The bottleneck isn’t creation anymore
“Some Simple Economics of AGI.” This paper boggled my mind this weekend. 🧠 I’m not usually an academic paper reader…but this one strikes home.
In short—it argues that the oncoming issue around AI is not about how much AI can DO. It’s about how much humans can VERIFY.
We’re already living this. It’s easy to fire off a 17 page Claude Memo or create a presentation deck in a few seconds. The bottleneck isn’t the creation anymore. It’s whether we can actually process and validate what’s in it. Agents can execute infinitely. Humans can only verify so much. That gap is widening.
The paper coins the “Hollow Economy”: that super high volume, unverified AI activity actually leads to too much noise and less total value for humanity.
From a people perspective, senior folks will spend their time verifying (and training the AI to replace them…the Codifier’s Curse!) and junior folks are being replaced by AI and aren’t getting the reps they need to become the future senior folks (the “Missing Junior Loop”).
PHEW. SO WHAT NOW?!
My brain is tied up with thoughts and future implications, but here are four quick items that struck home today:
-
Be aggressively anti-slop. Don’t send 17 pages when you can edit to a tight two. Don’t ship confusing decks because it’s fast. AI is amazing, but at its core is a prediction machine and isn’t always right. Review. Edit. Verify.
-
Practice matters. How do more junior folks get those missing reps? They use “synthetic practice” to simulate the role, get feedback loops, and speed up the usual on-the-job learning curve. This has me extra bullish on the AI roleplay and skills simulation we continue to build out in Docebo.
-
Skill verification becomes essential. Sorry for one more Docebo nod, but the new Verification Economy doesn’t care about your resume…it cares about what you can demonstrate and do. That makes skills intelligence like 365Talents (that’s real, verifiable, and continuously updated evidence of capability) absolutely foundational , not a talent perk.
-
Lean into the unmeasurable. Things that can be measured will be easily automated. Slightly more complex items will be verified. What’s left? Connection. Creativity. Humanity. Taste. How can we bring a quality of uniqueness, meaning, and that extra special touch to our work?
Lots more to process…what a wild time to be alive. If you or your robots want to read it, the paper is “Some Simple Economics of AGI.”