#61: Four quick AI-for-impact signals

Impact Signals #61: Four quick AI-for-impact signals A concise approval version of Episode 61: four source-bound items, each reduced to the useful fact and the practical implication. What changed - **Public-sector AI assurance:** Salesforce, Boston Consulting Group, and the Centre for Public Impact report that governments have AI principles, but approval processes are often complex, duplicative, and slow. The practical ask is lifecycle assurance that works inside service delivery. - **Haryana TB mapping:** The Indian Express reports that Haryana is using roughly thirty datasets and historical TB case data to map risk down to a 500-by-500-metre grid, then target active case-finding and AI-aided X-ray screening. - **AI accountability fellowships:** The Pulitzer Center opened its 2026–2027 AI Accountability Fellowships for eight to ten journalists, ten months of work, and up to $25,000 per fellow. - **Post-flood damage assessment:** Applied Geomatics reviewed deep learning and geospatial approaches for post-flood damage assessment, with the main constraint still being verified field data after infrastructure and communications are damaged. Why it matters The common test is simple: does the system name the workflow, user, evidence, limit, and handoff before AI output shapes a decision? Source trail - Salesforce / BCG / Centre for Public Impact — New Global Research Outlines How Governments Can Accelerate AI Implementation — https://salesforce.com/au/news/stories/new-global-resea…