definitions of fairness. This goes to show how nuanced and context-sensitive the concept of “fairness” is; its definition differs across cultures, across societies, and even over time.
Because there is no single, global definition of fairness, it’s important to thoughtfully consider what fairness means in the specific context of your use case. A key part of that work is describing the outcomes you don’t want. Without a thoughtful and robust approach to fairness at every step of an AI system build, that system could unintentionally produce unequal outcomes for individuals and cause them real harm .
Just how disparate were the results? The darker-skinned females saw error rates of up to 34.7%, whereas the error rate among lighter-skinned males was only 0.8%. The training dataset didn’t contain enough examples of darker-skinned females for the model to perform well for that group.Along the same vein, a training dataset could have societal and historical bias baked into the data.
googlecloud nothing.
googlecloud HectorVanegasF
googlecloud What happens when AI discovers that people behave in predictable patterns? That a single mistake is often indicative of a bad mortgage risk/credit risk/job applicant?
googlecloud Do we know what we mean ... by the 'I' in AI? ...
Technology Technology Latest News, Technology Technology Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
Source: BusinessInsider - 🏆 729. / 51 Read more »