Kars Alfrink pointed me to a report on AI Ethics by the Nuffield Foundation, and from it lifts a specific quote, adding:

Good to see people pointing this out: “principles alone are not enough. Instead of representing the outcome of meaningful ethical debate, to a significant degree they are just postponing it”

This postponing of things, is something I encounter all the time. In general I feel that many organisations who claim to be looking at ethics of algorithms, algorithmic fairness etc, currently actually don’t have anything to do with AI, ML or complicated algorithms. To me it seems they just do it to place the issue of ethics well into the future, that as yet unforeseen point they will actually have to deal with AI and ML. That way they prevent having to look at ethics and de-biasing their current work, how they now collect, process data and the governance processes they have.

This is not unique to AI and ML though. I’ve seen it happen with open data strategies too. Where the entire open data strategy of for instance a local authority was based on working with universities and research entities to figure out how decades after now data might play a role. No energy was spent on how open data might be an instrument in dealing with actual current policy issues. Looking at future issues as fig leaf to not deal with current ones.

This is qualitatively different from e.g. what we see in the climate debates, or with smoking, where there is a strong current to deny the very existence of issues. In this case it is more about being seen to solve future issues, so no-one notices you’re not addressing the current ones.