AI Isn’t Just Changing Work: It’s Rewriting Who Benefits

We’ve known all along that AI was going to transform the world of work. This week we’ve seen that translate into the harsh reality of mass layoffs in the very industry driving it. Meta and Microsoft have announced significant job cuts as they double down on their investments in technology, with Meta planning to sack around 10% of its workforce while aggressively increasing its spending in machine learning. Meanwhile, over at Snap, 16% of the workforce is being made redundant, with the company openly crediting AI with efficiency and cost savings.

The initial narrative was AI as a support tool, a way to increase productivity. No one wanted to talk about job replacement, and when they did, they wrapped it in nuances. In the meantime, companies have been crunching the numbers. And numbers aren’t interested in reassuring narratives.

What we are now beginning to see is a clear pattern: Big Tech is reducing its workforce while increasing investment in AI. A strategic decision. The time has now come for AI to make a real impact on employment, while in parallel we observe the phenomenon of AI-washing, blaming it for decisions that, in many cases, are simply about margin pressure and investor expectations.

Furthermore, AI is a technology that benefits some workers much more than others.

A recent Financial Times analysis based on a survey of 4,000 workers shows that more than 60% of the highest-paid employees us AI on a daily basis, compared to just 16% of those on lower incomes. The narrative of technological democratization falls apart as soon as it is contrasted with the data: using AI to its fullest potential requires education, abstract skills and familiarity with the technology that are not equally distributed.

The result is an unpleasant reality: AI is not leveling the playing field, but tilting it in favor of some even more. Daron Acemoglu points out in that same analysis, the most likely future is one of greater inequality between labor and capital.

For decades, automation was mainly driven by manual work. This time it is different. What is at stake now is skilled work, but not all skilled work. AI is amplifying the productivity of the most expert workers, while leaving behind, or directly replacing, the most vulnerable or early-stage profiles.

When AI replaces, not augments: a data-driven look at youth employment trends

I’ve been saying for years that AI will improve how we work. At this point in the process however, the data suggests…

medium.com

There is a further effect that almost no one is measuring yet: AI is destroying entry-level positions. Much of the work that used to be done by recent graduates is being absorbed by automated systems. And without entry-level positions, there is no training. And without training, in ten years, there are no qualified professionals.

And this connects with another worrying sign: the adoption of AI is not uniform, even within organizations. It happens among workers with more experience, more context and greater capacity for agency. In other words: AI does not replace human intelligence, it amplifies it… but only for those already in the game.

Adding to this gap between narrative and reality is another problem: regulation is way behind the curve: added to the delays in implementing the European AI Act, companies and governments seem to be moving at completely different paces. Business is making decisions based on AI, workers are beginning to feel its effects, but for most people this is still an issue that will affect the future.

In short, there’s an elephant in the room. Meanwhile, there’s talk of reskilling, adaptation, new opportunities. All of which may come to pass, but today it works more as anesthesia than diagnosis. Because at the same time we are being told about an orderly transition, many employees within companies are already warning of the impact of the AI Moon-race.

For years we’ve being talking about how AI might change work. That discussion is over. The question now is another, much more uncomfortable one: not only how much employment is going to disappear, but who is really going to benefit from the AI productivity surplus.

Because if one thing is beginning to become clear, it’s that this productivity is not being distributed evenly: on the contrary, it is ever-more concentrated. And when a technology increases efficiency at the same time as it deepens inequalities, it ceases to be simply an innovation, and inevitably becomes a political problem.

(En español, aquí)

—

This post was previously published on MEDIUM.COM.

—