New large-scale learning model shows how AI will shape LGBTQIA+ advocacy

AI Comes Out of the Closet is a large-scale learning model (LLM)-based online system that leverages artificial intelligence-generated dialogue and virtual characters to create complex social interaction simulations. These simulations allow users to experiment and refine their approach to LGBTQIA+ advocacy in a safe and controlled environment.

The works are published below Proceedings of the 29th International Conference on Intelligent User Interfaces..

This research is personal to first author D. Piris, an MIT graduate student in media arts and sciences and a research fellow in the Tangible Media Group at the MIT Media Lab, as it is rooted in the contexts that LGBTQIA+ people continue to navigate. And it's political. The complexities of identity, acceptance, and visibility. Piris' work is developing advocacy simulations that not only address the current challenges facing the LGBTQIA+ community, but also leverage the potential of AI to provide innovative solutions that build understanding, empathy, and support. Driven by necessity.

This project aims to test the belief that technology, when applied thoughtfully, can bridge the gap between diverse experiences, foster a more inclusive world, and become a force for social good.

Piris highlights the important but often overlooked connection between the LGBTQIA+ community and developments in AI and computing. “AI has always been queer. Computing has always been queer,” he said, starting with the story of Alan Turing, the founder of computer science and AI who faced legal penalties, and of queer people in the field. The contribution of is attracting attention. Chemical castration for his homosexuality.

Contrasting Turing's experience with the present, Piris points to OpenAI CEO Sam Altman's open acceptance of his queer identity, illustrating a broader shift toward inclusivity. This evolution from Turing to Altman highlights the influence of LGBTQIA+ individuals in shaping the AI ​​field.

“There's something about queer culture that celebrates the artificial through kitsch, camp, and performance,” Piris says. AI itself embodies a constructed, performative quality that deeply resonates with queer experiences and expressions. Through this lens, he argues for recognition of the strangeness at the heart of AI, not only in his history of AI but also in its nature.

Piris found a collaborator in Pat Pataranutaporn, a graduate student in the Media Lab's fluid interface group. As is common at the Media Lab, the partnership sees Pataranutaporn's work on AI characters meet with Piris' focus on 3D humans and his simulation of the institute's interdisciplinary pursuits. It started in culture.

Tackling the challenge of interpreting text into gesture-based relationships was a major technical hurdle. Pataranutaporn's research focuses on creating conditions in which people can thrive, rather than just solving problems, and how AI can contribute to human flourishing across dimensions of “wisdom, wonder, and happiness.” aims to understand.

For this project, Pataranutaporn focused on generating dialogue that facilitates virtual interactions. “It's not just about making people more efficient and productive, it's about how we can support multiple aspects of human growth and development,” he explains.

“AI offers significant new opportunities to support human learning, empowerment, and self-development,” said Patty Mays, the Germeshausen Professor of Media Arts and Sciences at the MIT Media Lab and an advisor on the project. We are proud and excited about the support this research will provide.” Rather than aiming for AGI, we aim for AI technology that benefits and enables people and humanity. [artificial general intelligence]. ”

Address pressing concerns at work

The urgency of this project is underlined by research showing that nearly 46% of LGBTQIA+ workers have experienced some form of unfair treatment in the workplace, from being ignored in employment opportunities to harassment. Approximately 46% of LGBTQIA+ individuals have to hide their identity at work due to concerns about stereotyping, potentially offending co-workers, or potentially jeopardizing professional relationships. I feel that.

The technology industry, in particular, presents a difficult situation for LGBTQIA+ individuals. According to the data, 33% of gay engineers perceive their sexual orientation as a barrier to career advancement. Additionally, more than half of LGBTQIA+ workers report encountering homophobic jokes at work, highlighting the need for cultural and behavioral change.

“AI Comes Out of the Closet” is designed as an online study to evaluate the impact of a simulator on developing empathy, understanding, and advocacy skills for LGBTQIA+ issues. Participants were introduced to an AI-generated environment that simulated real-world scenarios that LGBTQIA+ individuals might face, with a particular focus on the dynamics of coming out in the workplace.

work on simulation

Participants were randomly assigned to one of two interaction modes with a virtual character: “first person” or “third person.” In first-person mode, participants step into the shoes of a character navigating her coming-out process, creating personal involvement in the simulation. Third-person mode allowed participants to take on the role of observer or director and influence the storyline from an outside perspective, similar to the interactive audience in the Forum Theater. This approach was designed to investigate the effects of immersive and observational experiences.

Participants were guided through a series of simulated interactions, where virtual characters powered by advanced AI and LLM responded realistically and dynamically to participant input. The scenarios included key moments and decisions and depicted the emotional and social complexities of coming out.

The scripted scenarios in this study provided a structure for the interaction between the AI ​​and the participants. For example, in a scenario, when a hypothetical character discloses her LGBTQIA+ identity to her co-worker (represented by the participant), the co-worker can advance the conversation with multiple-choice responses. These options are designed to depict a range of responses, from supportive to neutral to even negative, allowing research to capture a variety of participants' attitudes and reactions. can.

After the simulation, participants were asked a series of questions aimed at gauging their level of empathy, sympathy, and comfort with LGBTQIA+ advocacy. These questions aimed to reflect and predict how the simulation might change participants' future behavior and thinking in real-life situations.

result

This study found interesting differences in how the simulation affected empathy levels based on third-person or first-person mode. In third-person mode, where participants observe and direct behavior from the outside, research has shown that participants feel more empathy and understanding towards LGBTQIA+ people in “coming out” situations. This suggests that by observing and controlling the scenario, we can better relate to the experiences of LGBTQIA+ individuals.

However, in first-person mode, where participants acted as characters in the simulation, empathy and the ability to support others did not change significantly. This difference suggests that the perspective we take can influence our reactions to simulated social situations, and that being an observer may be good for increasing empathy. I am.

Although the increase in empathy and sympathy within the third-party group was statistically significant, this study also highlighted areas that require further investigation. For example, the impact of simulation on participants' sense of safety and confidence in LGBTQIA+ advocacy situations has shown mixed results, indicating the need for deeper investigation.

The study also acknowledges the limitations inherent in its methodology, including its reliance on self-reported data and the controlled nature of the simulation scenario. Although these elements are necessary for the initial exploration of the study, they suggest areas for future research to validate the results and extend them. Future research may include additional scenarios, investigating diverse participant populations, and longitudinal studies to assess the lasting impact of the simulation.

“The most telling surprise was that there were many people who were accepting of LGBTQIA+ interaction in the workplace, and many who were against it,” Piris said. This stance highlights a broader trend in which people are accepting of her LGBTQIA+ identity but still don't fully recognize the importance of that experience.

Possibility of real-world application

Piris envisions multiple opportunities, such as simulations built for research.

In human resources and corporate training, simulators can serve as a tool to promote an inclusive workplace. By empowering employees to explore and understand the nuances of their LGBTQIA+ experiences and advocacy, companies can create more empathetic and supportive work environments, increasing team cohesion and employee satisfaction. You can increase it.

For educators, this tool has the potential to offer a new approach to teaching empathy and social justice and embed it into the curriculum to prepare students for the diverse world they live in. For parents, especially parents of her LGBTQIA+ children, this simulator may provide important insights and strategies such as: We support children through the coming out process and beyond.

Healthcare professionals can also benefit from simulator-based training to better understand the LGBTQIA+ patient experience and improve care and relationships. Mental health services in particular could use this tool to train therapists and counselors to provide more effective support to LGBTQIA+ clients.

In addition to Maes, Piris and Pataranutaporn were joined in the study by Misha Sura of the University of California, Santa Barbara.