AI data mining and emails in discovery
In the legal world, emails can be used in discovery and as evidence. When will mining AI platform data to gather insights into participants' mental models become mainstream? How are AI inputs similar or different from email in discovery?
I'm not asking about how to leverage AI to conduct research or to improve AI models. I am asking about mining people’s entries/results to understand their concerns, behaviors, and emotional responses regarding their AI use (e.g. retirement planning, proposal writing, health queries, food prep tips, drafting emails, navigating parenting challenges, etc.). In sum, conducting primary research on their inputs and the actions that result. Here’s what’s on my mind:
Capturing Participants' Mental Models:
Analyze queries/interactions for themes/concerns that reflect participants' understanding/beliefs about specific situations, (such as historical facts, weather and geography, travel planning, recipes, etc.).
Categorizing these entries and responses into themes, identifying common mental models or misconceptions.
Implementation and Impact of Acquired Insights:
Subsequent surveys/interviews gather data on if/how participants utilized the info or advice obtained.
Explore changes in behavior or decision-making processes that result from these AI interactions.
Emotional/Cognitive Responses:
Investigate emotional impact of the info obtained from their platform on the participants, focusing on whatever we are studying
Assess cognitive shifts in understanding topics of interest, post-interaction.
Authenticity and Diversity of Responses:
How might we ensure that responses genuinely reflect individual participants' views and are not from shared accounts or devices?
Verification methods or demographic data to support the analysis of diverse perspectives.
Ethical and Privacy Considerations:
Prioritizing participant consent and the ethical use of the platform’s generated data (e.g. confidentiality and respectful handling of sensitive topics.)
Protocols for anonymizing and securely storing data for participant privacy.
Analytical Rigor and Methodological Considerations:
How to navigate the limitations inherent in self-reported data and the potential for bias in AI-generated responses.
How participants use these platforms to explore and process significant, specific issues might be extremely ifruitful. This approach might reveal more human complexities and emotions (assuming we can account for and acknowledge AI’s evolution as a tool for personal exploration and problem-solving.)
I'll happily consider “donating” my AI data to an org formally doing this type of research. I am fascinated by the potential and would love to learn more about it, or from anyone interested in this, or those actually doing this type of work!
Would love to hear your thoughts!
