The Commonwealth of Pennsylvania is starting a pilot program in January for state employees to use generative artificial intelligence (AI) for copywriting, coding and updating documents.
Gov. Josh Shapiro announced Jan. 9 that the state is partnering with OpenAI, an AI research and deployment firm that offers a platform for generating content, images and code based on user prompts. OpenAI operates the ChatGPT language tool that has an Enterprise option with increased security and privacy to prevent any sharing of sensitive data.
“Pennsylvania is the first state in the nation to pilot ChatGPT Enterprise for its workforce,” OpenAI CEO Sam Altman said. “This initiative showcases Pennsylvania’s commitment to innovation, led by the strength of the state’s diverse communities and dynamic economy. Our collaboration with Gov. Shapiro and the Pennsylvania team will provide valuable insights into how AI tools can responsibly enhance state services.”
Shapiro anticipates the pilot will help employees better understand how and when to use AI tools daily in state operations. Initially, only employees in the Office of Administration will use ChatGPT for tasks such as updating outdated policy language, creating job descriptions and cleaning up employee policies that may have duplicated or conflicting information.
“Generative AI is here and impacting our daily lives already, and my administration is taking a proactive approach to harness the power of its benefits while mitigating its potential risks,” Shapiro said in the news release.
Shapiro has been preparing state employees for the pilot program since September when he issued an executive order that set the framework for the state to use generative AI. The order listed 10 core values for Pennsylvania to consider when using AI, including accuracy, employee empowerment, innovation, safety and security and transparency.
As part of the executive order, the state established an AI Governing Board with state administration officials. Pennsylvania State University is also partnering with the state to offer expertise from its Center of Applications of Artificial Intelligence and Machine Learning to Industry (AIMI) to the board. University faculty is offering guidance to the board on public and private sector opportunities for using AI.
“As government leaders, we need to lean into innovation and adapt to the changing tech environment in a responsible way while we educate ourselves and work proactively to minimize the risks and maximize the benefits of new technologies,” Shapiro said in the October announcement.
While Pennsylvania may be the first state to partner with OpenAI, other government entities have been using or preparing to use AI tools.
The state of Maryland is working on several digital initiatives to prepare for using AI and improving user experiences. The new AI Subcabinet will create an AI action plan, and the Maryland Department of Information Technology (MDIT) will form a digital services division to improve the website experience for residents. The state also plans to issue its first digital accessibility policy and form the Maryland Cybersecurity Task Force to take proactive measures.
“The collective actions we are announcing today are the beginning of where we need to go as a state and a nation,” MDIT Secretary Katie Savage said in the Jan. 8 announcement. “To be competitive and include everyone in our success, we must embrace new technology while ensuring it is secure and accessible to all.”
In October, New York City released its Artificial Action Plan as a framework for evaluating different AI tools the city could use and weighing any risks involved.
Several federal agencies have also begun using AI, including the Department of State, which uses Robotic Process Automation to automate the federal procurement data system for improved efficiency. The U.S. Mint is looking into how AI could improve the coin-quality inspection process.
However, other federal agencies are taking a more guarded approach until more can be known about the impact of generative AI in government operations. Last May, U.S. Sen. John Barrasso, R-Wyoming, issued a letter to the secretaries of the interior and energy departments encouraging them to block access to ChatGPT. He reported the Environmental Protection Agency had already done so until it could determine any risks.
“This security-focused mission makes DOE a prime target for adversaries that wish to undermine our national security and technological capabilities,” Barrasso wrote to the Department of Energy. “The use of ChatGPT and other third-party AI tools by DOE employees for official business only compounds the risk of compromise and attack of sensitive DOE information and systems.”