Is AI Affecting Employment?

The Real Estate Espresso Podcast

Is AI Affecting Employment?

Spotify Google Podcast Apple Podcast Subscribe via RSS Leave a Review

Is AI Affecting Employment?

Welcome to the Real Estate Espresso Podcast, your morning shot of what’s new in the world of real estate investing. I’m your host, Victor Menasce. On today’s show, we’re taking a look at a paper published last week by Stanford University.

But first, I’d like to let you know about an opportunity for one lucky individual to participate in the creation of the Real Estate Espresso Podcast. This is an unpaid internship position, which will give you the opportunity to learn and to contribute to how guests for the podcast are selected. This work represents a couple of hours’ work a week.

If you or someone you know might be a good fit for this position, we’ll be accepting applications to join our podcast team. Send an email to podcast at victorjm.com and let me know a little bit about yourself and how you think this might be a good fit for you. Again, send an email to podcast at victorjm.com.

On today’s show, we’re taking a second look at the paper published last week by Stanford University. There’s been a lot of news coverage about how AI has the potential to change productivity in business. And nowhere is that more true than in the world of software development. It’s been seen as a bellwether for what’s to come elsewhere in the economy.

Software is ideal for tools like artificial intelligence in that it’s a much more restricted language than natural language with a much smaller vocabulary. It’s at the intersection of math and language, so it’s much easier for large-language models to work on.

The Stanford paper was widely quoted in the media and it tells us a few things about the state of AI and how it’s affecting hiring. The discourse spans from utopian predictions of enhanced productivity to dystopian fears of mass unemployment.

Rather than engaging in speculation, the paper, called Canaries in the Coal Mine, cuts through the noise with an unprecedented, data-driven analysis. They use data from payroll company ADP, the largest in the United States. The authors, from Stanford University, systematically present six clear facts that characterize some recent shifts. This is not speculation, it’s using real data and drawing conclusions from that data.

The findings indicate a significant and disproportionate impact of AI, particularly on one specific and vulnerable demographic, and that’s…

The first and most striking fact is the substantial and measurable decline in employment for workers aged 22 to 25, in occupations deemed the most exposed to generative AI. That group has experienced a 13 percent decline in employment since late 2022. This period coincided with widespread adoption of tools like ChatGPT. To give you some context, think of professions like software developers and customer service representatives.

For these young professionals, the downward trend is pronounced. Conversely, more experienced workers in the same occupations, as well as workers of all ages in less exposed fields like nursing aides, have seen their employment remain stable or continue to grow. This divergence suggests a clear, targeted effect of the new technology.

Number two. Looking at the broader labor market, while overall employment in the economy continues robust growth, in line with low national unemployment, the employment trajectory for young workers has been conspicuously stagnant. The decline in hiring for those aged 22 to 25 is directly linked to the decline seen in jobs that are exposed to Artificial Intelligence. It’s a critical insight. The AI revolution isn’t causing total employment collapse. It’s driving a stagnation of a specific segment of the workforce.

For most AI-exposed occupations, young workers experienced a 6 percent decline while older workers saw an increase of 6 to 9 percent.

Number three. It introduces a crucial distinction between AI that automates versus AI that augments. It’s not all about AI itself but how it’s being used. The data clearly shows employment declines are concentrated in occupations where AI is used to automate work, essentially substituting for human labor.

The authors differentiate their uses by analyzing millions of queries to the LLM Cloud. The Cloud is the front runner when it comes to software development.

In contrast, occupations where AI is used to augment or compliment human work through task iteration, learning or validation have not experienced the same pattern of declining entry-level employment. This finding is consistent with a rigorous check of the data by accounting for potential confounding factors.

A plausible alternative explanation for these trends could be that they are driven by firm or industry-level shocks. By controlling for time effects, the authors demonstrate that the observed decline in young AI-exposed workers is not merely a symptom of a bigger, underlying contemporary event. The statistically significant decline remains, a finding that implies the employment trends are a genuine AI-related phenomenon, not a side effect of a differential shock to firms that hire these workers.

Number five. This highlights a fascinating aspect of the labor market adjustment. It’s visible in employment more than in compensation. The study finds little to no difference in annual salary trends across age groups or exposure levels. This suggests a phenomenon of wage stickiness in the short run.

Technology might reduce the demand for labor in certain tasks, but the price of that labor doesn’t immediately drop. Instead, firms adjust by reducing the number of people they hire for those roles, leaving employment as the primary margin for adjustment.

Number six is a confirmation of the robustness of these findings. The patterns hold across different alternative sample constructions and robustness checks. The trends are not driven solely by computer occupations or the information sector. They also are not the result of remote work disruptions. The same patterns are observed in non-teleworking occupations with high AI-exposures, like being a bank teller or tax preparation.

The paper further confirms these trends are a post-2022 phenomenon, not a continuation of pre-existing patterns, and they are visible in occupations with both high and low college graduate concentrations. The use of large-scale, close-to-real-time ADP data with millions of workers provides a far more reliable picture than public survey results like those from our census bureau, which suffer from small sample sizes and high volatility.

So, this particular paper shows young workers will continue to face an uphill battle when it comes to finding work. It’s because those entry-level positions are the ones that are, in fact, the most easy to automate. Those most experienced workers are using AI to augment their work, and their work is displacing the work that junior employees used to do in the past.

Now, intuitively, these findings actually make sense to me. I used to be a leader in the tech industry and have hired hundreds of software development and hardware development engineers. When I think about the work done by the most junior employees, those are, in fact, the tasks that are most easily automated, especially in the world of AI. So it makes sense that these people will be needed in fewer numbers.

Now we know, as real estate investors, that employment drives demand for housing. It affects household formation. Not only has it been more difficult for young adults to afford their first home, but it’s also now more difficult for young adults to even find their first job in their chosen field.

As you think about that, have an awesome rest of your day. Go make some great things happen. We’ll talk to you again tomorrow.

Stay connected and discover more about my work in real estate and by visiting and following me on various platforms: