The robots are doing the work, but nobody knows how to grade it anymore.

The Summary

The Signal

We're hitting the awkward middle of the AI transition. Half of managers can't figure out how to evaluate work when employees use AI tools. Not because the work is bad. Because the metrics broke. You can't measure hours anymore when the AI writes the first draft in three minutes. You can't measure effort when the hard part is knowing what to ask for, not producing it.

This isn't a problem to solve with better dashboards. It's a recognition problem. Companies optimized for industrial-era performance reviews: time in, output out, manager watches the assembly line. AI collapses that model. The person who prompts well and edits ruthlessly produces more than the person who grinds. But legacy systems can't see the difference.

"The robots are doing the work, but nobody knows how to grade it anymore."

The smart companies see it. Fortune reports that PepsiCo, Indeed, Dropbox, and Bristol Myers Squibb are redesigning how productivity gains get shared. Not just "use AI to do more with the same headcount." They're asking the uncomfortable question: if AI makes someone 3x more productive, does that person get 3x the pay, 3x the free time, or do shareholders pocket the difference?

This is the real fight. Not whether AI works. Whether humans capture the upside or just get speed-run into obsolescence. The companies asking this question now are the ones that might keep talent in 18 months when everyone else is managing a resignation wave.

Key tensions emerging:

  • Productivity is up, but so is manager confusion about what good work looks like
  • Old metrics (hours, effort, busyness) are dead but nothing replaced them yet
  • The gap between AI-native workers and AI-skeptical managers is a legibility crisis, not a skill crisis

The Implication

If you manage people, your job just became translation: between what AI outputs and what the business needs, between new productivity and old evaluation systems. The managers who figure out how to measure judgment, taste, and orchestration instead of hours and output will run the next decade. The ones still looking for butts in seats are managing a ghost workforce.

If you're the employee, this is leverage time. The company that can't measure your value is also the company that can't justify cutting you or underpaying you. Learn to articulate what you do that the AI can't. Not "I work hard." But "I know what to build, what to kill, and what question to ask next." Make that legible or someone else will define your value for you.

Sources

TechCrunch AI | Fortune Tech