Post #4259454

@Tourma@tech.lgbt Things I would ask knowing "AI Bros" struggle to answer: Who bears legal responsibility when AI provides incorrect information that harms a patron (wrong legal advice, medical guidance, or immigration information)? What is our written protocol for documenting and reporting these errors? What specific patron data will be transmitted to AI vendors, what guarantees do we have that it won't be retained or used for training, and can we independently verify vendor compliance? What specific problem is AI solving that cannot be addressed with existing tools or additional staff? What metrics will determine success, and what would trigger discontinuation if those metrics aren't met? Will staff receive compensated time for AI training, oversight, and correction of AI outputs? How will the library ensure this doesn't become an unfunded mandate that increases workload without additional resources? These force them to commit to specific answers on record. Good luck.