Originally published in the September 2025 issue of the Notes on Antiracism, Justice, and Equity newsletter.
I thought the future would be cooler. We were promised hoverboards and self-driving cars, but instead, we have the same old racism and misogyny remixed with new-ish technology. What seemed like a new, exciting way to learn, connect, and save time has been revealed as an ethical disaster on multiple fronts. This month, I am diving into how artificial intelligence, and specifically generative AI, clashes with some of the core values of public health.
Generative AI can be defined as a subfield of AI in which text, images, videos, and other forms of data are produced from input the models receive. Inputs used to train these models include books, visual art, audio, and yes, academic publications.
One of many ethical concerns regarding generative AI is the lack of appropriate credit and compensation of the creators of source materials. Campaigns and lawsuits have been started by creators to combat Meta and other platforms from using their work. As cultivators and stewards of knowledge, we should be concerned about work being unattributed or falsely attributed. Considering how our time is limited and often monetized, it is important that our work is appropriately recognized. Just as we hold students accountable for academic integrity, we should consider how we uphold these standards outside of the academy as well.
Another concern about generative AI is how frequent users become reliant on the technology. In one recent article, Dr. Fawad Khan suggested that dentists (and providers more broadly) may become reliant on AI, resulting in an erosion of their clinical skills. Another article found that students lost cognitive abilities and growth as a result of using AI on peer reviews. Students, who have presumably enrolled to learn how to research, synthesize, and share vital information, may be losing the opportunity and skills to learn, which involves anticipation, discovery, processing, and integration into our existing knowledge and beliefs. As Dr. McMillan Cottom put it, “Just knowing a fact is not the same as learning and that we are denying young people that experience of the pride of having acquired skill, talent and ability is, for me, just so sad.”
There are many other concerns with generative AI that I don’t have space to cover here, including major privacy concerns, the tendency for platforms to ‘hallucinate’ information, contributions to gender-based violence, the spread of misinformation, and the rise of unhealthy relationships with chatbots. Perhaps most relevant to public health however, is the impact AI is having on our environment. As we know, extreme weather has put us all at risk this year, whether it was from unsafe air, tsunamis, or devastating floods. Poorer communities, especially those in the global south, continue to suffer the most from our consumption, destruction, and disruption of our delicate ecosystems. The heavy amounts of data processing that generative AI require means that it needs to be cooled, quickly and en masse. In Memphis, a predominantly Black community, the supercomputer facility that powers Elon Musk’s AI chatbot Grok is using enough electricity to power 100,000 homes. It is also increasing Memphis’s smog by 30 to 60 percent, which is likely to lead to a significant increase in respiratory and cardiovascular diseases in the community. In Texas, data centers consumed over 50 billion gallons in 2024, enough to supply Austin for months. While water shortages continue in some parts of the country, AI manages to consume perhaps our most precious resource in unjustifiable amounts.
Perhaps AI is here to stay. Perhaps it is too late to put the toothpaste back into the proverbial tube. However, those that say it will make life easier are not telling the full truth. Yes, we may have the ability to compose emails faster, to create meeting agendas easier, to generate cute images to use for memes on social media. While these mundane tasks used to cost us small amounts of time, the cost of outsourcing them to AI are much more sinister.
