We tend to think data is impartial. As a result, there was a time when many thought technology and AI could deliver a world free from human faults and failing. But from misogynistic chatbots, to biased hiring algorithms - we now know tech data mimics human prejudice with potentially serious consequences. One algorithm used in US court systems predicted black defendants much more likely to repeat offenses than they actually were.
Do we need to give up the idea that data can ever be impartial, and was Thomas Kuhn right when he said 'the answers you get depends on the questions you ask'? Should we train our technology on "perfect" synthetic data in order to create the society we want? Or is impartial data essential to our culture and society and something we must strive at all costs to deliver?