As this week comes to an end, I am officially at the halfway point of my LiA. 

For my work with Data Pop Alliance, I carried out tasks closely resembling my research from last year, which involves testing ChatGPT. However, instead of measuring ChatGPT’s gender bias levels in its translations, I focused on how the wording of prompts can affect ChatGPT’s content in its responses. While I don’t want to get too nitty gritty into the details, I found that ChatGPT does alter its response to mimic the speech pattern of the user’s prompt, which is good especially in creating a more accessible and user-friendly environment. However, for certain prompts, ChatGPT provided me with very different recommendations and information when I changed the formality of my language. While this is not necessarily wrong, this has the potential of creating bias, particularly relating to education lines and accessibility to information.

I was also able to work with the GLBT Historical Society, an organization that “collects, preserves, exhibits and makes accessible to the public materials and knowledge to support and promote understanding of LGBTQ history, culture and arts in all their diversity.” For a couple hours this week, I was also able to help with secretarial work in the organization’s museum, by helping customers with purchasing tickets, answering emails, and tidying the space up. I was also able to receive my training for their auction fundraising event that I am helping with next week.

Next week, I will be focusing more on the Senegal Flood project at DPA, and I hope that I can begin to outline my final report for them. But overall, this week was productive and I am having a great time here in San Francisco!

Here are a couple of pictures of the GLBT Museum