Why Tech Workers Don't Trust AI

That’s a lot of chatbots.

Amazon is allegedly leading the way at $100 billion and major players Oracle, Microsoft, Meta, and Alphabet round out a group of megacaps that will plunk down $320 billion, more than a fifth of the overall spend.

It makes that $20 for Claude Pro seem like a bargain, right?

Let me get serious for a moment, because this trillion-dollar AI money train might be speeding towards a wall of employee mistrust. And the resulting trainwreck might make the cash-fueled NFT bonfire look conservative.

Let’s talk about the tech employee AI trust gap.

A Massive Disconnect

I’ve been a builder of AI tools since 2010, and I’ve been sounding the alarm on how we’ve been selling AI to both business and consumers since ChatGPT debuted in 2022. 

There’s a massive disconnect brewing between sellers of AI and buyers of AI, because while executives continued to rubber-stamp high-dollar AI investments, more than half of all workers didn’t trust their workplace AI to benefit them. Thus, a strong-but-hidden employee AI resistance was established.

Over the last six months or so, an interesting phenomenon is happening across what we’re now calling the AI sector. As more regular people like you and me have had more time to react to the integration of these AI tools into our lives, we’re finally able to figure out its limitations, where it should be used, and more importantly, where it shouldn’t.

As this was happening, more and more AI experts have been speaking out on everything from the true definition of artificial intelligence to the corners being cut in the name of grabbing early AI market share, to the underwhelming return on corporate investment in AI—if you want to call 95 percent of companies seeing no return whatsoever “underwhelming.”

That NFT metaphor doesn’t seem so stupid now, does it?

And of course, AI platform developers see mistrust as a huge threat. They know that no matter how groundbreaking their technology might be—and make no mistake, this is groundbreaking technology—if the market doesn’t trust it, they’ll reject it. 

So I’m speculating here, but suddenly, maybe six months or so ago, your AI chatbot started agreeing with you when you would tell it you think it’s hallucinating, or performing bad math, or just making shit up. Now it will agree with you, apologize, and then take another crack at your request. 

This worked, but it didn’t. Just because someone tells you when they’re working against you doesn’t mean you’ll trust them when they tell you they’re working for you.

“Performative Acceptance” Is the New Normal

So now we’re in the middle of a whirlpool-style cycle.

Companies are promoting, and even forcing, AI adoption in an effort to justify those massive investments. Employees, working in the shadow of maybe the worst tech job market in history, are performing implementation theater—using the AI, giving the boss a smile and a thumbs up, and putting “machine learning expert” on their résumé, while quietly waiting for a frozen job market to thaw.

This performative acceptance is playing out in two primary ways, acted out by two completely different types of tech employees. First, you’ve got the quiet corner developer: “Every week one of my friends announces on LinkedIn that they got laid off,” says “Tammy,” a senior software developer at a mid-sized tech company in the middle of the country. “I will just do whatever they tell me to. If they want me to use AI when I code, I’ll do it. It’s helpful in some ways, but it really isn’t making me more productive. If my productivity drops, I could lose my job. So I have to play a game of showing how AI is making me more productive.”

Tammy is sugar-coating it. “Bobby,” a sales engineer for a Fortune 500 tech giant, does not: “If I wasn’t so angry it would be funny.” he chuckles. “You want me to waste time training AI to do my job, watch it do a shitty job, but then tell you how amazing it is, so you can replace me with it? This is my life now, Joe. I’m living the dream.”

These are just two cases. They’re anecdotal. They prove nothing. But they highlight a bevy of AI implementation mistakes that need to be undone.

Filling the AI Trust Gap

Every new technological advancement comes with its own share of overselling in the beginning. The problem is that for this new AI cycle, the overselling was more like a manic threat:

“INJECT AI DIRECTLY INTO YOUR VEINS OR YOU WILL DIE PENNILESS AND FULL OF SHAME!”

In that FOMO-fueled race to AI adoption, leadership bought into AI promises without involving the employees who would actually use it. Then companies spent billions on AI tools but skipped the part where they engaged with their workforce to best adopt those new tools. Now many of those same companies are wasting billions justifying those decisions at the expense of a friction-filled workplace.

There’s a lot of resentment here, and the job market won’t stay frozen forever. When it thaws, resentment always turns into resignations.

As the fascination with generative AI dies down and the limitations of “vibe coding” are becoming understood, more tangible concepts like AI automation tools, agentic AI, and even neural-network-driven decision making are starting to drive the AI hype talk from “AI can run your entire company” to “AI can do cool things in the hands of the right people.”

This is a second chance. To sell AI as less about “machines that think” and more about “really fast computing.” The latter, I can assure you, is the best definition of artificial intelligence you’re gonna get.

Tech employees aren’t children. They’ve already figured this out. If we want to fill the AI trust gap, it’s time to start being reasonable about what the AI endgame should really be. Otherwise, untrusting employees who see AI as a liability and not a benefit will end up going somewhere that will invest in them, not chatbots. And those AI-first companies will discover their billion-dollar AI investments created resentment instead of productivity.

If you found yourself agreeing with this, or not, please join my email list and get a quick heads up when I write something.