What is gender bias in artificial intelligence (AI)? What contributed to the creation of that bias?

Gender bias is one of the forms of bias that infiltrates AI because it is created by humans. This bias does not generally stem from bad actors who want to create racist or sexist programs.   But implicit bias forms our opinions and enables us to act quickly. As a result, code inherits these human failings. Coders draft and input labels according to their own experience and perspectives; “deep learning” programs learn from scripts infused with this bias. Those programs build their own bias they've learned.

What are companies doing today to combat gender bias in AI?

Awareness of bias in AI is a recent development. Companies that have been burned by releasing AI-enabled products in which bias was later detected are some of the companies at the forefront. They are setting up guidelines and standards or they are funding studies to better answer this question. Sadly, the majority of companies have yet to realize that this problem could impact them in a significant way.

How does gender bias in AI negatively affect one’s business and employees?

Perhaps most notable are incidents where companies have released products that insulted or failed to serve potential consumers. When facial recognition tools fail to recognize black or Asian features, these products cannot be used by entire continents of potential consumers. There is also the human impact. Employees want to be a part of building products that add value. Any evidence of harm to consumers or the brand’s integrity invariably impacts morale.

What steps can businesses take to become more aware of biases within their hiring processes?

There are a variety of precautions that should be taken in both hiring and promotions within companies. A growing number of companies rely on AI systems to identify candidates, select resumes and conduct initial rounds of interviews. Companies should ask what steps are being taken to avoid infusion of bias in the AI and test out AI products to ensure satisfactory results. Companies should also work with those responsible for hiring and promotion to ensure avoidance of selection and confirmation bias.