AI, large language models, and government
Finding where AI and CX meet
As artificial intelligence (AI) tools reshape consumer technology, so will they soon reshape how the public expects to interact with their government. By carefully applying AI tools, leaders in government have an opportunity to increase the efficiency of their teams while closing the gap between how people experience consumer services and public services.
At Ad Hoc, we’ve implemented AI tools and large language models (LLM) to help our customers and we’re exploring new ways to apply these tools to government customer experience challenges. We base these efforts on our foundational approach we’ve developed over the last ten years using human-centered design and research thinking to enable agencies to improve the experience of customers using their most critical services.
Seven ways AI tools can help
Insights from unstructured data
Many government programs have databases of unstructured data hiding information from decision-makers. LLMs can help teams organize that data and find insights.
Semantic search
Traditional search engines rely on keywords. Generative AIs and LLMs offer the next evolution in finding information: searching by meaning or intent.
Question answering
LLMs are powerful tools for understanding context and have the ability to summarize large amounts of text. Combined with semantic search for finding relevant and precise content, an AI-powered search can directly answer user questions, saving them time from wading through pages of search results.
Plain language
Generative AIs excel at translating the style of a piece of text, like writing a State of the Union address in the style of Shakespeare. Agencies can also apply that skill to dense, legal information to create plain language resources.
Help the helpers
Call center representatives already know their domains well, but an LLM trained on their specific program could boost efficiency and customer service.
Intelligent coding assistance
Moving beyond drag and drop low code solutions to a tool that allows you to describe the functionality of a website and have an AI write the code, increasing team productivity and enabling agencies to tackle more impactful challenges.
Your agency’s challenge
AIs and LLMs are just tools, not silver bullets. Ad Hoc wants to work with agencies on ways to improve their customer experience and explore whether AI is the right solution for their problem.
Get in touchAddressing AI compliance in government
Three core concepts to think about when applying AI:
Safety
There are many legitimate concerns about the safety of AI tools and AI decision making. Those use cases deserve far more study to ensure equitable use in government. There are also many pragmatic applications of AI in government that can improve the speed and experience of applications while avoiding more experimental approaches that increase the risk for agencies and the public.
Accuracy
One hurdle of generative AIs and LLMs is they are not a database of facts or a search engine that crawls the internet. They are a language tool that can return grammatically correct responses that are inaccurate. Government agencies can improve accuracy by controlling the text that is fed into the model and using techniques such as Document Question-Answer patterns.
Control
While the most popular Generative AI tools are run by private organizations, there are open-source tools that agencies can host locally. This offers increased control over how a particular instance of an LLM is trained and used and can help agencies ensure they meet their goals and any future government regulations without sending data to third-party applications.
Closing the gap between consumer technology and public services
Like mobile applications and cloud hosting before it, AI is going to change the public’s expectations of how they interact with services and technology.
Government agencies must find meaningful ways to adopt these technologies or public services run the risk of falling even further behind consumer services.
Ad Hoc can help agencies apply the right type of AI tool to the right government challenge to close that gap in expectations.