An interview with Rui Correia
“Is the notion of utterly banishing bias from AI a utopian ideal? Perhaps. Is it a social imperative and the moral obligation of AI technologists to make it our common Big Hairy Audacious Goal (BHAG)? Absolutely.”Daniela Braga, founder and CEO of Defined Crowd
Addressing gender bias and the disproportionate representation of women in the data and AI field is not only a problem for women to address. It’s something that society as a whole has to recognize, talk about and find solutions for.
That’s why, although we’re celebrating International Women’s Day, we thought it valuable and indeed necessary to get a man’s opinion on the challenges facing the tech industry.
Today, we’re talking to Rui Correia, a lead machine learning engineer at DefinedCrowd. Rui’s team works on developing machine learning models and techniques to improve the quality of data we supply to clients. Here’s more with Rui.
Reports show that only 22% of AI and data professionals are women. Have you noticed this disparity in your career, or during your studies?
Oh yes. While I was studying for my undergraduate degree, there were five women for every twenty men in a classroom. It is clear the issue exists!
Why do you think there are so few women in the industry compared to men?
I think because society has clearly defined the roles of each gender. From very early on, we learn stereotypes based on gender, and begin to take for granted that certain genders are more suitable to certain jobs. It’s done in a very subtle manner and from early on. We give little girls dolls to play with, while boys receive science kits or toy cars. We don’t teach girls in the same way we teach boys, so it becomes a structural issue. It also becomes a vicious cycle. Because there are so few women in tech roles, young women have no female role models they can look up to. Without female role models shattering conceptions, it is difficult to show young girls that they can do it too.
How do we solve this?
Education is important. We need to give young girls the confidence at an early age to pursue whatever dreams they may harbor. However, it’s also important to educate parents to not place expectations on their children according to gender. They must realize that gender is not a factor in success.
The private sector also has a role to play. Representation is key to social impact. When a young girl sees a female CEO doing amazing things, she will aspire to that; she will believe she can do it too.
You work in a team that is predominantly female. What’s that experience been like for you?
(Laughs) It’s definitely not something I think about. If there is one thing I can say, it’s that I am proud of the representivity of my team. Women make up the majority of my team, and I feel they are role models for the new students who complete work experience with us while they study towards their Masters degree. It’s really great.
Do you think there is gender bias in AI?
Yes, there is and sometimes it is not hidden at all. Sometimes, biases are profitable, and companies are not shy to exploit this fact. Adverts that target women or men specifically are obvious examples of intentional bias, but there are many more. I think intentional bias for profit serves to perpetuate the issue of gender bias in AI. Another factor to consider is that the AI used in production is typically a black box system, in the sense that the public they target do not know or are not told what variables the models take into account. Finally, models can learn bias when using real-world data, as the bias is in the data to begin with.
How can machine learning teams mitigate gender bias AI?
There are certain techniques for checking your algorithms or models against this type of phenomenon. I think the most important thing is to create accountability. By doing this, you are going to force the industry to create measurements or metrics that tell you if your algorithm is biased. You need to be held responsible for the actions of your algorithms. We have to break away from the black box systems and show people how the models work and what variables they take into account. The systems need to be transparent.
Why do you think addressing gender bias in AI is important?
It’s of great importance. We’re using technology to accomplish so many wonderful and useful things. We have cameras that can see at night and cars that drive by themselves. We are using technology to do things that we, as humans, cannot do. Likewise, bias is a human limitation, so technology can also help us here. If built and used correctly, AI can actually help us address bias systematically overcoming something we are flawed at.
Will we ever totally eliminate bias from AI?
We are able to mitigate some biases we know exist, but we also have to know that AI, at least for now, is made by humans. We may not even be aware of all the biases that exist. We are humans, and we create those biases, so it is a construction that is difficult to overcome.
Keen to hear more from our professionals?
Click on the images below to read more opinions and thoughts from our machine learning engineers and software engineers!