According to a recent Washington Post article, some companies are setting expectations for how much employees should use AI. Some have even experimented with usage targets, including tokens consumed.
I get it. Companies have invested heavily in AI, and they want employees to use it. They want to see a return on their investment.
But what does “using AI” mean?
I have some questions.
If I ask ChatGPT to summarize a document I could have read in three minutes, does that count? What if I ask it to rewrite an email six times before deciding the first version was the best? Does this make me six times more productive (or seven, if I count the original)? If I spend two hours making a presentation after my boss literally told me to just give him a short email explaining the situation, is that a sign of my productivity? Mind you, I’m just asking for a friend…
Back when WFH was just an AM station on the radio, I remember a colleague commenting on how another colleague stayed late every night. “She works so hard.” I couldn’t help wondering whether she was working harder or simply taking longer.
Is tracking AI usage the tech equivalent of noticing who’s still at their desk at 7 p.m.? Have we confused spending time with accomplishing something? Let’s compare tools with tools:
Excel: Great news, Tasha. You opened 47 spreadsheets this week. We’re putting you on the leadership track.
Email: Brian sent 136 messages yesterday. Clearly a high performer.
Or Presentations. That deck had 27 slides and 2 appendices. Obviously Janet knows her stuff. (Actually, scratch that. I think we may have actually done that one.)
There’s something especially funny about measuring AI activity when a key promise of AI is that it will help us work more efficiently. If I figure out how to accomplish something in half the time, with fewer prompts, am I doing better or falling behind on my AI engagement goals?
To be fair, some companies have already recognized the potential problem. Meta reportedly moved away from using AI token counts to evaluate employee impact, and Amazon has warned against treating AI usage as a performance target.
And if employees think they’re being judged on how much they use AI, I suspect they'll get remarkably creative about using more AI.
Whether that makes them better at their jobs is another question.
We’ve always been pretty good at finding ways to look productive. AI just gives us a whole new set of options.