Personalizing experiences with GenAI
The speed at which AI has grown has drawn both proponents and skeptics into a debate about the utility, ethics and value of these tools. The long-term prospects of AI may seem both unimaginable and unsettling, but there are some immediate opportunities that AI can help government agencies address in the near term.
A highly personalized digital experience requires a combination of technological capabilities, creativity, and multiple variants of content fine-tuned to unique audiences. A digital video streaming service will present two different users with different still images, synopsis and featured actors depending on what its algorithm believes will appeal more directly to a specific viewer. Creating these content variants has historically been a costly, highly specialized and labor-intensive process. But GenAI algorithms, written to parse high-volume data in real time, can solve this by dynamically sorting and creating content versions based on audience variables.
For a complex government agency that offers numerous programs or services, it can take average users a long time to click through the maze to find what’s relevant to them. Search functions are imprecise and often they expect the user to “speak government” or at least know exactly what they are looking for, which, in a site full of regulatory language, is too high of an expectation.
Ideally, such websites need to offer a more personalized experience to a broad range of specific user types and needs. GenAI can create multiple versions of written and visual content that a personalization engine can then present to different user segments, ultimately helping these agencies achieve a similar level of personalization as the private sector. While a human will need to perform quality review of these content variants, the review time is a fraction of the typical creative time for these types of tasks.
Improving navigation and information access
Perhaps the biggest challenge with government digital experiences is information density. A typical government website has a non-intuitive menu navigation, an overwhelming number of links, and lots of text that, though well-intentioned, often confuses or overwhelms average users who value simplicity.
Some agencies are seeking customer-centric redesigns, and these are worthy efforts, though they can be costly and time-consuming. A faster path to making CX improvements can be a sophisticated conversational AI that can help users get direct access to content while redesign efforts are being evaluated and executed. These are not simple FAQ-style chat bots that have started to wane in popularity, but active conversational agents. FAQ bots suffered from somewhat static content pegged to specific questions, and an inability to learn from what users were asking to improve accuracy and usability.
Newer, conversational AI agents can utilize generative AI techniques to better simulate talking to a human, to surface deep links to specific tools and information, and can summarize complex regulatory language into bite-sized concepts that are easier for laypeople to understand. In many cases, this type of agent can index a site and start generating improved usability in a matter of days. These interfaces can dramatically increase task completion while other improvement projects are underway, and these interfaces get better with time and more usage.
