Jeff Berkowitz’s Post

It was a privilege to sit down with Connecticut State Senator James Maroney for one of the most grounded and forward-looking conversations I’ve had on how AI will transform our work in government affairs. While Sen. Maroney is well known as a major voice in tech policy, at State Government Affairs Council (SGAC)’s Leaders’ Policy Conference yesterday, we shifted the discussion to how he and other legislators are adopting AI and how the technology is evolving to actually support the realities of policymaking. A few insights that stood out: 🔹 𝗔𝗜’𝘀 𝘂𝘀𝗲𝗿 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲 𝗶𝘀 𝗺𝗼𝘃𝗶𝗻𝗴 𝗯𝗲𝘆𝗼𝗻𝗱 𝘁𝗵𝗲 𝗰𝗵𝗮𝘁𝗯𝗼𝘁. In my presentation, I walked through how t𝘩𝘦 𝘯𝘦𝘹𝘵 𝘸𝘢𝘷𝘦 𝘰𝘧 𝘈𝘐 𝘪𝘴 𝘴𝘩𝘪𝘧𝘵𝘪𝘯𝘨 𝘧𝘳𝘰𝘮 “𝘢𝘴𝘬 𝘢 𝘤𝘩𝘢𝘵𝘣𝘰𝘵 𝘢 𝘲𝘶𝘦𝘴𝘵𝘪𝘰𝘯” 𝘵𝘰 𝘢𝘨𝘦𝘯𝘵𝘪𝘤 𝘴𝘺𝘴𝘵𝘦𝘮𝘴 𝘵𝘩𝘢𝘵 𝘤𝘢𝘯 𝘱𝘭𝘢𝘯, 𝘤𝘰𝘰𝘳𝘥𝘪𝘯𝘢𝘵𝘦 𝘵𝘢𝘴𝘬𝘴, 𝘢𝘯𝘥 𝘸𝘰𝘳𝘬 𝘢𝘭𝘰𝘯𝘨𝘴𝘪𝘥𝘦 𝘩𝘶𝘮𝘢𝘯𝘴 to prepare hearings, analyze legislation, triage constituent concerns, and monitor policy developments across jurisdictions. The future isn’t a blank input box. It’s workflows designed around how humans actually do the work. Senator Maroney shared how valuable it is, even in this early stage of AI’s development: legislator don’t just need answers; they need systems that help them manage the work of serving their constituents more effectively. 🔹 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗺𝗮𝘆 𝘀𝗼𝗼𝗻 ‘𝗷𝗼𝗶𝗻 𝘁𝗵𝗲 𝗼𝗿𝗴 𝗰𝗵𝗮𝗿𝘁,’ 𝗯𝘂𝘁 𝘁𝗵𝗲𝘆’𝗹𝗹 𝘀𝘁𝗶𝗹𝗹 𝗿𝗲𝗾𝘂𝗶𝗿𝗲 𝗵𝘂𝗺𝗮𝗻 𝘀𝘂𝗽𝗲𝗿𝘃𝗶𝘀𝗶𝗼𝗻. Senator Maroney was intrigued by the idea that legislative offices could eventually have AI agents assisting staff, coordinating tasks, and handling repeatable workflows. But he was clear that 𝘫𝘶𝘴𝘵 𝘭𝘪𝘬𝘦 𝘵𝘩𝘦 𝘵𝘦𝘢𝘮 𝘮𝘦𝘮𝘣𝘦𝘳𝘴 𝘰𝘧 𝘵𝘰𝘥𝘢𝘺, 𝘵𝘩𝘦𝘴𝘦 𝘴𝘺𝘴𝘵𝘦𝘮𝘴 𝘸𝘪𝘭𝘭 𝘴𝘵𝘪𝘭𝘭 𝘯𝘦𝘦𝘥 𝘢 𝘩𝘶𝘮𝘢𝘯 𝘴𝘶𝘱𝘦𝘳𝘷𝘪𝘴𝘰𝘳 to review, guide, and ultimately take responsibility for decisions. That human-in-the-loop model is where the opportunity — and the accountability — will sit. 🔹 𝗪𝗲 𝘀𝗰𝗿𝘂𝘁𝗶𝗻𝗶𝘇𝗲 𝗔𝗜’𝘀 𝗮𝗰𝗰𝘂𝗿𝗮𝗰𝘆, 𝗯𝘂𝘁 𝘄𝗲 𝗿𝗮𝗿𝗲𝗹𝘆 𝘀𝗰𝗿𝘂𝘁𝗶𝗻𝗶𝘇𝗲 𝗵𝘂𝗺𝗮𝗻 𝗲𝗿𝗿𝗼𝗿. When we discussed a case study of Maryland finding its AI constituent response agent to be 97% accurate, one legislator in the audience asked the natural question: 𝘸𝘩𝘢𝘵 𝘢𝘣𝘰𝘶𝘵 𝘵𝘩𝘦 3%? Our discussion highlighted a truth we often overlook: we don’t often know the error rate for human staff today. 𝘜𝘯𝘭𝘪𝘬𝘦 𝘩𝘶𝘮𝘢𝘯𝘴, 𝘵𝘩𝘰𝘶𝘨𝘩, 𝘈𝘐 𝘴𝘺𝘴𝘵𝘦𝘮𝘴 𝘤𝘢𝘯 𝘱𝘳𝘰𝘷𝘪𝘥𝘦 𝘵𝘳𝘢𝘯𝘴𝘱𝘢𝘳𝘦𝘯𝘤𝘺 𝘵𝘩𝘢𝘵 𝘮𝘢𝘬𝘦𝘴 𝘵𝘩𝘦𝘪𝘳 𝘮𝘪𝘴𝘴𝘦𝘴 𝘮𝘰𝘳𝘦 𝘷𝘪𝘴𝘪𝘣𝘭𝘦 𝘢𝘯𝘥 𝘧𝘪𝘹𝘢𝘣𝘭𝘦. That’s a powerful shift for public-facing government work. The big takeaway: for both government and those who have to navigate it, AI is becoming an operating layer that can help us stay ahead of complexity, not get buried under it. If you want to see what that could look like for your organization, reach out!