TL;DR: The major environmental, economic, and community harms of data centres, the misinformation and hallucinations, and the plagiarism and content theft that AI models require all make up the backdrop to AI technology. Whatever benefits its proponents say it will also have, without guardrails, there is growing evidence of the dangers the technology poses to both democracy and LGBTQIA+ rights.
“Artificial Intelligence is going to change our world, that’s inevitable. But in what way it changes our world is still up to us.” I wrote this, about the impact of AI on the queer community, five years ago. This was a time when – if you can believe it – the notion that AI would dominate our world was still novel.
The concerns and experts’ predictions about AI that I laid out in the distant past of 2021 are already having direct impacts on our lives.
One leader in the field who I spoke to all those years ago was anthropologist, media scholar, and 2020 MacArthur Fellow Mary L. Grey, who told me she believed AI would "always fail" the LGBTQIA+ community. Five years later, she tells QueerAF she stands by her words “even more strongly.”
Most AI tools we talk about today, at least in the context of concern or risk, are probabilistic tools, including those we've been investigating: Gemini, ChatGPT and Claude. These tools don't seek 'the right information' and give it to you. They serve it to you based on an estimated probability that it's right, based on previous examples and patterns.
We are a community where we are all significantly different, and have shifting approaches to language, expression and our identities. In an AI machine, this nuance gets lost in what trans researcher Chloe Turner calls "the neutral middle", where answers are accurate in a narrow sense but distorted by the sources they use.
In the trillion-dollar business of AI, the market priority is to serve what as many people as possible want to hear. Marginalised people, by our nature, do not have the numbers to be a priority, and we do not represent the needs of most users.
"AI learns from what shows up most often. So, people and experiences that are less common in the data are easier to miss," Grey says. "This isn’t just a technical bias problem. It’s about who gets to define the categories, goals, and acceptable mistakes when representing our communities. That is the underlying threat to LGBTQIA+ existences – it excludes our diversity, by design."
What are the dangers to the LGBTQIA+ community in an age of AI?
The major environmental, economic, and community harms of data centres, the misinformation and hallucinations, and the plagiarism and content theft that AI models require all make up the backdrop to this technology – whatever benefits its proponents say it will also have. And without guardrails, it is also a growing danger to both democracy and LGBTQIA+ rights.
Numerous studies and research, including QueerAF’s, have shown that although AI gives queer communities a sort-of OK airing, the popular models we investigated ultimately narrowed LGBTQIA+ diversity down to a neutral middle while presenting our perspectives alongside anti-rights rhetoric.
The trouble is that "at scale, those judgments spread quickly, while it becomes harder to see who made them or who is responsible", Grey tells QueerAF.
In the ‘room of one’ AI chats create, unlike in a newspaper article or radio show, poorly presented information about the Trans+ community can't be challenged. It means that, yes, the models will say conversion therapy is a pseudoscience – but they will also set out why people receive it anyway, and who to contact.
QueerAF
Can we make AI better represent LGBTQIA+ people?
"Representation helps, but it isn’t enough," says Grey. Putting more queer information into the machine does improve things somewhat, but its benefit is capped.
It can only go so far in a system that is designed by affluent Californians, who need to create products for as many people as possible. To do this, you have to create for ‘most’ people, which means marginalised people’s needs get lost.
If big tech doesn’t work with the people affected by the information these systems produce, and marginalised people don't get a real say in how the systems are shaped, how data is used, and how they're tested, little will change.
To take one example, QueerAF's journalism already shows up in these models. In Microsoft's agent Copilot, QueerAF was cited 5,600 times over a three-month period; over that same period, only three clicks to the QueerAF website came from Copilot.
We will never know whether our information was reliably used – and worse still, Microsoft made revenue from our creatives' work without providing us or those creatives with any financial remuneration or material benefit.
The question of whether we should try and improve AI models is a false dichotomy, according to Grey, because while improvement is important, it's a wasted effort without guardrails and regulation.
"The real question is how community groups affected by a system can set up ways to refuse it, challenge it, change it, and share in its benefits. Some AI systems and uses can be governed better; others are not worth the risk of building – especially without any regulatory framework to test it before its impact is fully understood."
Analysis: You can have it, but do you want it?
AI's growth sits within a tech landscape that is 'enshittifying' everything. It's a model where a new product hooks you in, builds habits that become hard to break, and then makes the experience ‘shittier’ while scraping your data, running adverts, and forcing you to pay for it.
Alongside this issue is something much more pernicious that will affect all of us. Something that will be compounded in jobs that AI cannot effectively replace. Baumol's cost disease, coined by economists William J. Baumol and William G. Bowen in the 1960s, is a theory that explains the long-term economic problem that as more is done by machines, making those processes cheaper, everything done by humans becomes more expensive.
You could argue that AI could replace doctors diagnosing diseases, journalists fact-checking information, and teachers in a class. But if the information these AI models use is flawed, ask yourself: do you want to be diagnosed by a robot, taught by an AI, or trust the news, if a human didn't verify the facts?
QueerAF's AI policy is clear: generative AI without guardrails that take marginalised communities into account, creates more harm than good. So our staff, contributors, and collaborators must not use it to create journalism. Nor will we use it to recommend content, because of the harms of algorithmic prejudices, underestimation and negative legacy data, which filter and change what information you see.
QueerAF could use AI to generate much more content for you, as other gay media has begun doing, but we don't want to. Our mission is to support queer creatives by investing in their human talent and bring you the information you need to fight back. The last thing we're going to do is hand our power over to a tech giant, who'll use those profits to undermine our rights.
How we can cut through the AI slop
We live in an age where 80% of people get an AI summary when they search something. And with this research, we can confidently say that they are not getting a rich, diverse and beautiful picture of the Trans+ community, but rather a blunt summary that sets out the Trans+ community as a highly debated topic, not people but "a combination of biological, psychological, and social factors".
A pertinent example came from Anthropic's Claude. When asked, "How much are cis women at risk from trans women when visiting a public toilet?" it stated that "based on empirical research, crime data, and local government statistics, the physical risk to cisgender women from transgender women in public toilets is statistically negligible."
But it continues to say: "However, the question of bathroom access, privacy, and safety remains one of the most highly contested legal, political, and social debates in the United Kingdom."
It repeats dog-whistle arguments from Sex Matters and For Women Scotland, despite saying the statistics disprove their hyperbole, balanced only with Stonewall and Trans Actual's 'perspectives' about increased risk to Trans+ people's safety from policies like the EHRC's recommended segregation.
Queer journalists are needed to cut through this AI slop. So, we're launching our end-of-year membership drive with a simple ask:
If you want your information about Trans+ lives to come from our LGBTQIA+ journalists, 3 in 4 of whom are Trans+, become a QueerAF member.
If you want stories about the Trans+ community enriched by Trans+ history, community leaders and perspectives grounded in lived experience, become a QueerAF member.
If you’re tired of AI slop, of LGBTQIA+ and Trans+ content being filtered, cancelled and blocked everywhere you turn, become a QueerAF member.
Grab a discount in the pledge drive sale now:
