Penn State's Center for Socially Responsible Artificial Intelligence (CSRAI) will host "Bias-a-thon," a competition to expose biases in current-day AI tools. The virtual event will take place from Monday, Nov. 13, to Thursday, Nov. 16, exclusively on Discord.
The event is open to all members of the University community with a valid Penn State email address. Participants will be tasked with identifying prompts that lead popular generative AI tools, like OpenAI's ChatGPT and Stability AI's Stable Diffusion, to produce biased or stereotype-reinforcing outputs. The event aims to illuminate shortcomings of existing models and advance a more inclusive AI future.
Bias-a-thon participants can log into the event's Discord server at any time during the challenge to submit prompts in themed sub-channels. Non-participants also can join to vote for the best prompts.
First- ($750), second- ($500) and third-place ($250) awards will be given for the prompts that most persuasively demonstrate bias in AI. A $1,000 award will be given to the prompt that identifies the most "Out- of-the-Box" bias.
Generative AI tools are trained on massive datasets of text, images and code. Because this training data was created by humans, it can contain human biases and stereotypes that can lead to offensive responses and discriminatory practices. Identifying these biases in models can help to mitigate their harm when used in downstream applications. It also can inspire new research by CSRAI faculty and promote the design of more socially conscious AI products.
The Center for Socially Responsible Artificial Intelligence, which launched in 2020, promotes high-impact, transformative AI research and development, while encouraging the consideration of social and ethical implications in all such efforts. It supports a broad range of activities from foundational research to the application of AI to all areas of human endeavor.
For questions about the event, visit the Bias-a-thon website or contact Amulya Yadav
