Slashdot Mirror
AI Programs Exhibit Racial and Gender Biases, Research Reveals (theguardian.com)
An anonymous reader quotes a report from The Guardian: An artificial intelligence tool that has revolutionized the ability of computers to interpret everyday language has been shown to exhibit striking gender and racial biases. The findings raise the specter of existing social inequalities and prejudices being reinforced in new and unpredictable ways as an increasing number of decisions affecting our everyday lives are ceded to automatons. In the past few years, the ability of programs such as Google Translate to interpret language has improved dramatically. These gains have been thanks to new machine learning techniques and the availability of vast amounts of online text data, on which the algorithms can be trained. However, as machines are getting closer to acquiring human-like language abilities, they are also absorbing the deeply ingrained biases concealed within the patterns of language use, the latest research reveals. Joanna Bryson, a computer scientist at the University of Bath and a co-author, warned that AI has the potential to reinforce existing biases because, unlike humans, algorithms may be unequipped to consciously counteract learned biases. The research, published in the journal Science, focuses on a machine learning tool known as "word embedding," which is already transforming the way computers interpret speech and text.
Most spoken languages exhibit a lot of bias. For example, Deutsch means people or folk, and that lightly implies what is not Deutsch is not people. A lot of languages have that mindset, and it's not surprising. Language evolved during times when people had values we disagree with.
This is my signature. There are many like it, but this one is mine.
Or global cooling, oops I mean global warming, oops I mean climate change, oops I mean government control.
Simple. 'decolonize' it.
https://www.youtube.com/watch?...
SocJus taken to its 'logical' conclusion. Reality is bigotry.
"Simple". The ML community is very aware of this problem, but sanitizing real-world data that may be shaped by subtle biases is really, really hard. You'd need a dedicated sociology PhD involved in every ML research project - a ludicrous load - and even then you wouldn't catch everything. This is a Hard Problem to be aware of for a long time to come.
Reality doesn't care about feelings or trying to make sure that outcomes are equal across groups, so we conclude that some group is a worse risk. It probably is
Except the latest interpretation of the Civil Rights Act by the courts is that Disparate Impact counts the same as direct discrimination. If your company adopts a policy that has a negative disparate impact on different groups, then it's deemed in violation of the law, so even if your AI is making a correct decision, it may be deemed racially biased by the courts, and require your company modify its policies to compensate for the bias.