GUEST COLUMN.

In 2013, a group of experienced radiologists were tasked with reviewing lung cancer scans. The only difference between these scans and the thousands they had previously reviewed over multiple decades, is the researchers inconspicuously inserted a picture of a gorilla about 48 times the size of a typical lung cancer nodule. 83 percent of the radiologists failed to see the gorilla. 

The issue wasn’t carelessness or a lack of expertise. These were all veteran radiologists. They just weren’t expecting a gorilla; they were looking for cancer. Their training, their attention, and their expectations were tuned to one task so that something wildly out of place, something obvious in hindsight, went completely unnoticed.

The term for this is inattentional blindness. It happens in local government too, Consider procurement: Many of our procurement processes are so rigid and structured, that we can miss the most obvious obstacles and opportunities. The process is designed to tell us if a solution works, not what the results are or could be. 

The promise of AI is real. We see the demos and hear the carefully crafted pitches at conferences. We then draft RFPs, attend meetings with vendors, and get excited about the potential. But are we missing a gorilla in all of this?

I want to propose that we need to go beyond asking whether AI works and begin to question what happens when AI works. The stakes for local governments are too high for us to risk not asking the right questions.

I want to give you two practical ways to slow things down so you can see the gorilla in your AI procurement process.

First, understand motivation. I’m often asked, “How do I implement AI in my organization?” I answer with “Why do you want to?” The responses are usually something about productivity, efficiency, and cost savings. But then I ask: “Do you use AI in your personal life?” When the answer is yes, people report that they use it to brainstorm recipes, look something up, write poems, cartoonize a photo of our dog. So why do we use it there? Because it’s fun. It sparks creativity. It helps us feel more capable.

At home, we treat AI like a partner; but at work, we often treat it like a software upgrade. Efficiency is a good, but fragile, foundation. AI is efficient until it breaks down – much like a car is a more efficient mode of transportation than a horse and buggy. You’ll have days when the model gets it wrong, when the output derails a workflow. If the implementation framework is built on efficiency, it will collapse the moment the engine stalls. 

Government efficiency alone, in a way that AI can foster, doesn't build public trust, and front-line employees often lack incentives to prioritize it. Consider firefighters or front desk staff: they must remain at their stations for full shifts regardless of how quickly they complete tasks. Without meaningful rewards for efficiency, why would they care about streamlining processes?

While I support efficient government over wasteful alternatives, efficiency should not be the ultimate goal. Residents trust government when officials take time to explain decisions, listen to concerns, and respond thoughtfully to community needs. Those are deeply inefficient processes, when done right. Beneficial efficiency creates space for essential human connections. When we streamline routine tasks and processes, we free up capacity for the deeper engagement that actually builds public trust.

What should the motivation be? How about this: to enable staff to bring their full human intelligence to work. To help residents interact with information in ways that build trust. To support elected officials in understanding the downstream impacts of policy. That is a more solid foundation. One that can withstand the complexity and storms of civic life. The goal of implementing AI in local government must be to enable more meaningful relationships between government and the people it serves.

Second, take advantage of demos. Not only to verify functionality, but to sense impact. Many AI vendors offer demos and limited trial periods. Take advantage of them. Explore what happens to your organization when it works. In my new book Fragile Systems, I argue that AI does more than automate work, it alters the civic ecosystem. Too often, AI procurement is so procedural, so carefully bounded, that it blinds us to the gorilla: the factor that doesn’t fit neatly in a rubric but still reshapes everything. Maybe it’s the way a platform quietly reshapes interdepartmental communication. Or a slight shift in staff discretion that becomes a long-term dependency. 

The promise of AI is not really speed. The promise is that AI, if used wisely, might help us listen, adapt, and care at scale. And, we can do that as long as we step back and look for gorillas along the way.

The contents of this Guest Column are those of the author, and not necessarily Barrett and Greene, Inc. 

#RethinkingArtificialIntelligencePurpose #CityManagementArtificialIntelligence #StateandLocalGovernmentArtificialIntelligence #CountyManagementArtificialIntelligence #TrustInGovernment #StateMotivationForArtificialIntelligence #CityMotivationForArtificialIntelligence #GovernmentAIMotivation #PublicSectorAIMotivation #StateLocalGenerativeAIPolicyandManagement #CityAIPolicyandManagement #ArtificialIntelligenceAndCreativity #ArtificialIntelligenceAndTrustInGovernment #StateandLocalTrustInGovernment #FragileSystems #MicahGaudet #CityOfMaricopa #ArtificialIntelligenceAndCitizenOutreach #AIVendorDemos #ArtificialIntelligenceAndHumanIntelligence #StateandLocalGovernmentStreamlining #StateandLocalGovernmentEfficiency #StateandLocalGovernmentPerformance #AIandCityCitizenOutreach #AIandStateCitizenOutreach #AIandCountyCitizenOutreach  #ArtificialIntelligenceAndUnderstandingPolicyImpact #InattentionalBlindness #BandGGuestColumn #BarrettandGreeneInc