This DOGE staffer’s GitHub posts might help us understand how Elon Musk wants to bring AI into the government

The internet posts and side projects of Department of Government Efficiency (DOGE) worker Jordan Wick could give some clues for how Musk’s efficiency group might attempt to use AI to downsize and retool the government.

During the last half of February, Wick, who has a DOGE email account associated with the Executive Office of the President and now is embedded in the General Services Administration (GSA), posted to his GitHub page the code for several tools that appeared to be related to DOGE’s work. The page was discovered by political reporter Roger Sollenberger at the end of last month. 

Wick posted the code for a tool that automatically downloads DMs from Twitter accounts. The code specifies Twitter accounts, which existed only until the social platform rebranded to “X” in October 2023, suggesting the possibility that the tool could be used to search through the digital past of government employees looking for disagreeable opinions or references. 

Another tool appeared to be designed for collecting sensitive data from government agency org charts. The tool contained fields for capturing the employee’s office, a 1-5 satisfaction rating, union status, and whether or not their position is statutorily mandated. 

Both the Twitter DM and org chart tools could be used to capture and structure data for use in an AI model. One source within the GSA believes the sensitive data collected by the tools could be used in a “potential AI federal employee evaluation/firing-bot.” Wick has now set his GitHub account to private. (DOGE didn’t immediately respond to Fast Company’s request for comment on Wick’s GitHub posts.)

Wick’s tools appear to be designed for evaluating and firing more federal employees, but his professional background suggests a deep interest in modernizing government IT systems using AI.

After graduating from MIT in 2020 and working at the self-driving car company Waymo, Wick cofounded (with ex-Palantir employee Anthony Jancso) Accelerate SF, which put on hackathons looking for ways to use large language models to improve local government services in San Francisco. In 2024, Accelerate SF renamed itself AccelerateX and began courting federal agency contracts with its own “modern OS for government.” A recruiting post on X said the company hopes to use AI to reduce expenditures, automate tedious work, and help interpret government policy language.

AccelerateX claimed last May that it already had contracts with “two of the largest transportation agencies in America,” but a search for the company’s name in the federal procurement database came up empty. Those contracts could be with local or state governments. (AccelerateX did not respond to Fast Company’s request for comment.)

One key challenge of using AI tools to streamline the government is that they must be deftly integrated into existing mission-critical systems—some of them old and brittle like the COBOL-coded mainframe systems used to process Social Security payments. Any interruption to those legacy systems could deprive millions of people of benefits they depend on to live. And there are many such systems active in the government.

https://www.fastcompany.com/91290599/this-doge-staffers-github-posts-might-help-us-understand-how-elon-musk-wants-to-bring-ai-into-the-government?partner=rss&utm_source=rss&utm_medium=feed&utm_campaign=rss+fastcompany&utm_content=rss