AUGUST 2023
This was a group, solutions-based project. The objective was to explore and develop a creative solution to an issue that came up during the course (Meaning of Information & Technology).
SUMMARY
Our particular solution was a response to the issues around AI-generated content with respect to bias in the data (amplification of bias in the output), lack of transparency to users, and algorithms optimized only for engagement or "accuracy" with no quantitative way to integrate the human values that may be of importance to the user and the output processing.
Our solution concept combined tools that already exist, such as bias auditing algorithms, Harmful Content filters, and integrates these with a human-values integration interface for the user as well as a transparency report for each input/ output of an AI generated platform such as ChatGPT .
We also generally explored tools that already exist for purposes that would/should require a human values component to the generation.
Being able to properly define a human value is tricky, but a vital step to the process. For this, we chose to adopt the standard Social Worker definitions of these values, generally, because they are in many ways interaction specialists.
PowerPoint File:
Source Summary & Research Details:
Presentation:
* All research, development, and presentation was done by myself, Cole Metcalf, and Sena Uctuk
Комментарии