## Notes from 04 May 2025 [[2025-05-03|← Previous note]] ┃ [[2025-05-05|Next note →]] I saw [Talita Elizeu]() share the [[Good Government Organizations#Govern for America|Govern for America]] [report](https://www.govforamerica.org/ai-tech-talent) with PIT-UN ([The Public Interest Technology University Network](https://pit-un.org/)) and [Center for Public Sector AI (CPSAI)](https://www.cpsai.org/), and I have to say I love when we look at the subnational level because the variation across states is telling. The study finds that at least 32 US states have created AI task forces, but most remain temporary or volunteer groups rather than permanent units. Talent management emerges as the biggest hurdle: AI is a clear priority, but very few states have dedicated public sector teams. There is no shortage of motivated, early-career tech professionals, but recruiting, integrating, and retaining them remains painfully difficult. The report argues that AI could help alleviate chronic staffing shortages in government's back-office functions, by making employees more efficient before moving on to citizen-facing services. However, these efforts are still in the early stages, pilots fail nearly half the time, and meaningful impact metrics beyond simple time savings remain elusive. As solutions, the report points to fellowship programs like those from Govern for America and [Coding it Forward](https://codingitforward.com/), university partnerships like those promoted by [The Volcker Alliance](https://www.volckeralliance.org/initiatives/government-university-initiative), and a critical role for philanthropy. This research was supported by grants from Microsoft and the [Public Interest Tech Fund](https://www.publicinteresttech.fund/). Take [[AI and Civil Service#^pennsylvania-openia-partnership|Pennsylvania]] for example. It recently published [partial results](https://www.pa.gov/content/dam/copapwp-pagov/en/oa/documents/programs/information-technology/documents/openai-pilot-report-2025.pdf) from its OpenAI partnership, and those early insights are surprisingly promising. Imagine applying lessons like these across Brazil’s states.