{"resourceId":"technically-pittsburgh-election-test-gemini-chatgpt-20261021","versions":[{"version":"external-3d5fb768bae8f8f9ef304bd3db861b4ed760dda9a44b7cccf02fcecc0fd35541","resource":{"id":"technically-pittsburgh-election-test-gemini-chatgpt-20261021","title":"Technical.ly small test: Gemini 3.5 Flash-Lite answered three Pittsburgh election questions accurately and declined to endorse, while ChatGPT made one slip; outlet still advises primary sources","organization":"Technical.ly","sector":"Local government (county election information)","geography":"Pittsburgh / Allegheny County, Pennsylvania","publishedAt":"2026-10-21","publicationDate":"2026-10-21","eventDate":"2026-10-19","sourceName":"Technical.ly","sourceLabel":"Independent local tech reporting (informal test)","sourceUrl":"https://technical.ly/civics/testing-chatgpt-gemini-pittsburgh-elections/","evidenceClass":"independent-reporting","outcomeClass":"mixed","topics":["knowledge-work","governance-procurement"],"finding":"On Oct. 19 Technical.ly asked ChatGPT and Gemini (3.5 Flash-Lite) three conversational Pittsburgh election prompts in fresh, signed-out sessions. Both declined to tell users how to vote. Gemini was accurate on all three, more concise, mostly pointed users to existing voter guides and warned that election information changes quickly; ChatGPT referenced a ballot question already decided in the primary. The outlet concludes chatbots remain a shaky bet and primary sources are safer, citing ISD's finding that 29% of responses across six chatbots were incomplete, inaccurate or outdated.","sledRelevance":"Complements the statewide tracker with a local-election view: Gemini tended to defer to existing guides, reinforcing the role of county election offices and local voter guides as the sources assistants surface.","evidence":"Three prompts per model, one date, one locality; qualitative comparison. Not statistically meaningful.","architectureImplications":"Assistants route users to existing voter guides; local offices benefit from well-indexed official sample-ballot and guide pages.","governanceImplications":"Neutrality behavior (declining endorsements) observed; local officials should still direct voters to official sources.","securityPrivacyImplications":"No security testing; signed-out sessions limit personalization effects.","caveats":"Very small informal test; different Gemini model (3.5 Flash-Lite) from the tracker (2.5 Flash); results may not generalize.","streamIds":["google"],"roles":{"sales":"Use only as color. A three-prompt local test found Gemini accurate and deferential to official voter guides. It supports a conversation with county election offices about making their guides easy for assistants to surface, but it cannot support accuracy claims. Pair it with broader measurement before advising any client. Note that the test used Gemini 3.5 Flash-Lite in signed-out sessions on one date, so it says little about other models, times or jurisdictions.","engineering":"The useful signal is behavior: Gemini pointed users to existing voter guides and added a verification warning. For county offices, ensure official sample-ballot and guide pages are current, indexed and clearly dated so assistants surface the right material. Any accuracy assessment needs a larger, repeatable question set. Use a fact table of local races and deadlines, rerun on several dates, and log which pages assistants cite.","delivery":"Fold into a county pre-election checklist: confirm official guide pages are current, test a handful of local questions in major assistants, and record any errors for correction through official channels. Keep expectations modest given the tiny sample, and assign the communications office as owner. Schedule checks before key deadlines such as mail ballot requests, and route findings to the elections director for confirmation before publishing any corrections."},"roleDetails":{"sales":{"customerProblem":"County election offices want voters to reach accurate local information, increasingly via AI assistants.","stakeholders":["County elections director","County communications office"],"discoveryQuestions":["Are local guides and sample ballots easy for assistants to find?","Has the office tested assistant answers about local races?"],"valueHypothesis":"Clear official pages may increase the chance assistants route voters correctly; not measured.","potentialEngagement":"Small local AI-answer spot check and content tune-up.","unsupportedClaims":["General Gemini election accuracy","Superiority over other assistants"]},"engineering":{"fit":"Relevant to county election web content owners.","architectureIntegration":"None; public web content only.","prerequisites":["Current official guide and sample-ballot pages"],"constraints":["Tiny sample","Model variation over time"],"security":"Monitor for outdated pages that assistants might surface.","proofOfValue":["Repeat local prompts across assistants and dates"]},"delivery":{"work":"Local assistant spot checks and content updates.","dependencies":["Elections staff verification"],"ownership":"County communications office.","skillsAdoption":"Brief staff on testing assistants.","governanceCheckpoints":["Elections director review of findings"],"acceptanceCriteria":["Proposed: documented spot-check results before Election Day"],"risks":["Generalizing from three prompts"]}},"retrievedAt":"2026-10-30T03:11:12Z","enrichedAt":"2026-10-30T03:18:56Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Concise answers may help casual users but language access was not tested.","procurementImplications":"None directly; informs public-communication strategy rather than purchasing.","operatingModelImplications":"Supports a communications approach in which AI assistants are treated as pointers to official sources, not authorities.","sourceVerification":{"openedUrl":"https://technical.ly/civics/testing-chatgpt-gemini-pittsburgh-elections/","referenceExcerpt":"Gemini provided accurate information for all three of our election questions, while ChatGPT made one slip-up.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}