A Data-Driven Strategic Model of Common Sense in Machine Ethics of Cares

摘要

When adopting artificial intelligence in organizations, we face machine behaviors that are problematic ethically. Tay, a chatter robot in Twitter, learned what and how to speak from twitter users without having ethical common sense. Eventually, Tay was shut down after it tweeted segregative and violent words to people. Besides, Amazon's AI recruit system showed a sexist behavior that it preferred male than female candidates. The present studies focus on how to apply the ethics of justice to artificial intelligence by feeding data on standards and rules in our society. However, we claim that the ethics of cares is also necessary to have artificial intelligence in our daily life. While one-way data is necessary in the ethics of justice, two-way data is required in the ethics of cares. Namely, one learns how to care others by having feedback from others after taking an action. The one learns if the action is offensive to others or not.