Government agencies see many possibilities for using AI tools in their daily work, but in reality, only a small portion are put into practice. The GAO found that 20 non-defense agencies identified over 1,200 use cases within the government, but they have only implemented about 16% of those AI use cases. AI poses risks to the public, such as producing biased outcomes that could 'amplify existing inequalities', especially among historically underserved populations. For the 2023 fiscal year, federal agencies requested $1.8 billion in funding for non-defense AI research.