Artificial intelligence, as we know, does not always work very well for people belonging to minorities. But the LGBT community is perhaps even more affected by this IT fairness issue. In any case, this is what four DeepMind researchers think, who have just published a scientific article on this subject.
One of the main problems would be that data on LGBT people may be non-existent, because not recorded – a lack of “registration”, which is not necessarily bad news … However, in the area of health, the authors explain, the research data used rarely includes LGBT indicators, whether through negligence or protection of privacy. An omission which, of course, does not allow us to create effective prediction models for these people.
The LGBT community can also suffer from bias induced by learning data. The risk is particularly present in recruitment. If the software used relies too much on historical data, it risks reproducing biased judgments of the past and disadvantages LGBT people.
This risk also exists for software related to language processing, such as speech recognition, translators or chatbots. Learning data can easily drag this software into abusive terms and stereotypes, when instead an inclusive way to deal with LGBT topics should be found.
Also to discover in video:
Also to discover in video:
Finally, some states could use artificial intelligence to identify, discriminate or censor LGBT people within a population, much like the Chinese government does for the Uyghur population.
Indeed, there are still many countries where LGBT identities and culture are not accepted or even repressed. For all these reasons, DeepMind researchers believe that a special research effort must be made to ensure that the artificial intelligence of tomorrow does not become a burden for this community. This research may also be useful for other minorities.
Source : DeepMind Scientific Paper
Source: Technos – 01net by www.01net.com.
*The article has been translated based on the content of Technos – 01net by www.01net.com. If there is any problem regarding the content, copyright, please leave a report below the article. We will try to process as quickly as possible to protect the rights of the author. Thank you very much!
*We just want readers to access information more quickly and easily with other multilingual content, instead of information only available in a certain language.
*We always respect the copyright of the content of the author and always include the original link of the source article.If the author disagrees, just leave the report below the article, the article will be edited or deleted at the request of the author. Thanks very much! Best regards!