エピソード

  • Just Because genAI Can, Doesn’t Mean genAI Should - in City Hall
    2024/10/01

    In Episode #3, hosts Mike Sarasti and Mark Wheeler respond to questions from listeners about generative AI. Mike and Mark make it clear: for local governments, early experimentation is valuable, but do not downplay the risks and execution considerations. Slapping on “use at your own risk” warning on your city hall genAI chatbot is maybe not the best way to actually serve the public?

    • Not All AI Risk Profiles Are the Same: Local governments are conflating the very different risk profiles of generative AI vs. predictive AI, and by doing so, are creating unnecessary work and confusion about how to assess the impacts of "genAI-ifying" SaaS tools.

    • Maybe Don’t Let a Copilot Take the Wheel: An immature, all-purpose generative AI, known to fabricate, is given access to the unclassified data of your workforce. What could go wrong?

    • A Creative’s Organization Tool: Mike offers a more personal perspective, sharing how he uses AI tools like OpenAI’s ChatGPT to streamline his daily tasks and organizational workflows for his consulting work and his band’s music production/performances. He makes the case that while all-purpose genAI is great at organizing, he’s not relinquishing creative control.

    • Focused is Better than “All-Purpose” genAI: Mark and Mike advocate for a sandboxed, focused, and restrictive use of a genAI model tailored (aka fine-tuned) for tasks on specific knowledge domains. By limiting the AI’s scope to a particular field, local governments can gain more accurate, reliable, and contextually relevant outputs, while minimizing the risk of errors.

    In the closing music segment, Mike gives an update on his band, Raker, and their upcoming shows and competition showcase for the Miami Three Points festival—and the band’s latest single, “Relevant,” streaming now on all music platforms.

    Featuring a “Just Say No” faux PSA by Earl Grant, local Philadelphia actor.

    Follow Raker on Instagram: @rakermusic

    Listen to "Relevant" on Spotify: https://open.spotify.com/track/4JqywsINKEWZucDNYlnBBU?si=da45e2c66825465b



    続きを読む 一部表示
    30 分
  • Cities, Slow Your Roll on Enterprise Generative AI
    2024/08/10

    In Episode #2 of No BS Bureaucracy, hosts Mike Sarasti and Mark Wheeler explore local governments' readiness for generative AI. They tackle the notion that cities must rush to put enterprise-level gen AI tools into production. Their bottom line?

    1. Don’t relinquish the wheel to any gen AI co-pilots until city leaders understand the contours of their departments’ data sprawl.
    2. Leaders must be familiar with gen AI tools themselves.
    3. AI governance is a natural extension of all your existing frameworks for governing data, privacy, and ethics - not a standalone program to be built from scratch.

    Mike shares his personal journey with gen AI, explaining how he jumped into a relationship with ChatGPT rather quickly (hmmm… U-Haul has never seen a moving van lease get signed so fast!). Mike likens the process to "carving a sculpture," where the first output of gen AI is like a blob of potential, but you’ve got to spend time picking it apart and refining outputs. He recommends that local government leaders begin with familiar topics (where they are the subject matter experts) and see gen AI governance through the lenses of all existing policies on privacy and equity.

    Mark shares recommendations for crafting AI governance from existing data security and privacy programs. His main point is that AI is data. To govern gen AI effectively, cities must first be able to govern ALL of their data. He recommends taking stock of an organization’s data sprawl and asking some critical questions: What’s the current state of data governance? Is there a classification scheme for unstructured data? Are data retention policies enforced for legal and/or historical preservation purposes, or are hoarding tendencies running unchecked? Plus, where to go for help—like the GovAI Coalition for policy, procurement, and education templates and guidance.

    Mark also provides some stats to back up the feeling that gen AI adoption is a “must-do” for leaders but questions why the evidence of its transformation on industries and everyday life hasn't materialized.

    ADVERT: The Sara Connor Anti-AI Plug-in

    In the closing segment, aka Mike & Mark’s music picks, Mike announces his band Raker is releasing new music in August! After a few years' hiatus, Raker is back with the release of the single “Hologram” on all music streaming platforms!
    https://open.spotify.com/track/6Qf228NOvNIG3MjoCc9ZeX?si=a866719ff1b242a6
    Instagram: @rakermusic https://www.instagram.com/rakermusic

    Mark describes what it's like to fully experience Philly’s eclectic Martha Graham Cracker Cabaret.

    Thanks to Earl Grant for lending his thespian skills to the “Anti-AI plug-in” advert.

    #innovation #AI #genAI #localgovernment #govtech #publicsector #smartcities

    続きを読む 一部表示
    39 分
  • Portable Digital IDs and the City - Ready or Not?
    2024/07/06

    The pilot episode! The ex-CIOs are “unleashed” and talk about the state of innovation leadership in local government (whose job is it really?) and delve into the value of digital IDs and state-issued mobile driver licenses (mDLs) to cities. They discuss how different approaches to digital ID can seamlessly, and securely, unlock municipal services for residents. Mike provides details on a recent project exploring emergency relief payments with decentralized IDs and verifiable credentials. Mark answers the question, how do local governments get ready for the adoption of portable digital IDs and mDLs?

    Plus

    • Recap their digital ID panel at Smart Cities Connect with guest speakers and digital ID experts from Socure, Ying, and Spruce ID.
    • Hot topics at the recent Smart Cities Connect conference in Raleigh, NC.
    • Highlights from the local music scene - the Sweeping Promises show in downtown Raleigh.

    Thanks to local Philly actor Earl Grant for his voice on Translator Gator.


    続きを読む 一部表示
    40 分