"Alexa, play something from the 80s." 

How quickly have we become accustomed to having a small speaker in the house that we can ask for all the things we could just as easily have done ourselves. Artificial intelligence has been a part of our everyday lives for some time now—yet we can still expect some major breakthroughs ahead of us. Soon, AI will be performing surgical operations more meticulously than humans, it will give children a better education and it will take us wherever we want to go—under water, on the ground and in the air.

Already, AI is much more than just an unobtrusive, affable assistant intended to make home life easier. In the recruitment industry, AI is employed to ensure—thanks to its data-driven approach—that applicant shortlists are free from prejudices, are gender-neutral and culturally inclusive, which is often not the case when a prejudiced manager opts for the candidate who is culturally closest to them. 

But can AI really be trusted to do that? Is artificial intelligence truly free from prejudice?

Artificial intelligence is only as good as the society that created it.

It's not surprising that AI can only be as smart and unbiased as the data it is fed. And if that data is incorrect or incomplete, then you can no longer rely on Alexa and her comrades. This is also true when it comes to bias towards stereotypes or the mainstream. We have to assume that, following our request for hits from the 80s, Alexa will play Depeche Mode rather than a song that was playing on the radios in Tanzania at the same time. Which is why it is so vital to ensure cultural equality when delivering data to Alexa. There's just one problem: Not every society is equally documented and digitized. We cannot escape the fact that cultures that are less “present” or represented have a disadvantage when it comes to data administration. In comparison to western cultures such as the USA, they simply fail to exist and are therefore not taken into account by Alexa and Co.

Science is still a man's world.

Much has been discussed on the topic of gender equality in recent months. Medical findings are still based too heavily on data collected from the male body. Diagnoses, therapies and research itself are still very much a male preserve. As a result, female patients sometimes have to wait years for a diagnosis because there is little comparable data available to alleviate their suffering. 

The example of crash test dummies is often discussed in technology and industry. There is rumored to be only one company in the world that manufactures dummies with a female body for use as models. The consequences are clear: Safety in the car, safety on the road, general vehicle construction—everything is based on male test subjects. For women, the result is an increased risk in everyday life. Artificial intelligence, too, has little choice but to process data based on its biased and unbalanced inputs. Ultimately, AI emulates a man's world.

So despite her soft feminine voice, Alexa is ... a man? 

It's true! At her very data-driven core, Alexa is, in fact, a white male. Naturally, psychologists are quick to express their concerns and point out that Alexa is not able to break down gender clichés. On the contrary, Alexa only deepens the social rifts. Artificial intelligence cannot (yet) relieve a society of its commitment to socio-cultural progress. We have to first create an inclusive and equal world free from barriers and "teach" it to the AI machines. Each and every one of us.