About That AI Jobs Apocalypse…

If so, you may have (in an act so ironic, the word “irony” isn’t strong enough) turned to AI to help you find a new job. By, for example, writing a cover letter to go with your resume.

That’s exactly the sort of thing an AI chatbot can do really well. Just prompt it with your background info and a list of job postings you want to pursue. The AI will crank out as many cover letters as you need, all customized to make you look like a perfect match.

But it turns out, according to a recent Financial Times report, this doesn’t help anyone. Here’s John Burn-Murdoch, a human who is the FT’s chief data reporter:

“We’ve been blessed with some really great recent research on this one. There was a particularly comprehensive paper out last week from economists Anaïs Galdin at Dartmouth and Jesse Silbert at Princeton, looking at how the ability to use generative AI to write cover letters for jobs on an online freelancing platform affected hiring.

“At first it all sounds quite promising: using AI not only sped up the process of writing applications, they were higher-quality letters too, tailored more to the specific job than those from the pre-LLM period.

“The bad news is everything that follows from this.

“Before LLMs, a good cover letter signaled both candidate quality and commitment. After LLMs it signaled neither. In the pre-AI period people who wrote better cover letters were much more likely to get the job. Afterwards, hiring rates were no better for good pitches than bad ones, and the overall hiring rate fell too. Would-be employers no longer trusted quality applications to be from quality applicants and questioned every pitch they read.”

Another way to look at this: AI creates a “collective action problem” for jobseekers. Using AI to write a superior cover letter makes perfect sense for any individual. When everyone does it, then no one gets the desired benefit.

It hurts employers, too. The study FT cites found hiring rates fell for the most qualified applicants – and rose for the least qualified ones. 

Source: Financial Times

Moreover, AI-assisted jobseekers are blasting employers with more applications, which require more time to review but don’t enhance candidate quality. As Burn-Murdoch concludes, “Good job, everyone!”

Now, this isn’t necessarily good news for human workers. As the AI systems improve, more employers may be able to reduce payrolls, in which case your cover letter’s quality could be irrelevant. But we’re not there yet.

In fact, employers who lay off their human workers are increasingly re-hiring the same people they just fired. That’s not efficient, to say the least.

Visier, a workplace technology platform, examined data covering 2.4 million employees as 142 companies around the world. They found about 5.3% of laid-off employees are later re-hired by their former employer. This number was stable for years but is now ticking up.

Why would that be? Maybe because companies are learning AI isn’t the magic headcount reducer/profit booster they thought it was.

That’s not to say AI won’t help businesses perform better. Many freelancers find it helpful. But integrating AI into a larger company without causing other problems is a complex management problem no one has nailed yet.

To be fair, plenty of people expected the machines to have much greater impact, much sooner than reality is showing. I was one of them. Here’s a story I wrote in 2018.

Source: Mauldin Economics

I cited a World Economic Forum report predicting machines would do more than half of all labor by 2025. And by 2022, it said, some 54% of all workers would need significant retraining just to stay employed.

Obviously, that didn’t happen. And mea culpa on me for believing WEF, which I later learned is a less than reliable source on these things. But at least I didn’t take my own advice. I said this in the 2018 piece:

“Everyone should stop assuming their current role will still exist by 2025. It probably won’t. Identify something else you can do and start preparing for it… now.”

Yet here I am, still in the same role I said would probably disappear. I’m also doing it better, in part because I learn new things every day, and in part because I have new tools like AI to help me.

That word “tools” is key. Nobody wants to be replaced by a machine. But new tools are great.  They make work more enjoyable and workers more productive.

The real question, then, is whether AI will be an amazing new tool – or a new kind of labor, competing against humans instead of making us better.

We’ll find out soon.

See you at the top,