We tend to believe that smart machines are always objective, cold and unbiased in their decisions – be it self-driving cars, assessing credit scores, or classifying images in photo albums. We trust these systems to be genuine and the metrics to be complete and impactful. The truth, however, could not be further from this.
The lack of diversity in human brains behind these algorithms has resulted in multiple sources of bias. Many scholars believe that the Artificial Intelligence (AI) and Automation industries together today are a “Sea of Dudes,” mostly white dudes. This homogeneity in the back-end data has reached a very alarming stage.
To understand the evolving gender trends in Automation and AI industries, we need to first understand the difference between the two. While automation is the process devised to run without or with very little help of human “thinking”, AI is the science and engineering of making intelligent machines that can imitate human behavior and intelligence, and evolve with changing environmental parameters.
Every smart device’s output is a result of either the data received in the backend by its human brains or data received through interaction.
Automation has been less of a victim of gender biases than AI due to its inherent programming of choosing suitable option from a list of predefined choices, as opposed to AI which functions on evolving choices with human and environmental interaction. Nevertheless, both have been guilty of reflecting our biases.
Automation and AI have rapidly become an integral part of the 21st century society. Robots are predicted to replace 47 percent of U.S. jobs, according to a study by the Oxford Martin School. Developing world countries such as Ethiopia, China, Thailand, and India are even more at risk. However, the good news is: being in incumbent state, we still have time to address their inappropriate approaches towards gender definitions. And the only way to do this is by understanding bias themselves as the source of problems, and actively designing systems to avoid them.
The severity of the issue can be gauged from the fact that only 17 percent of computer science graduates in the U.S. today are women, down from the peak of 37 percent in 1985. The figures are worse in AI. This inadvertent gender bias in the workforce of AI and Automation is resulting in a kind of myopia. “If everyone teaching computers to act like humans are men, then the machines will have a view of the world that's narrow by default,” opines Bloomberg in a recent article titled Artificial Intelligence Has a ‘Sea of Dudes’ Problem.
The lack of diversity has been resulting in very close-minded outcomes. Soaring examples of these are Siri and Cortana. Biology which predisposes us to prefer the sound of woman since the time we are in a womb must have been an integral part of their gender definition, but the logic somehow is very ill-equipped to counter the straight up sexism that argues, “Assistants by default mean women.” Fortunately, IT giants have rather been quick in accepting the criticism and identifying the bias but still have a long way to go to rectify.
AI and automation can facilitate closing the gender gap, as long as it is not vulnerable to human bias.
And as Stephen Hawking puts it, “The real risk with AI isn’t malice but competence. A superintelligent AI will be extremely good at accomplishing its goals, and if those goals aren’t aligned with ours, we’re in trouble.”
Photo credit: iStock