You won’t believe this, but a grant for a museum’s HVAC system was actually canceled because an AI flagged it for Diversity, Equity, and Inclusion (DEI) reasons. Yes, you read that right. The Department of Government Efficiency (DOGE), a federal agency once led by Elon Musk, used ChatGPT to review grants, and this seemingly straightforward repair project was caught in the crossfire.
The High Point Museum’s HVAC Dilemma
The High Point Museum in North Carolina was set to receive a $349,000 grant from the National Endowment for the Humanities (NEH) to replace its aging HVAC system. This was crucial for maintaining a proper preservation environment for the museum’s diverse collections. However, after the establishment of DOGE to review government spending, particularly concerning DEI, the grant was abruptly canceled.
Museum Director Edith Brady stated that the project had already begun but had to be terminated. Fortunately, they were able to recoup about 70% of the award through a termination clause. Still, the cancellation raises serious questions about the criteria being used for funding decisions.
ChatGPT’s Role in Grant Review
Court documents reveal that DOGE employees Justin Fox and Nate Cavanaugh used ChatGPT to assess grants. They fed specific prompts into the AI, asking if certain initiatives related to DEI and requesting factual responses under 120 characters, starting with ‘Yes.’ or ‘No.’ The prompt for the museum’s HVAC project included details about improving preservation conditions and providing greater access to collections.
ChatGPT’s response? “Yes. Improving HVAC systems enhances preservation conditions for collections, aligning with the goal of providing greater access to diverse audiences. #DEI.” This response, linking improved access to diverse audiences with DEI, led to the grant’s cancellation. It’s a fascinating, and perhaps alarming, example of how AI is being interpreted and applied in policy decisions.
Broader Implications and Deficit Reduction Claims
Acting NEH chairman Michael McDonald, appointed by President Donald Trump during his anti-DEI push, admitted in an email that some of the canceled grants didn’t actually promote DEI ideology. However, he also noted that DOGE’s stated goal was to assist with deficit reduction. Yet, when questioned in a deposition, Cavanaugh admitted that these efforts did not actually reduce the federal deficit.
This situation highlights the complex and sometimes bizarre outcomes when new technologies and political agendas intersect. The reliance on AI for such critical decisions, especially when the AI’s reasoning seems so broad, raises concerns about fairness, efficiency, and the actual impact on government programs and the institutions they support.
What do you think about ChatGPT flagging an HVAC grant as DEI? Does this seem like a reasonable use of AI in government? Share your thoughts in the comments below!