"Did you use AI?" is the wrong question. It's the one every hiring team is stuck on right now, be..."Did you use AI?" is the wrong question. It's the one every hiring team is stuck on right now, be...
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
"Did you use AI?" is the wrong question.
It's the one every hiring team is stuck on right now, because take-homes stopped meaning much the moment a model could do most of one in four minutes. The usual responses: ban AI, proctor the screen, go back to whiteboards. All three try to un-invent a tool those engineers will open on their first day on the job.
So Builders' Echo doesn't ask whether you used AI. It asks whether the submission is actually yours — meaning distinct from everyone else's in the same contest. That's a question you can answer with math instead of a policy.
Two passes on every submission:
Structural. The code gets parsed into an abstract syntax tree and compared on shape — control flow, nesting, how the problem was decomposed — with identifiers, comments, and formatting thrown away. Renaming variables and reflowing whitespace doesn't move the tree. This catches copy-paste with a fresh coat of paint.
Semantic. AST comparison misses the rewrite that keeps the idea and changes the structure. So submissions also get embedded and compared in vector space, which catches "same solution, different syntax."
Both roll into a 0–100 originality score that rides along with the submission, visible to whoever is judging.
The thing I got wrong at first: treating the score as a verdict. It isn't one. Two people reaching for the same obvious library on a scoped brief aren't cheating, they're converging — and on a tight problem, convergence is the expected outcome, not the suspicious one. Boilerplate makes it worse: scaffolding, config, and standard error handling are near-identical by design. So the signal has to be weighted toward the parts of a submission where a real decision got made, and the number has to sit next to the work rather than in front of it. A human still makes the call.
That's the honest version of AI policy in hiring, as far as I can tell. Not surveillance, not an honor system — just checking that the thinking on the page is yours, and letting you use whatever tools you want to get it there.
Curious whether people think a number like this belongs in front of a judge at all, or whether it just anchors them. I go back and forth on it.
If your team is doing tech hiring, give it a try here 👉🏼buildersecho.com
Post image
Post image
Back to feed
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started