We should not fear the routine application of AI, and AI generally should not be found guilty of pre-crime. We should, however, take the opportunity to use AI to hold a mirror up to humanity, to improve ourselves.

AI is not a meaningful category to which new blanket rules — including ethical rules — should apply. To treat it as such would deny opportunities for innovation and use in relation to a promising new general-purpose technology.

That in turn would forego societal benefits, including potential lives saved; and that hardly seems ethical. It also risks entrenching power in groups of experts — equivalent to the psychics in Minority Report — who may unduly limit innovation, liberty and be insufficiently accountable both in terms of the trade-offs involved and politically.

Where we think higher standards should apply, we should apply them to all decisions, whether by machine or human; and that is likely to be an intensely political decision rather than a detached ethical one — since the lives of voters are directly involved. If we are going to contemplate a future challenge now, it should be this one.