If you want your company’s stock to go up, hire wonkier IT people

Stay informed with free updates

Simply sign up to the Artificial intelligence myFT Digest — delivered directly to your inbox.

Before taking all our jobs, AI sure is creating a lot of jobs.

The chart above uses data from Barclays, and it may not tell the whole story, because not all AI jobs are equal.

Adam Lauretig and Ryan Preclaw, of Barclays’ data science team, have been collecting job ad data to determine which companies are hiring AI specialists, rather than creating “digital prophet” type roles.

More than 85 per cent of S&P 500 companies had advertised for AI people by March 2023, they found, so by that measure there was not much to tell them apart. A more useful approach, they reasoned, would be to filter the ads for specifics like natural language processing and computer vision, or for software brands like TensorFlow and Pytorch.

Any company that has a high ratio of specific AI roles relative to general ones has self-identified as a “sophisticated end user”, Barclays says. A basket of these stocks could to give investors exposure to AI productivity benefits in a more subtle way than buying more Nvidia.

The experiment has been running for a year and results are . . . fine? An equal-weighted top quintile of AI sophisticates returned 31.78 per cent since mid-October 2023, versus 30.16 per cent for the S&P 500 Equal Weighted index.

More compellingly, there does seem to be a longer-term correlation between stock performance and the targeted recruitment of wonky IT people:

Quant research like this is more about idea generation than applicable strategies. AI specialist-to-generalist is a crude ratio that consistently rates companies including termite exterminator Rollins (two AI job ads, one wonky) and screw maker Fastenal (ditto) above all of the Magnificent Seven. Meta recently dropped out of the top quintile entirely, in spite of remaining by far the biggest AI recruiter, because it ran so many general ads its specialist-to-generalist ratio fell to just 0.23. Selling Meta while holding Rollins would be an unusual way to gain exposure to bleeding-edge AI development.

Is there a more elegant way to show the data? Possibly, but we’ve gone instead with a clickable colour-coded dot plot with a time slider and log-scale axes. You’re welcome.

What we have here is choose-your-own-adventure research.

Maybe you’re looking to future-proof your real estate investments by buying shares in the sector’s most AI-attuned company? Or maybe you’re looking to short the property developer whose HR department appears to have caught the AI bug? Or maybe you’re an IT recruiter who wants to place clients with a bricks-and-mortar operator that doesn’t seem to be too fussy about its hires?

The answer in all three cases is Equity Residential. How you apply his information is up to you. The algorithms can only do so much.