Forbes: How Public-Sector Organizations Can Factor ‘Step 0’ Into AI Adoption

How Public-Sector Organizations Can Factor ‘Step 0’ Into AI Adoption

July 1, 2024, JENNIFER SANDERS, Forbes

Don’t worry—this article isn’t going to ruminate on the world’s interest in artificial intelligence (AI)—that’s a fact.

Instead, it highlights the often-overlooked Step 0, the initial stage critical to successfully implementing AI programs. Are you truly ready?

As with new technologies and tools preceding it—big data, analytics, machine learning, Internet of Things (IoT), etc.—the internal understanding of and the dedicated exercise to establish where on the readiness scale an organization is is a gap worth emphasizing. Agnostic to the sector, Step 0 is the equivalent of "Measure twice and cut once."

Both private and—uniquely—public sector organizations are moving quickly toward implementing AI solutions, frankly, wildly more quickly than the majority of new paradigms over the past 10-20 years. Best-in-class providers will provide Day 1 plans, but the best may not realize the need to take a step back and ensure Step 0 is addressed.

There are three ways to look at Step 0.

1. Does the organization truly understand the solution?

Let's start with a prime early example—are you comfortable explaining the difference between AI and generative AI (GenAI)? Where did the interest in AI solutions originate? Was it a collaboratively identified problem to solve—or a shiny object that came down from the mayor’s office? At this point, AI cannot fully replace humans; is the organization looking to leverage AI to assist in making processes more efficient, or are they looking to leverage AI to fully replace select internal tasks?

A key early example of learnings in deployment comes from the internal integration of AI sandboxes, where many employees had experience using platforms like ChatGPT, but it was not clear that they had to use the internal solution for any work-related functions to not put non-public information into the publicly available model and risk unintended exposure of data.

Is there sufficient internal support to undertake something new and previously unproven? Risk thresholds can be a barrier (often a necessary one) to innovation, and there are models to support the identification and stress-testing of solutions—for example, through programs implemented by my own company, North Texas Innovation Alliance, and the Colorado Smart Cities Alliance and through technology policies established by Peachtree Corners. (Full disclosure: Both companies are peer organizations we collaborate with.)

Internal use cases are a key entry point to the use of AI. External and infrastructure-based programs benefit greatly from a measured approach.

2. Step 0 must include temperature checks on public trust in the government’s deployment of new tech.

For better or worse, the government is often held to a higher standard than the private sector (how long do youspend reading terms and conditions before hitting accept?), and the pre-work and intentionality are key components of Step 0 readiness.

Has attention been paid to potential public pushback on AI given widespread and justified, but widely misunderstood, concerns about privacy and governance? Will the specific uses of AI be ethical, and how are the benefits, protections and guardrails being communicated to residents and stakeholders?

There are organizations and cities that are taking this step in an intentional way, including Helpful Places, led by the former director of digital integration at Sidewalk Labs. Through these frameworks and infrastructures, a commitment is being made to supporting the public and private sectors in successfully evaluating, engaging and effectively utilizing technology for social benefit.

3. We need more about readiness.

Technology has always outpaced the ability to quickly implement in the public sector. For those embarking on new solutions, is your data cleaned and standardized for training AI models to ensure the solution is accurately presented? For already innovative public sector groups, is wanting to integrate AI just continued muscle memory for innovation? Organizations analyzing aspirational versus pragmatic approaches toward this know all about backing into the quick wins; the trick is balancing the political will necessary—for the time necessary—to reap the benefits.

To providers, there is not a product that is plug-and-play for all entities and systems. I beg of you, remove that statement from your pitch decks—and ensure a customer has hit Step 0—to set everyone up for success.