ArticleAugust 24, 2022by included

Is artificial intelligence biased?

We often think that algorithms and machines can’t be biased. But, this technology takes on the bias of those who built it. It’s crucial that diverse, inclusive teams are designing and innovating in the tech space to ensure bias is not carried through to the final product.

Tackling bias is a central to Included’s purpose. We exist to build more meritocratic workplaces and an essential step of ensuring merit is judged equitably is addressing our own biases as well as those we then build into technology and systems. We’re supporting Forever Beta’s campaign to tackle embedded bias in AI. Take a look at these images that show how AI is reflecting existing stereotypes back at us.

The words 'leader', 'employer', 'CEO', 'business person', 'company dirctor', 'city worker' business owner' and 'professional' generate images of white men in suits.

It’s crucial that diverse, inclusive teams are designing and innovating in the tech space to ensure bias not carried through to the final product.

Why is this important?

The tech industry has the opportunity to build products that both generate revenue and can even change the way we live. Those who are investing in tech have the ability to determine how inclusion manifests in this space. If they choose to build diverse teams from the very beginnings, these teams can go on to crack new markets and build innovative products and solutions. Without this, the lack of diversity in a company will be felt by the end users once the product lands. We can see this in Kodak’s use of “The Shirley Card” that used a photo of a white woman and therefore built bias against Black people and people with darker skintones into the way their products functioned. We can also see in this automatic hand dryers that take longer to respond to hands with darker skin tones than with lighter.

Increasingly, technology, AI, machine learning, and algorithms are part of how the world is run and how we operate in the new world of work. It’s a common misconception that these machines will not be biased. But, we will all naturally have unconscious bias as humans and therefore will include this into how we build technology, especially if the technology development team is not diverse. As we use this technology more and more, we will see a greater impact through people not being able to access technology in the same way and through stereotypes being entrenched such as what we have seen in the AI created imagery.

How does it tie into Included’s work?

Groupthink is a significant barrier to long-term growth. organisations must include diversity of thought in their teams. Tech companies can often be born of like-minded people, then meaning that as they grow they are in a ‘diversity debt’. Proactive inclusion is a key way to address this. In Included’s work with Tech Nation, increasing the understanding of D&I in tech companies was key. We ran internal Inclusive Leadership Labs around the UK and then co-created an incubator for tech start-ups to access D&I training free of charge. As a result largely male tech start-ups started to consider D&I in their scale plans for the first time.Additionally, in our work with Cloudflare, worked on embedding inclusion into their product design to drive innovation. After this, the organisation also took the profound step of starting a conversation about and then reforming their coding languages and terminologies, removing racial or offensive wording.