John Burn-Murdoch

Although it might be true that within particular workplaces, AI can kind of level the playing field, across the whole economy, it seems to have the opposite effect.

Sure.

So we worked with Focal Data on this.

We went out and surveyed about 4,000 people across the UK and US, about 2,000 in each country.

And we asked dozens of questions about how people are using AI at work, whether they're using it at all, what they're doing specifically with AI in their jobs, as well as various characteristics of their jobs, including...

what they do, what industry they're in, how much they're paid, that kind of thing.

Yeah, so this is just a fairly straight finding that workers on the highest salaries in the economy are much heavier users of AI than workers lower down the salary distribution.

So among the people who are in the top 10% by income, about two thirds of them are using AI in their jobs.

Whereas for those in the bottom 10%, it's only about 15%.

So we see this consistent gradient where the highest earners, the most highly educated workers, people who've been in their jobs a bit longer are all using AI in their jobs more than those who are at the other end of things.

Now, if we had to take a stab at why this might be the case, what are some working theories?

So one way of unpicking this is we looked at whether that gradient of sort of pay and economic status, as it were, and AI usage, whether that applied within particular sectors and jobs as well as across the whole economy.

And within, say, the tech sector or the white collar sector in general, that was still true.

Like the most highly paid white collar workers are using AI more than the least highly paid white collar workers.

But when we drilled right down to look at, say, accountants, management consultants, lawyers, there, there was actually less of a relationship.

So what it looks like we're seeing is

is that it's people who are in higher paying, more high status jobs, use AI more than those who are in lower paying and low status jobs.

So it's not that you've got two people sitting on desks next to each other and one of them's paid more than the other and uses AI more.

It's that different occupations have different pay patterns and AI usage patterns.

So various things came out of this.