The Queer in AI workshop at NeurIPS asks its participants to question the status quo of machine learning research and applications in society, in a world ravaged by queerphobia, heteropatriarchy, corporate hegemony, racial disparity and global economic inequality. The Queer in AI membership survey shows that nearly 70% of queer scientists are not publicly out and many have faced discrimination or even violence due to their existence. We call on our community to face these challenges head-on, to advocate for and build a future where technological progress empowers marginalized people and does not ossify the status quo of the past and present.

In the last months, large language models have made impressive progress on well-established AI benchmarks, to a point where some believe that the continuous increase in the size of language models can bring about AGI. This vision ignores the real and well-documented harms that this paradigm presents for marginalized communities and concentrates power in the hands of corporations and other actors who can afford to collect data and scale systems without considering the lives of those impacted. It also silences those who work diligently to explore the weaknesses, problems and short-coming of these models.

As such, it is imperative to bring this workshop to NeurIPS and invite all participants of the conference to reflect, discuss and build a future of equitable AI for people of all identities and backgrounds, beyond simplistic questions of the raw scale of computational resources and data. In this year’s workshop, which will be held both virtually and in-person, we want to focus in particular on the ongoing tension of ethical AI research at large corporations whose profit motives are intrinsically linked to AI, and on the impact of AI technology on people with intersectional identities, such as queer neurodiverse people and queer people of color.