AI Done Right: What We Learned When We Asked “Should We?”

On September 25th, we gathered virtually for the second volume of Button's Lunch n' Learn series to tackle a question that's becoming increasingly urgent in public sector work: How do we know when AI is actually the right tool for the job?

The timing couldn't be more critical. Generative AI has become ubiquitous, accessible, and, perhaps most importantly, expected. Product owners, directors, and managers across sectors are hearing the same refrain: "use AI to be more efficient." But as our host Elliott Lee, PhD, pointed out, this rush to adopt AI mirrors Maslow's famous observation from 1966 about treating everything like a nail when your only tool is a hammer.

What We're Actually Using It For

Before diving into the "should we" questions, we started with the "are we" reality check. The conversation revealed something interesting: most attendees were already AI users, but their approaches varied dramatically between personal and professional contexts.

Outside of work, AI has quietly become the opinionated search engine people didn't know they needed. Instead of wading through ten blue links, people described turning to AI for detailed, contextual responses on everything from travel planning and cooking to health questions. One attendee shared a practice that resonated across the group: using voice input to transform rough bullet points into coherent paragraphs, freeing themselves from the paralysis of perfecting every word on the first pass.

At work, the picture shifted. AI emerged as a tool for research, analysis, and decision support. Project management. Content editing. Subject research. Grant writing. Some even used it to generate illustrations. But here's where it got interesting.

The Critical Thinking Difference

The group's reflections revealed something that Elliott found particularly encouraging: people aren't just accepting AI outputs at face value. They're treating AI with the same healthy skepticism they've learned to apply to Google search results.

"Instead of just going with the first answer, people said they usually dig deeper or refine their prompts," Elliott noted. This wasn't universal AI enthusiasm—it was thoughtful, iterative engagement. People described using AI for first drafts and design work, then applying their own expertise to elevate the results.

The distinction mattered. Attendees saw AI as a tool to complement their skills and expertise, not replace them entirely. They were curious but cautious, asking questions about ethics, environmental impact, and accuracy – especially in legal contexts and non-profit work where getting it wrong carries real consequences.

The Evolution of AI Interaction

What emerged from the discussion was a snapshot of how people are evolving their relationship with AI. Some are still exploring basic use cases. Others have developed sophisticated prompt engineering skills. But across the spectrum, one pattern held: the most effective AI users weren't the ones who trusted it most – they were the ones who questioned it best.

This matters because it speaks to a fundamental tension in how AI is being deployed, particularly in public sector work. When the pressure is to "use AI to be more efficient," the risk is treating it as a solution looking for problems rather than a tool that might, or might not, be appropriate for the challenge at hand.

Deciding Before Diving In

The Lunch n' Learn deliberately kept slides light and thinking heavy. The goal wasn't to provide definitive answers about when to use AI. It was to foster the kind of critical evaluation that attendees were already practicing in their best work.

Because here's the reality: AI can be transformative. It can also be overused, misapplied, and deployed without full understanding of its implications. The difference between these outcomes isn't the technology itself. It's the decision-making process that comes before implementation.

What's Next

The conversation we had on September 25th is one that needs to continue happening across organizations, teams, and projects. Not because AI is inherently good or bad, but because thoughtful adoption requires ongoing dialogue about appropriate use, ethical implications, and the irreplaceable value of human expertise.

As Elliott reminded the group, you don't need to be an expert to engage with these questions—just curious. And based on the depth of reflection and critical thinking that surfaced during our hour together, curiosity might be the most valuable tool we have.

Interested in joining future Button Lunch n' Learn sessions? We keep them light on presentation and heavy on thinking, tackling the questions that matter most to people building digital services in complex environments. Stay tuned for Volume 3 next month.