AI Engineers Say They’re Burning Out as Bosses Whiplash From One Desperate Idea to Another

Full Stack

As the AI bubble continues inflating at lightning speed, the people doing the industry's grunt work are feeling the churn.

In interviews with CNBC, AI engineers from giant companies including Amazon, Google, and Microsoft say that they've been under immense pressure to build new products for their employers — except that on many occasions, the features they work tirelessly to deliver are shelved at the finish line.

One Amazon AI engineer, who spoke to CNBC anonymously over concerns of retaliation, said that he was assigned an urgent project on a Friday night that was due Monday morning by 6 am. Despite having out-of-town company, he blew them off to spend the entire weekend working on the project — only to learn later that it had been "deprioritized."

This sort of outcome isn't out of the ordinary at Amazon or at other tech companies, workers told CNBC. These sorts of stop-and-go projects are being assigned and shelved at such breakneck speeds, in fact, that they often go untested, the AI engineers said. And when the tools that do make it to market inevitably break, it's the engineers' job to fix it, often resulting in middle-of-the-night troubleshooting sessions between coworkers.

Across companies, workers say that the general vibe is that their superiors are more interested in satisfying investors and keeping neck-and-neck with competitors than benefitting users. Others said their initial jobs didn't even have to do with AI to begin with, but were switched over from other teams without much training due to high demand.

Trend Chasers

Naturally, this sort of atmosphere doesn't exist in a vacuum. Time and again, we've seen the tech industry pivot from the next big thing to the other, with AI replacing crypto in industry trends that, despite technical restraints and massive warnings of over-hype, CEOs are pouring money into as if it grows on trees.

What's worse, because AI is theoretically supposed to be able to replace human labor, the heads of AI-investing companies are not just having their workers waste time on products that either don't work or won't be deployed, but also having them build their own potential replacements.

The same Amazon engineer who lost his weekend to an ultimately kiboshed project told CNBC that it doesn't appear that higher-ups really care about what their subordinates are doing as long as they "tick a checkbox" — and do it fast.

Translation: AI engineers are doing busywork because the people in their company's C-suites want to give shareholders the illusion of progress, which is a surefire way to accelerate the bubble burst that the industry is careening towards.

More on the AI bubble: AI Startups Are Suddenly In Big Trouble