The Wild Wild West of AI: What I Heard at MMANC
AI feels like the Wild West
In California's gold country, thousands of people poured into territory that had almost no rules, staking claims and hoping to strike it rich. A few did. A lot more people who were sure they'd get rich went home broke. Whole new industries sprang up almost overnight, and California grew faster than anyone could govern it.
AI is more than a little like this.
In 2024, I was at MMANC's Summer Symposium, where the organizers had used AI to draft the speaker bios. I was incredulous. AI writing actual words about actual people? It felt like a miracle.
Fast forward to MMANC's annual conference this fall: The exhibit hall was full of vendors selling dedicated AI products to public sector leaders, and just about every other vendor was touting the AI tools built into whatever they were selling.
From attendees I heard just about every possible take on AI. Some folks started from the assumption that AI is destructive, even evil (it's consuming all of our water!). At the other end of the spectrum, I met one agency that had essentially given a staffer free rein to build whatever he wanted with it. And everything in between: worries about jobs, privacy, security, the union, the environment, council, the public, and plain old fear of getting it wrong.
At the plenary session, the conference ran a quick poll asking people about their biggest obstacles to AI adoption. Almost everybody in the room landed on one of two answers. Either they don't have a policy or a plan, or they are worried about risk and security.
Here's the thing: I think those are two sides of the same answer. "We're worried about risk" usually means "we don't have a framework that tells us which risks we're willing to take." Without a solid framework, it doesn't feel safe to act at all.
Management analysts and middle managers are desperate for their leadership to just give them some parameters to work within. My session, "What's Your AI Plan?", was packed with management analyst types, and at least three different people told me they were texting their city manager during the session, because their organization needed to get focused on this. Now. But that doesn't make it an easy task to focus on.
Why is AI so overwhelming?
"AI change" doesn't feel like one problem; it feels like twenty. "Can staff put resident information into ChatGPT?" sounds totally unrelated to "What will the union say?" And "What do we tell council?" sounds unrelated to "Who approved this tool?"
But these aren't twenty separate problems, and you don't need twenty separate management systems to deal with them. These distinct issues come down to four leadership questions:
Direction: Where are we trying to go?
Decisions: How will we make AI decisions?
People: How will our people adapt?
Evidence: How will we know what's actually happening?
Every one of those twenty anxieties lives under one of these four. And each one has concrete, practical moves a city can make right now, without waiting until it has everything figured out.
Next week I'll be running my MMANC session again online, and we'll cover all four of these decisions. But as a teaser, here's an example of the "Decisions" question:
A Quick Tip: Decide in Advance
Most cities are trying to evaluate AI one proposal at a time, which is exhausting and inconsistent. Instead, sort use cases ahead of time into three groups:
Definitely allowed. Uses you're comfortable with, so you can give your people explicit permission to experiment.
Definitely not allowed. Your red lines: the things you won't do with AI, at least given its current capabilities and limitations.
Everything in the middle. A decision framework for evaluating the gray areas, so every new idea doesn't turn into a fresh argument.
This removes a lot of the anxiety from the decision making process.
By clearly defining expectations, you minimize surprises, make defensible choices, and have more constructive conversations. Your staff are given as much freedom and autonomy as they can be to take initiative. Boundaries are clearly defined to protect everyone. And lastly, leadership has a cohesive and aligned decision structure for evaluating the grey areas.
Want the rest? Join me on October 21
On October 21, I'm bringing my MMANC session online as a webinar, The Wild Wild West of AI. We'll walk through all four leadership questions, the practical tools behind each one, and how to figure out where your organization should start. You'll walk away with a readiness check and a reference guide that walks you through how to answer each of those four leadership questions, as well as templates to write your own agency's AI plan.
You've got a choice: you can let AI happen to your agency, or you can take a proactive position and make a plan.
Comments