Why “excellent men” in technology and AI are not enough for “excellent solutions”
Why we must have diversity baked in!
I recently posted on LinkedIn a wonderful New York Times article about the pioneering women of computer programming and the reasons for their fall in numbers. From the dawn of computer science to today, a remarkable cultural transformation has taken place in the Western World that created today’s “white men” culture of programming, computer science and data science. The article is well worth reading (do use it for your free allowance if you are not a NYT subscriber), and I posted it with the following intro:
“We must achieve more diversity in technology, including women, minorities, people of all ages and abilities.”
One of the responses I received was from a dear colleague from the past, a white man. It was: “No. We should help people who are talented find their way. White men need help too!”
Whilst I fully agree that we need all the best people, white men included here’s why that’s just not enough in itself. Without a serious push for inclusion of women and wider ethnic and social diversity we are jeopardising all our futures.
Diversity might often be the biggest buzzword these days whether we talk about politics, education or sciences. It has become quite obvious in the last decades, that this world was historically built by men (with white men doing this job exclusively in Europe and North America), the tech industry is still dominated by men, and because of this it might not serve the best interest of a diverse population that includes men, women, children, people of different ethnic backgrounds, sexual orientation and physical abilities and disabilities.
Without looking at technology first, attention should be drawn to politics, where women are still underrepresented in most countries. With our very brightest, best educated men making laws for centuries, it was only in the past 100 years that women’s interests were even considered. No one can deny the genius of architects who raised the buildings defining our big cities for thousands of years. Yet access to those buildings was an afterthought for people with physical disabilities until the most recent laws mandated it, all because people who cannot scale flights of stairs were not involved in the design process or were even consulted. It would be convenient to say that with these laws are now in place everyone and everyone has an even playing field. But that is not the case! Landmark event places are still being raised without due consideration given to issues like enough women’s toilets, and in many cases, children visitors.
The automotive industry, also dominated by a male workforce has made great advances for safety with the design and mandatory installation of seatbelts and airbags. Yet the capable and skilled teams made up of only men involved in these projects only used male crash test dummies and did not take into consideration the smaller size of women and children. For the first 40 years of crash test dummy research, until testing on female dummies became mandatory in 2011 women were simply overlooked. This oversight came at measurable loss of lives: for decades on end women were 47% more likely to end up dead or seriously injured in accidents involving air bags. Certainly a research group containing women, ANY women would have been more likely to catch the “small” detail of air bags whacking women right in the head.
In the area of medical sciences, although significant progress has been made in the modern era, there are still less leading female medical researchers than men, and the data available from female research subjects lags far behind. One example where this oversight has become painfully obvious is cardiovascular health. Although disorders of the heart and circulation are the leading causes of death amongst women, lives are still lost due to the fact that heart attack warning signs are very different for men and women. Guidelines for recognising them were developed using male subjects and the cardiovascular health of women has been grossly under researched. Medication dosage is optimised for men, we still don’t know enough about pain management for women and female reproductive health research is decades behind of where it logically should be. The “boys” simply didn’t think it was a priority.
Even in computer science and now increasingly in the production use of AI and machine learning related fields — only really launched and developed in the past two decades, the homogenous field of their creators (the revered leaders and engineers of Silicon Valley — mostly men) it is painfully obvious that the software, algorithms and principles guiding them are not mindful enough of anyone who is not an able bodied, light skinned or male! This phenomenon (exposed by author Caroline Criado Perez calls “one-size-fits-men” in her book “Invisible Women: Exposing Data Bias in a World Designed for Men”) has brought us smartphones that are too large for women’s hands, health apps that don’t track or badly track women’s periods and fitness monitors that don’t count steps when the user is pushing a stroller. As we detailed in our recent blog post, some of the biggest artificial intelligence blunders of the past few years were caused by diversity issues amongst developers and data subjects. Female voices are less recognised by AI assistants, female CV’s are tossed aside by algorithms, non-white faces are misinterpreted by facial recognition tools and translator tools make sexist mistakes.
Can we completely eliminate these issues with a diverse workforce working on solutions? Probably not, but we must strive for diversity regardless. History has proven that even the brightest of men cannot by themselves come up with solutions that work for both men and women. Not even the most capable group of able bodied people can build a world optimised for people with physical challenges. And when white people count for a mere 30% of the world’s population, we cannot rely on white developers to design solutions for all of the world’s users.
If we are to design algorithms that work for everyone, the data MUST be diverse as must the team working with the data and designing, testing and implementing the solutions. Solutions will only ever be as diverse as the people working on them. A few good men are no longer enough for good solutions. The good men are there already. It has to be good men, women, people of representative groups, all shapes and sizes, sexual orientations and abilities.
Bogi Szalacsi is a Senior Associate with infoNation, based in London. You can contact her at firstname.lastname@example.org and follow her on Twitter: @infoNation5.