executives across 28 countries and 21,000 workers across 22 nations, IBM found that business leaders expect 40% of their workforce will need to reskill in the next three years due to AI implementation.

Applied to the World Bank's tally of the 3.4 billion global workforce, that is roughly 1.4 billion people heading back to the training room.

Soft skills eat the world

The most eye-opening part of IBM's findings is what companies actually care about now.

Back in 2016, STEM proficiency was ranked as the single most critical capability by surveyed executives. In 2023, that same technical proficiency crashed straight to the bottom of the priority list.

Why? Tools like ChatGPT now let everyday workers produce technical output, from basic coding to drafting, with far less underlying domain knowledge.

Instead, people skills have taken the crown. Team management, effective collaboration, communication, and adaptability to change are now the top priorities for employers trying to navigate AI adoption.

The revenue divide

Despite the massive reskilling required, 87% of surveyed executives expect generative AI to augment existing roles rather than outright replace them.

IBM summarized the dynamic with a phrase that has quickly become an industry mantra: AI won't replace people, but people who use AI will replace people who don't.

The business upside looks substantial. IBM reported that tech adopters who successfully retrain their workers see a 15% average revenue growth rate premium over their competition.

For businesses that lean heavily into AI-driven transformation, that revenue growth premium jumps to 36%.

Why it matters

For workers, this study draws a very clear line: sweating over routine technical mechanics is becoming a dead end. What matters right now is learning how to direct these models and, more importantly, how to communicate and coordinate around them. Knowing how to manage projects and adapt quickly is suddenly worth more than remembering syntax.

For businesses, the competitive battleground is shifting from tool procurement to human execution. Buying access to an LLM is easy; retraining thousands of employees to overhaul their daily workflows within 36 months is an operational nightmare.

We think IBM's focus on the decline of STEM prioritization is spot-on, though it introduces a real dilemma. Hard technical skills are being commoditized faster than universities and corporate training programs can adapt. The tools are here, but whether companies can actually retool 1.4 billion people before they fall behind is an open question.