(Image Credit: rixie / 123RF Stock Photo)
Artificial intelligence (AI) and machine learning are both becoming more commonplace than ever. The good news is these technologies make people’s lives easier across the globe. However, the industry faces growing criticism that it doesn’t consider diverse perspectives when working with AI and machine learning. This goes beyond the obvious demographic factors of race, gender, and ethnicity that people cannon change and should include several other factors that are more changeable such as religion, education, and language. Failing to account for these fundamental differences in people means that technology will reflect biases by not considering the wider human experience.
As an example of how lack of diversity can play out with AI, imagine that a group of white male technology professionals in their 30s designed a personal assistant intended to make suggestions to the user. With such a homogenized group, it should come as no surprise that the answers provided by the personal assistant reflect that portion of the population. One particularly offensive example of this problem is when an image service from Google categorized all African-American people as gorillas. This example reflects a larger issue that people who develop AI and machine learning receive their training from Internet images of primarily white people.
The Three Most Important Dimensions of Diversity
Diversity in technology includes several factors, the most important of which are cultural diversity, human diversity, and systems diversity. The first category includes qualities that play a large role in who a person is but that he or she ultimately be may able to change. Some examples include ethics, religion, and language as well as working, thinking, and learning styles. Finding people to work on AI and machine learning teams with a diverse range of cultural diversity ensures that the technology is effective for the most people.
Human diversity refers to the non-changeable aspects of a person, such as ethnicity, gender, and race. Systems diversity considers how systems such as performance management, empowerment, and education interact together. When creating a new technology team, human resource representatives should consider whether they have overlooked any of the crucial diversity dimensions to the detriment of the team.
How to Incorporate More Diversity into Technology Hiring
Intent is a good start when it comes to creating more diverse AI and machine learning teams, but it only goes so far. Organizations must create a formal plan in writing to reach the goal of diverse hiring. Including some of these strategies helps take the concept from an idea to something actionable and measurable:
- Publish and then implement specific goals for the recruitment, retention, and advancement of diverse talent in technology
- Create specific measures to start and maintain a culture of inclusion
- Publish progress metrics and data on the diversity of technology employees across several departments and levels of seniority
- Create and invest in partnerships with educational facilities and other employers to develop a diverse technology pipeline
There’s no doubt that diversity in technology is the right thing to do from a moral perspective. It also makes sense from an economic perspective because products that reflect the broader culture will be more successful and sell better than those that do not.
