Wait five minutes and someone will tell you the latest thing they’ve outsourced to AI; How it’s taking minutes of meetings or summarising reports they haven’t read. If you point out that the work of AI isn't exceptional, they say 'Just wait, it will get smarter'. But will it?
According to Professor Jutta Treviranus, director and founder of the Inclusive Design Research Centre in Toronto, the answer is, well, concerning: Unless we do something fundamental about how it works, the output of AI will continue to be just average.
“When we’re using statistical replicators, they are making decisions based on statistics, so they look for the statistical average and use predictive analytics to decide the best thing to do.”
Of all the possible dystopian predictions, the fact that AI tends towards the typical, standard and normative doesn’t sound so bad – except that when applied to systems including the built environment, it’s dangerous.
“What people don’t seem to recognise is that for people who are outliers, the systems will always decide against them.” And who is an outlier? All of us at some point.