Generative AI has the potential to transform public service experiences. Will it deliver?

By Eyal Darmon and Avik Batra

The most popular Generative AI solution, ChatGPT is currently used by over 100 million people. It’s passed the bar exam at a higher ratio than any other artificial intelligence (AI). It can write epic poems, create workout plans, develop business strategies, give movie recommendations—all in an instant, with minimal context provided by the user.

But can it connect constituents with the right department to learn about food assistance programs?

Before we explore that question, let’s look at why generative AI has dominated headlines for the past few months. Generative AI captured global attention for its ability to mimic human dialogue and decision-making.?Accenture’s research found that 40% of all working hours can be impacted by large learning models (LLMs) like those that power Generative AI solutions. That’s nothing short of revolutionary.

While Generative AI is an incredible tool, it has its limitations. It can’t replace what people do. Case in point: Generative AI lacks real-world understanding and has difficulty with specialized topics that are domain specific.

That said, given the transformative promise of generative AI and there is still much to learn, how should public service agencies be thinking about this technology? And will it ultimately be able to help them serve constituents in a better, faster, and more efficient way?

The evolution of AI in government services

Let’s take a step back and look at the evolution of AI in government services, and where we are on that continuum today.

• Evolution 0: No capabilities.?For years, there were no AI capabilities supporting government services. For instance, when a constituent called for services, a contact center representative answered the phone call and provided the information needed or connected the constituent with another department.
• Evolution 1: Virtual agents. AI was introduced to government services as virtual agents. Constituents most often interacted with these agents, which used?natural-language processing, as chatbots that provided automated service, transactions, or guidance. Virtual agents needed to be trained; their capabilities are limited to what they’ve been trained to do.
• Evolution 2: Generative AI.?This is where we are on the continuum today. The second evolution of AI is a significant step forward. For one thing, agencies don't need to train a virtual agent. Foundational models exist that have thousands of data sets already in place as the baseline. Once the container/context is created for a new use case, the AI consumes and interprets vast quantities of data; interprets the data; establishes context of the data; and remembers the data to create content and experiences.

That, combined with the ability to learn language, context, and intent, offers the potential to elevate customer experiences to new levels of speed, efficiency, and proactive helpfulness.

What does generative AI mean for government services?

The COVID-19 pandemic accelerated the transformation of government services, spurring agencies to push the envelope in terms of delivering modern, digital public service experiences. Generative AI has the potential to take that transformation another step forward, almost at that pandemic-era speed.

In short, generative AI can enable agencies to continue to improve services by empowering public servants with technology. To be clear, jobs won’t go away, and the need for people to provide service won’t disappear either. What will change is what human intervention looks like in the service space, along with training and job functions.

For instance, with the enhanced automation and predictive intelligence enabled by generative AI, call center reps will be able to improve customer service. They’ll have better information about why constituents are calling and the kind of information they’re calling about, and so they will be able to give more relevant, reliable, and accurate information, faster.

Or let’s go back to our original scenario: a constituent looking for the right department to learn about programs for food assistance. Cities are organized into numerous departments. Since most people only interact with a select few of them (such as the Department of Motor Vehicles), it takes some searching to find the right department for their queries. To find food resources, is this a question for the housing authority? Human services? Or somewhere else entirely?

Generative AI could understand the user’s intent and direct them to the appropriate department. It would also be able to make recommendations about additional services available to the constituent or answer other questions they may have about assistance programs.

A pivotal moment for public service delivery

So, to answer the question: “can generative AI connect constituents with the right department to learn about programs for food assistance?” That’s a yes, right? Well… it’s not a clear-cut yes or?no.

Remember: this technology is still new . It needs extensive pressure testing before we can make a final determination as to whether this is truly the next big thing for public service.

What we are sure about is that generative AI looks incredibly promising. If it lives up to its potential, then we’ve got the foundation for exponentially revolutionizing how government services are delivered. It will enable agencies to innovate on public service experiences faster, at greater speed and scale than ever before—such as the ability to write code better, faster, and cheaper than today.?As we are seeing in the market today, many state and local governments are trying to modernize their digital channels.?Prior to the availability of large language models to help write code, public service agencies would traditionally need to budget significant money and effort to complete these projects through the entire lifecycle.?With Image models, design systems could be created rapidly with agency style guides, and responsive front-end code could be written quickly with code models.?Imagine after the initial code is generated, another model could help with WCAG 2.0 Accessibility code generation and tagging!

Such a transformation will increase capacity for agencies, and it will be exciting to imagine what they could do with this potential extra bandwidth and budget.

This is a pivotal moment for agencies to set a new performance frontier for public service experiences.?We are helping public service organizations move from AI interest to action in a?responsible way with clear business cases and by studying its potential and ramifications. Partnering with our public sector clients we are exploring how both apply while also seeing what other industries have learned in their early journeys.

It’s an exciting time to see how these new technologies can revolutionize service delivery for your constituents—to food resources and way beyond.

How do you see generative AI being used in public service? Let’s start the conversation to explore this technology’s potential, together.