Purpose: The aim of this project is to provide resources that can help in recreating experiments on rating of AI services. See more details of ideas, papers and patents at: https://sites.google.com/site/biplavsrivastava/research-1/trustedai
Directories
- BiasSpecReusable: machine-readable specification of bias. Provides specification of group bias used in work around language translators
- TranslatorDataShare: data for rating of translators for gender-bias
- ChatbotRatingShare: rating of conversation agents (chatbot) for trust. Provides experimental setup to combine ratings of 4 issues related to a chatbot's trust.
Credits Biplav Srivastava (2017-), Francesca Rossi (2017-), Mariana Bernagozzi (2019-) and Sheema Usmani (2019-) have contributed to the ethics rating work.