Knowing Is Half the Battle
What a week of licks, beef, Mikey, and rewinding adds up to, and what to do about it.
If you watched Saturday morning cartoons in the 1980s, you remember how G.I. Joe ended. A kid is about to do something unwise, like getting into a stranger’s car or playing with matches. A Joe shows up just in time and explains the danger. The kid says, “Now I know!” And the Joe answers with a line a whole generation can still finish: “And knowing is half the battle.”
This week was the knowing. This piece is about the other half.
Now you know
Four posts, one argument:
In How Many Licks Does It Take?, efficiency isn’t an outcome. A dashboard built from usage data can tell you how much you produced and at what cost. It can’t tell you whether any of it mattered.
In Where’s the Beef?, vendors are building that blind spot into their products. Anthropic’s smart reports fill the outcome slot with outputs, and they’re designed to support renewals.
In Mikey Likes It, outcomes live with the people the work is for. When researchers measure outside the tool, AI’s gains are real, but smaller and more uneven than they feel.
In Be Kind, Rewind, the way out is to start with the decision you need to make and work backward to the evidence.
Why knowing isn’t enough
The psychologist Laurie Santos and the philosopher Tamar Gendler have a name for the belief that awareness solves a problem: the G.I. Joe fallacy. Knowing about a bias, they argue, rarely stops us from acting on it.
Measurement works the same way. You can understand perfectly well that efficiency isn’t an outcome and still end up managing to the efficiency dashboard, because it’s the one in front of you. Three forces keep pulling in that direction.
The first is cost. Measurement inside the tool is cheap and automatic. Outcomes take time, relationships, and the patience to wait for change to show up.
The second is that saved time gets spent. In 1865, the economist William Stanley Jevons noticed that as coal engines grew more efficient, Britain burned more coal, not less. Efficiency made coal cheaper to use, so people used more of it. AI time savings tend to follow the same path. When proposals get cheaper to write, organizations write more of them. Unless someone decides otherwise, efficiency becomes volume.
The third is incentives. Vendors need renewals. Boards like a clean dashboard. Funder templates ask for counts. Nobody in that chain is rewarded for reporting that a use of AI didn’t help.
The other half
Knowing gets you to the starting line. Here’s what doing looks like, depending on where you sit.
If you use AI in your work, write one sentence about what your use is for. Not what it produces: what it’s for. Then name one outcome you’d expect to see if it’s working, and one stand-in you can watch until that outcome has time to show up.
If you lead an organization, decide where saved time goes before it gets spent. If the hours AI saves should go to clients, relationships, or rest, say so out loud, or they’ll quietly become more output. And when a vendor report arrives, read it for what’s missing.
If you fund nonprofits, resist asking grantees for AI efficiency metrics. Asking for tokens, sessions, or time saved will teach them to count licks. Ask what changed for the people they serve, and whether AI had anything to do with it.
This week started with my partner’s dashboard, so it seems right to end there. The clearest outcome in this whole week of writing never showed up on a dashboard. It was my partner’s learning, compounding the way it has since 2022. Nobody counted it, and it was still the most valuable thing in the room.
Now you know. And knowing is half the battle. The other half is what you do next.
Catch up on the week:
Monday: How Many Licks Does It Take?
Tuesday: Where’s the Beef?
Wednesday: Mikey Likes It
Thursday: Be Kind, Rewind
*If you aren’t a US Gen Xer, I recommend searching the above. You may be entertained.
Anthralytic helps organizations connect strategy, evidence, and responsible AI adoption to the outcomes that matter. Human-Led. AI-Powered. Impact-Driven.

Discussion about this post