Lessons Learned

AI & Community History

Gaining knowledge together \u2014 what AI gets wrong about LGBTQ+ history, why it matters, and how community expertise is the only real corrective.

This page grows from a real incident in our own community. We share it because we believe transparency about AI\u2019s limitations is part of doing honest history work \u2014 and because we hope you\u2019ll want to participate and teach us.

A Case Study from Our Own Community

In June 2026, a Facebook reel used AI-generated imagery to illustrate the story of the first Michigan Pride march in Detroit \u2014 a march that began with a pageant at Lansing\u2019s Joe Covello Bar in 1972.

The AI conjured a generic, sanitized, and whitewashed \u201cMiss Capital City\u201d \u2014 erasing Aretha (also known as Mike Scott), a Black drag queen from Lansing who actually won that pageant on June 22, 1972, and then led the procession down Woodward Avenue. The AI image also omitted that the pageant was sponsored by the Michigan State University Gay Liberation Movement, and called the event a \u201cparade\u201d rather than a \u201cmarch\u201d \u2014 the word participants themselves used.

The authentic photograph \u2014 scanned from the personal papers of Greg Kamm \u2014 shows Aretha winning the Miss Capital City pageant. It tells a completely different story than what AI produced.

\u2014 Analysis by Tim Retzloff, shared with the Lansing Queer Almanac community

What We\u2019ve Learned

Six Lessons About AI & History

AI can erase people of color from history

When AI generates images or summaries of LGBTQ+ history, it often defaults to white, gender-normative representations — erasing the central roles that queer people of color played. A real example from our own community: an AI-generated image of “Miss Capital City” whitewashed Aretha (Mike Scott), a Black drag queen from Lansing’s Joe Covello Bar who won the pageant in 1972 and then led Michigan’s first Pride march in Detroit. The authentic photograph, scanned from the personal papers of Greg Kamm, tells a completely different story.

AI flattens language and erases identity

AI tools often strip away the specific names, terms, and self-descriptions that communities used for themselves. Calling the 1972 Detroit demonstration a “parade” instead of a “march” — as most participants called it — is not a small error. Language carries political meaning. When AI smooths over those distinctions, it sanitizes history.

AI-generated content can spread quickly as fact

Social media platforms now label some AI-generated posts, but many people scroll past those labels. AI content about local history can circulate widely before anyone with firsthand knowledge sees it. Community historians and archivists are often the only people positioned to catch these errors — which is exactly why projects like this one matter.

Primary sources are irreplaceable

Photographs scanned from personal papers, oral histories, local newspaper clippings, and community-held documents cannot be replicated by AI. They are the ground truth. The Lansing Queer Almanac is built on the belief that community members are the keepers of their own history — not algorithms.

AI can be a useful tool — with critical oversight

AI can help with tasks like transcribing handwritten documents, suggesting search terms, or drafting outlines. But every output needs to be checked against primary sources and community knowledge. The question is always: who is being centered, who is being erased, and who gets to verify the answer?

Community knowledge is expertise

People who lived through events, who knew the people involved, who held the photographs and letters — they are experts. AI systems are trained on whatever text existed on the internet, which skews heavily toward dominant narratives. Local, queer, and community-of-color histories are systematically underrepresented in those training sets.

Keep Learning

Key Resources

Community archives, oral histories, and podcasts that model what rigorous, community-centered LGBTQ+ history looks like.

Podcast

Queercore Podcast — LGBTQ+ Health Program

Conversations centering LGBTQ+ lived experiences, health, and identity — a model for community-led knowledge-sharing.

Digital Transgender Archive

An international repository of trans history materials — a powerful example of community-controlled digital archiving.

Oral History

ACT UP Oral History Project

Video interviews with surviving members of ACT UP — demonstrating how oral history preserves what written records miss.

GLBT Historical Society

One of the world’s largest LGBTQ+ archives — a resource for understanding what rigorous community history looks like.

Michigan

Michigan LGBTQ Remember

A Michigan-specific memorial project preserving names and stories of LGBTQ+ community members — local history done right.

Academic

University of Michigan LGBTQ+ Heritage Initiative

Scholarly resources and exhibits on Michigan LGBTQ+ history, including materials on early Pride organizing.

More resources are available in our full Resources Library.

Hope You\u2019ll Want to Participate

This page is a starting point, not a finished document. We want to hear from people who have encountered AI errors about LGBTQ+ history \u2014 locally or nationally \u2014 and from anyone who has found useful ways to work with or around these tools.

Teach us. Share a story, flag a resource, or bring a case study to one of our monthly meetings. Community knowledge is how we keep the record honest.

Share Your Story Join a Meeting Email Us