Possible dangers of artificial intelligence that can make sexual judgments

Recently, two researchers at Stanford University, Yilun Wang and Michal Kosinski, created an artificial intelligence system that can judge whether a person is a homosexuality simply by observing a person's photo.

However, whether it is the current research results or the future direction, this research has caused great moral controversy. Kosinski mentioned: "We are very upset about these results and have spent a lot of time thinking about whether they should be made public."

They eventually published the paper, and then it caused an uproar.

Some people blame this study as "pseudoscience" because the study is limited to white people in the United States and cannot judge whether a person likes bisexual dating or threesome love.

Kosinski retorted in a note: Although this study does not prove that the same conclusion applies to other races, we find that the conclusion is applicable. The same biological, evolutionary, and cultural factors contribute to the formation of differences between homosexuals and heterosexuals, and these factors are likely to affect other races. The failure to identify bisexual relationship will not invalidate existing conclusions.

But more discussion has focused on whether the study violates the sexual orientation of others.

Some people think that if you intend to become friends or establish a closer relationship with one person, then it is necessary to know each other's sexual orientation. An author of the Washington Post wrote an article saying: "I used to meet a man for the third time before I realized that he was planning to date me." I didn't notice this romantic trend in time. I feel very embarrassed.

More people think that a person's sexual orientation should be his or her privacy, no one can know at will; in some places, if the sexual orientation is disclosed, it may have serious consequences.

Someone sent an angry email to the researcher: "You must know that in some countries, homosexuality is a crime. So I think you are a homophobic bastard who supports the murder of homosexuals. If not, please destroy all work related to the topic, otherwise, I hope to be able to personally kill you, because your work will cause many people to suffer and even die."

Kosinski responded to these warnings and threats, indicating that they did not create a tool to invade privacy. The research work is limited in content and only studies people who claim to be homosexual or heterosexual, but he also acknowledges these technologies do have such risks (invasion of privacy).

If this artificial intelligence system spreads, your photos may be tested without permission. After all, in this era, users who want to "internet mass hunting" is not difficult to get a few photos from the internet. Think about it: If you have not married at the age of marriage, and then someone takes your photo to test, it just happens that you may have a homosexual orientation, then you may be entangled in gossip.

Xiaoyan also wants to extend the discussion: We said that the research method can be extended to detect emotional intelligence and even criminal tendencies. If a person's information in these aspects can be judged by artificial intelligence, it will be dangerous for normal people?

Artificial intelligence has obtained some general conclusions through a large amount of data analysis, and then applied the conclusions to new individuals. This practice is also common in human society, but it is not advocated.

A study by the Center for Information Technology and Policy at Princeton University found that machine learning and AI algorithms inadvertently reinforce and expand established prejudices in society or the subconscious of users. For example, the sexual orientation study above, if widely used, makes it easier for people with soft temperament or handsome women to associate with homosexuality.

In 2013, a Harvard study also found that African-American names are more likely to be paired with wanted orders. If the data enters the artificial intelligence system, then when judging a new sample, an African-American name will be more likely to get a "criminal tendency" prediction. This is already suspected of racial discrimination.

But artificial intelligence itself is not guilty. It is like a mirror that reflects the true side of human society. This real side has been hidden in the United States after decades of baptism. Everyone just keeps saying it, but data revealed it. If we don't take the initiative to correct "stereotypes," artificial intelligence as a tool will reinforce these impressions. In the final analysis, the subjective initiative is in the hands of human beings, and the advancement of artificial intelligence, like nuclear weapons, is indeed likely to be used by users to do bad things.

Therefore, artificial intelligence can accurately determine your sexual orientation. Do you think this is a dangerous thing?