Open Research Series
Research
We publish what we find, including the parts that complicate the argument.
Claims about AI adoption are easy to make and hard to check. So we publish ours the way a reader would need in order to disagree with us properly: the analysis code, the denominators, the robustness checks, a file manifest with SHA-256 checksums, and the strongest counterargument we could construct against our own result.
Survey data belonging to third parties is never redistributed here. Each release documents how to obtain the original data and how to verify that what you obtained matches what we analyzed.
Publications
Open Research Series · v3.1.0
The Developers Using AI Without Trusting It
What 26,102 current users in Stack Overflow’s 2025 survey reveal about adoption, sentiment, and calibrated skepticism
A secondary analysis of 26,102 current AI-tool users finds favorable stance toward AI carries substantially more information about daily use than trust in output accuracy.
Josh Penzell
Replication and corrections
Corrections, replication attempts, and methodological criticism are welcome. Material revisions are documented in the repository release notes and reflected in the versioned research note.