Whether training produces measurable gains in everyday work and public service quality is still an open question. The Observatory says it plans to study that next. The report does not yet measure outcomes at that level. The state examples cited alongside it are also not tested against the same methodology. Pennsylvania reported saving an average of 95 minutes per employee when staff used ChatGPT to draft and summarize long documents. New Jersey surveyed its public workforce before finalizing its AI governance framework.
For attorneys advising agencies or public-sector clients, the report's central claim is that buying software or publishing a policy does not change how work gets done. Watch for whether the Observatory's follow-up study measures actual productivity and service quality, not just completion rates. Clients drafting AI use policies should note the report's finding that permission confusion is a bigger barrier than resistance, which makes clear written guidance on permitted use a practical priority. The timing matters too: economists are tracking early productivity gains among private-sector AI users, so the public-sector gap is likely to draw scrutiny from oversight bodies and budget writers.