Racist Soapdishes and Rebellious (?) Children: Towards Human/AI Cooperation

Fiction is replete with tales of machine minds overthrowing their human creators. These archetypes are powerful, and durable, but they must be confronted and challenged if we are to truly address the real issues that the wholesale introduction of AI presents. This paper will argue that the problem lies not with AI in itself, but in its ability to act as a force multiplier for our own innately human biases and prejudices; as Safiya Noble and Caroline Criado Perez among others have shown, the underlying algorithms driving much of today's AI are based on entirely prejudiced assumptions, and build racism and sexism into the automated systems on which the modern world relies. As with so much within the cyber domain, the problem is not technological, but human, and it requires a human solution. What will be argued here is that greater use of AI is both inevitable and potentially hugely beneficial, but that it requires close examination of the underlying assumptions on which the fundamental programming of machine intelligences is/will be based. More than that, the paper will suggest that a truly successful future will lie in humans working with non-human intelligences, rather than in their exploitation.