America’s biggest firms still bang on about artificial intelligence in earnings calls, investor days and shiny events, but one term has vanished from the script.
Running a small-scale AI test used to charm investors because it suggested progress and a foothold in a potentially market-upending technology. Over the past year, “pilot” has started to sound like code for doing nothing useful.
AT&T’s chief data and AI officer, Andy Markus, said: “We are not a pilot company.” “What we’re doing truly is driving value.”
Companies still do smaller-scale testing before committing to full deployments, but they are doing it more sparingly than a year ago. When they do, they dodge the word “pilot” because it now reeks of experiments that never ship.
Worldwide, the word “pilot” in reference to AI appeared in about 18 per cent fewer earnings calls in the fourth quarter of 2025 than in the third quarter, according to AlphaSense.
An infamous 2025 study by the Massachusetts Institute of Technology found that 95 per cent of enterprise AI pilots fail to produce a measurable financial impact, and the phrase “Pilot Purgatory” has given the problem a name.
Markus compares AI pilots to flipping a coin, with roughly a 50/50 chance of success. He said AT&T is only chasing use cases where value is clear from the outset.
If something needs tweaking before going live, Markus avoids calling it a pilot. “Maybe you can call it a ‘development plan to go to production,’ but I wouldn’t call it a pilot because most things actually do work.”
Bristol-Myers Squibb’s chief digital and technology officer, Greg Meyers, said pilots do not represent the company’s AI work. He said the term is wrapped up in “PowerPoint washing” by firms that want AI credit without doing anything meaningful.
Meyers said: “Pilots are a really safe way to say that you’re doing something with AI without actually having to change anything or take any risk,” he said.
He said employees often ignore pilots, making it harder to judge whether the work aligns with real business needs.
Meyers calls the testing “proof of concept”, borrowing language already used in drug development. “To me, a pilot often means, ‘I’m just going to try something with no commitment,’” he added.
