What Happens When AI Job Postings Go Down?

The global labor market is behaving strangely. 

Or at least: it’s behaving in a way we’ve never seen it behave before. After the turbulent years post-pandemic when workers changed jobs at record pace and then future of work conversations about quiet quitting and return-to-office debates dominated national news, we’re opening a new chapter in the history of the labor market. Big-picture patterns reflect this, like in the way Federal interest rates have risen sharply over the past several years but the unemployment rate has remained stubbornly low. As of this morning, JOLTS showed that job openings are holding steady at higher-than average rates and layoffs have declined again, staying well below their pre-pandemic baseline.

This isn’t an environment marked by the reliable norms of years past. We can’t rely on hunches, anecdotes, or conventional wisdom like we used to. When you’re making decisions about the labor market, you can’t afford to commit time and resources based on bad intel, either—you need something you can count on, and that’s what our software, data delivery, and Professional Services all provide. When we understand the labor market better, we can make better decisions about it.

Understanding the labor market also means making it more accessible. At Lightcast, we’ll often say we want to “create a job market that works for everyone,” and so we want to share what we know with more people than just those who understand data science really well. What happens in the labor market happens to all of us. 

So that’s what’s brought us here: On The Job. We’re here to provide a regular roundup of ideas, trends, and research that shape how we at Lightcast see the world of work.

Up first: the technology that’s generating more buzz, more anticipation, and more anxiety in the labor market than any other.

A New Index For AI

A few weeks ago, the Center For Human-Centered AI (HAI) at Stanford University released the 2024 AI Index Annual Report, and Lightcast is one of its Analytics and Research Partners (others include Accenture, GitHub, LinkedIn, GitHub, McKinsey & Company, and the International Federation of Robotics). The report is a huge project, essentially just a big summary of everything that had happened in the field of artificial intelligence that year—and 2023 was a big year for AI! Research covering topics as diverse as model structure, overall productivity, and new legislation and regulation are all covered in the report, alongside job posting data from Lightcast.

But here’s what’s interesting: we actually saw AI job postings, and overall demand for AI skills, go down this year. In 2022, AI-related positions made up 2.0% of all job postings in the US, a figure that decreased to 1.6% in 2023.

And this isn’t what one might expect, given the buzz that AI has enjoyed. So what’s behind the change?

This is a good example of a situation where data can’t think for itself: the lines on the charts can help guide our understanding of AI in the labor market, but they can’t tell the full story. Here’s one potential explanation, which we offered in the report:

“The 2023 decrease in AI job postings is driven by many top AI employers (such as Amazon, Deloitte, Capital One, Randstad, and Elevance Health) scaling back their overall posting counts. Additionally, many companies shifted the occupational mix of their postings. For example, Amazon, in 2023, posted a higher share of operational roles like sales delivery driver, packager, and postal service/mail room worker than in 2022. At the same time, there was a lower share of demand for tech roles like software developers and data scientists.” 

If we limited our scope to only the years 2022 and 2023, we’d conclude that demand for AI in the labor market is dropping fast. But we know that’s not true; investment in AI firms is soaring, and companies like Samsung are reporting strong earnings based on AI investment. 

The full picture shows us that demand for AI jobs and skills is still a steady climb upward. Ironically, this illustrates a core principle about how Lightcast interacts with data and AI: any output must be transparent and explainable. We don’t trust a “black box” algorithm to just spit out numbers, we need to know how it got there. In this case, we know that there have been organizational shifts in the way big employers like Amazon and Deloitte are shifting their hiring strategies because we have human experts who study this kind of thing. They looked into it and found a likely answer.

AI at Lightcast

One reason it’s important we recognize that AI isn’t a passing fad that peaked in 2022 is because we’re continuing to deploy new features and resources that help process the change that’s happening throughout our workplaces. One big one is the AI Skills Sector in Analyst. In the second chart above, we’ve sorted dozens of specific AI-related skills into eight distinct clusters. Identifying those dozens of skills is how we were able to provide this data to Stanford, but it’s useful for all kinds of other purposes (“Which AI skills are most in demand in my region?” for instance). So that list is now available in our software.

The Lightcast x AI page starts with skill clusters and then highlights the other work we’ve done in the space, including our guide to using AI in HR, our core principles for deploying and understanding AI, and research, like the Stanford index, that tracks where AI is now and where it’s going in the future. 

Thanks for reading On The Job. We’re already working on the next issue, which you can expect on May 15, and you can subscribe to receive it in your inbox right here.