OpenAI’s leadership must think carefully about the remaining board members they add, and not just to look progressive. They need women, people of colour and other diverse voices for whom biased language models are most likely to cause harm, and who will speak up about those risks