How can AI be bias-free?

The more diverse the better: A principle that applies not only to data, but to research teams as well. Without diverse viewpoints when flagging terms for content moderation, the resulting algorithms will be faulty.