I've been noticing something pretty wild happening in tech companies right now. AI coding tools were supposed to free engineers from drudgery, but instead we're seeing this weird new anxiety epidemic where people are more burned out than ever. It's like the promise of "AI saves time" has flipped into a workaholic nightmare for everyone involved.

Here's the thing that got me thinking: executives are suddenly obsessed with measuring productivity in the most literal ways possible. We're talking CTOs coding at 5 a.m. because they want to "reconnect" with the underlying code, CEOs tracking AI tool usage bills down to the minute, and startup founders literally telling employees they're "not working hard enough" if their Claude Code interactions don't hit some arbitrary quota. One CEO at an AI company openly admits he reviews his team's AI bills like a financial audit—the more they spend on AI tools, the "better" they're performing. That's... a lot.

But here's where it gets interesting. There's this massive gap between what executives think is happening and what's actually happening on the ground. Surveys show over 40% of C-level execs believe AI saves them 8+ hours weekly, while 67% of regular employees say AI saves them less than 2 hours or nothing at all. UC Berkeley researchers studying a 200-person org found something even more telling: even with AI handling tons of work, actual working hours are still climbing. People aren't getting freed up—they're getting more anxious about optimizing every single moment.

What's driving this? I think it comes down to how "efficiency" itself got redefined. When you can measure productivity by interaction counts and code generation speed, suddenly every moment your AI isn't running feels like wasted time. That's not efficiency—that's just a different flavor of the workaholic culture that was already eating tech alive. Some engineers are openly talking about "AI fatigue" now—this constant worry that the next breakthrough is just one prompt away, and if you're not grinding on it, you're falling behind.

There's also this "task expansion" problem nobody's really talking about. When non-technical people start using AI to generate code, engineers end up spending hours cleaning up half-baked implementations. It's actually making their workload heavier, not lighter. Plus you get this weird hybrid role situation where the lines between who does what are getting blurry, and suddenly there's more code being written than ever before—but is any of it actually valuable? Intuit reported a 30% productivity bump in code output, but if most of that code ends up being disposable or abandoned, what's the real gain?

The deeper issue is that this efficiency obsession is creating what some researchers are calling "busyware"—minor tweaks nobody asked for, custom dashboards for one person, prototype projects that get abandoned halfway through. Each one seems justified in the moment, but most end up in the graveyard of obsolete code. It's like we've created a system where people feel compelled to build things just to prove they're productive, whether those things matter or not.

I think what we're missing is a real conversation about what should never be built in the first place. That's where actual efficiency lives—not in how fast you can generate code, but in choosing what's worth building at all. The workaholic cycle is just accelerating the wrong things.