What looks finished
The recommendation
Your document is coherent, well-structured, and may even be correct. AI has helped you move faster and produce something useful.
A companion to Undelegatable
Synthetic
competence.
The work is better. What about the capability behind it?
AI can make our work faster, sharper, and more persuasive. That matters. But the quality of what we produce doesn't always reveal what we can explain, question, or adapt when circumstances change. That's the gap worth examining.
Explored by Lyndon Cerejo · Published in California Management Review ↗
Two different questions01 / 02
What I can produce with AI
Polished, persuasive work
Doesn't tell me
What I can still do
Explain, question, and adapt the result
Better output is valuable. It's just not the whole measure of capability.
01 / Understand the idea
When output and capability part ways.
AI can extend what you're able to accomplish. The question is whether you can still exercise the understanding and judgment needed to own the result.
“Synthetic competence is the gap between the capability an AI-assisted deliverable appears to demonstrate and the capability its accountable owner can exercise when the work must be questioned, adapted, or recovered.”
The distinction is ownership, not authorship. You don't need to recreate everything AI helped produce. You do need to recognize what matters, challenge what looks wrong, and respond when the situation changes.
Not a criticism of AI use
Work can be excellent precisely because you used AI well. Synthetic competence is not about failing to do everything unaided. It's about a gap between what the work seems to prove and what you can actually explain, question, and adapt.
02 / Recognize it in your own work
A good result.
An unanswered question.
Imagine you've used AI to turn a complicated set of inputs into a convincing recommendation. It looks good, the logic sounds persuasive, and you're ready to share it.
What remains to be exercised
Your judgment
Someone challenges a key assumption. Can you trace the reasoning, identify what might be missing, and change your recommendation without simply accepting the next AI answer?
That's the distinction. The issue isn't how much AI you used. It's whether the result leaves you able to make the next consequential decision.
03 / Try it on your own work
The Ownership
Check.
Before you share a consequential piece of AI-assisted work, try this on yourself. Don't ask whether you could have made it without AI. Ask whether you can still own what happens next.
01
Where might this fail?
02
If it did, what would I do?
Four moves to take it further
Locate
Can I pinpoint a real vulnerability, not just list generic risks?
Explain
Can I connect that vulnerability to my evidence, assumptions, and choices?
Stress test
If a key condition changed, would I know what changes and why?
Reconstruct
Could I revise the plan, analysis, design, or recommendation deliberately?
AI isn't disqualified from helping you do any of this. The purpose is to keep your judgment engaged and notice where you may need more understanding or practice.
The broader exploration
Not less AI.
More of you in the work.
Synthetic competence is one reason to pay attention to our capabilities. But the bigger question is what we choose to keep thinking, practicing, and owning as AI takes on more of the work.
Undelegatable is the larger project: six human skills worth practicing, and the PAIR approach to using AI without quietly handing over your judgment.
Curiosity · Critical Judgment · Empathy · Advocacy · Orchestration · Shaping
Explore Undelegatable
05 / The broader picture
The work and the person are two different signals.
A useful way to see the distinction is to consider output quality and exercisable capability separately. Neither tells you everything about the other.
Explore the four possible combinations
Output quality × exercisable capability
Synthetic competence
Strong results suggest capability the owner cannot reliably exercise under changed conditions.
Augmented competence
AI improves the work while the person can still challenge, adapt, and recover it.
Visible capability gap
The gap is more apparent because it shows up in the result.
Under-realized capability
The person has more capability than this particular result demonstrates.
This 2×2 is a conceptual heuristic, not a validated assessment instrument or a way to label people.
A note for leaders
Polished performance is not proof of readiness.
If you develop or evaluate others, ask about their reasoning, their response to changed assumptions, and how they're building capability—not only the quality of what they deliver. The Ownership Check can begin that conversation without turning AI use into a compliance exercise.
Read the leadership perspective in California Management Review ↗Read the full exploration