Questions remain.

• How DID the soap dispenser in this video pass QA, let alone user testing?

• Who was included (or indeed excluded), in the testing data?

• Technically, this likely occurred because the infrared sensor was not designed to detect darker skin tones effectively. However, this isn’t limited to AI. We’ve seen similar issues in medical devices including pulse oximeters (which use light to measure how much oxygen is in the blood), critical during the pandemic)), where non‑inclusive datasets and design decisions led to poorer outcomes for people with darker skin.

• While some of these issues are now being addressed and, to be fair, may be more a design and testing failure, rather than outright AI failure, the root cause is the same i.e non-inclusive teams, lead to non-inclusive data, resulting in non-inclusive outcomes.

• Critics argue that such incidents highlight the pervasive diversity challenges within the tech industry, especially when systems are built by third parties and then integrated into clients' supply chains.

This is why AI isn't solely a tech problem; it is a societal one and should concern all of us; morally, and because many of us serve and work with global, diverse customers and colleagues.

​

• Organisations therefore need to embed Diversity and Inclusion (D&I) across both their supply chains and their value chains. Unlike SaaS, AI permeates all aspects of an organisation. So even when people may not be aware they’re using AI, their tech vendors almost certainly are.

• This is one of the reasons why, by not having diversity across most departments, organisations could open themselves up to greater risk.