Woman-Run
In this episode of the Woman Run Podcast, the Meredith Lowry interviews Deanna Ray, the Chief Legal Officer for the Little Rock Water Reclamation Authority, to unpack the intersections of local government, vendor negotiations, and the implementation of organizational artificial intelligence policies. Deanna describes her organization as a municipality or government agency that reclaimed used water, treated it, and discharged it back into the Arkansas River, adding that "water wastewater treatment actually is very technology-based." The catalyst for their conversation stems from recent state regulation, specifically Act 848 passed during the 2025 Arkansas legislative session. Deanna explains that under this act, "if you are a government agency that uses any type of AI, then you must have an AI policy." She clarifies that the law defines AI broadly, encompassing everyday features like text autocorrect, spelling tools, and automated note-takers rather than just generative text programs. An essential pillar of Act 848 is ensuring that artificial intelligence does not operate unchecked. Deanna emphasizes that under the law, "you must have a human component to it. You cannot let AI make any decisions... there has to be a human check on any of those decisions." Beyond state compliance, the episode highlights a collaborative, multi-departmental approach involving leadership, human resources, legal, and IT teams to refine these guidelines. A key area of concern is customer and employee data privacy, which Deanna describes as a major responsibility since "we are responsible for shepherding our customers and their information and our employees and their information." The hosts also discuss the practical challenges of software terms of use, noting that platforms like Microsoft Copilot often contain clauses designating the tool "for entertainment purposes only," meaning tech providers carry no liability for professional errors. This lack of vendor liability emphasizes the necessity of internal policies, as Deanna points out that "the AI vendors, they don't care if their stuff is misused because they've already decried all liability." The discussion covers a real-world learning experience where an IT director innocently used an AI note-taker during a Microsoft Teams meeting. This highlighted a massive legal risk, as using automated transcription can inadvertently destroy confidentiality. Deanna warns that if a note-taker records an internal conversation, it can "destroy attorney-client privilege" and create a record that becomes "foiable" under public records requests. This risk is reinforced by a recent ruling from the Southern District of New York. In that case, the government successfully subpoenaed a defendant's search and prompt history from a generative AI platform because information shared with commercial AI platforms lacks confidentiality. The episode addresses other technical shortcomings, such as AI "hallucinations." Deanna mentions that even top legal research databases have faced scrutiny because their specialized research bots were "putting out incorrect summaries of cases." This reinforces why continuous verification remains mandatory, since a language model "is a language prediction tool, and it's trying to give you the next word in the sequence." Ultimately, the episode encourages leaders to proactively engage with these technologies rather than avoiding them. Deanna concludes by urging organizations to adapt continuously, advising managers "to not be afraid of AI" but to "respect it" and commit to ongoing policy reviews as the software evolves.
24 episodios
Comentarios
0Sé la primera persona en comentar
¡Regístrate ahora y únete a la comunidad de Woman-Run!