Purpose of Study

We aimed to comprehend which aspects of digital technology Black users found to consider their lived experiences and to emphasize potential pitfalls in the design of current digital tech that needed addressing. The study sought to understand the perception (interview), expectation (design fiction), and experience (observations) of Black users with (conversational) AI-supported text technologies.

Phase 1: Interviews + Observations, 45–60 Min

Phase 2: Speculative Design Workshops, 30 Min

Interview Protocol

Given the demanding schedule of conducting 6 interviews per week, we promptly dived into the development of the Interview protocol. I contributed by reviewing the draft crafted by the lead researcher, improving its structure, differentiating interview questions, and closely aligning them with the research question. To ensure the entire process was well-organized, we created the following procedure:

Welcome → Introducing Process → Privacy & Verbal Consent → Interview Contents (Warm Up → Establish baseline → Black Lived Experience → AI-supported text technology) → Observing Technology Use → Closing Script

In pursuit of more substantial data, I recommended that the team avoid close-ended questions with "Yes/No" answers and focus on more open-ended questions. Here are some sample questions:

• What are your thoughts on AI text generators, such as smart text assistants, chatbots, and ChatGPT?
• If there are opportunities to enhance AI text generators to be more inclusive of the Black community, what suggestions do you have for potential changes?
• In your opinion, what challenges faced by the Black community have AI text generators addressed?
• In what ways do you believe AI text generators can be destructive or serve as a hindrance to the Black community?

Design Fiction Workshop Protocol

In the development of the design fiction workshop protocol, I offered essential instructions to participants to enhance understanding and improved time management. This adjustment was prompted by my observation during interview sessions, where some individuals faced hard stops or were less willing to participate beyond the scheduled time.

Note Taking

I took on the role of the note-taker in 5 interviews, documenting participant experiences related to each section of the interview, alongside an interviewer and a participant. Following the interviews, I transcribed the audio into text, facilitating the process of data analysis.

Code the Findings

Raw Coding to Affinity Mapping

To comprehend participants' perceptions and experiences with ASWT, I conducted a thematic analysis of the interview data collected by our team. After cleaning the Zoom audio transcriptions with Otter.AI, I proceeded with the inductive coding of two interviews. I performed blind coding for each of the selected interviews, generating an initial list of codes. The next step involved gathering all produced codes and merging similar ones, utilizing an affinity map to create larger coding groups.

Codebook

To comprehend participants' perceptions and experiences with ASWT, I conducted a thematic analysis of the interview data collected by our team. After cleaning the Zoom audio transcriptions with Otter.AI, I proceeded with the inductive coding of two interviews. I performed blind coding for each of the selected interviews, generating an initial list of codes. The next step involved gathering all produced codes and merging similar ones, utilizing an affinity map to create larger coding groups.

Conference Paper

The major output was a paper highlighting the urgent need for ASWT to evolve into a more inclusive, nuanced, and culturally aware technology. Participants expressed a desire for their language and culture to be respected, understood, and celebrated by ASWT. They emphasized the crucial role of technology in addressing complex issues related to language, culture, and inclusivity rather than exacerbating them.

The insights gathered from participants suggest a way forward: the development of technology that respects and adapts to diverse linguistic and cultural expressions, promotes language autonomy, and strives to understand, rather than merely imitate, the rich tapestry of human communication. Overall, the study advocates for a more considerate and culturally sensitive approach in the design and implementation of ASWT to ensure fair and equitable outcomes for all users.

Where is AI Headed?

The significance of the study lies in its potential to serve as a blueprint for addressing text misclassification due to dialect differences for minority and multilingual dialect speakers. It also emphasizes the importance of building trust in AI and raises ethical considerations related to manipulating trust factors to enhance the user experience.