AI Blindspot is a discovery process for spotting unconscious biases and structural inequalities in AI systems.
Archive (Page 1 of 2)
I am profoundly envious of people who get to write about settled domains or sort of settled states of affairs in human events. For me, I was dealing with a set of technologies which are either recently emerged or still in the process of emerging. And so it was a continual Red Queen’s race to keep up with these things as they announce themselves to us and try and wrap my head around them, understand what it was that they were proposing, understand what their effects were when deployed in the world.
Computers can tell stories but they’re always stories that humans have input into a computer, which are then just being regurgitated. But they don’t make stories up on their own. They don’t really understand the stories that we tell. They’re not kind of aware of the cultural importance of stories. They can’t watch the same movies or read the same books we do. And this seems like this huge missing gap between what computers can do and humans can do if you think about how important storytelling is to the human condition.
Increasingly we’re using automated technology in ways that kind of support humans in what they’re doing rather than just having algorithms work on their own, because they’re not smart enough to do that yet or deal with unexpected situations.
I teach my students that design is ongoing risky decision-making. And what I mean by ongoing is that you never really get to stop questioning the assumptions that you’re making and that are underlying what it is that you’re creating—those fundamental premises.
If you have a system that can worry about stuff that you don’t have to worry about anymore, you can turn your attention to other possibly more interesting or important issues.
In a world of conflicting values, it’s going to be difficult to develop values for AI that are not the lowest common denominator.
I think one of the things I want to say from the start is it’s not like AI is going to appear. It’s actually out there, in some instances in ways that we never even notice.
Machine learning systems that we have today have become so powerful and are being introduced into everything from self-driving cars, to predictive policing, to assisting judges, to producing your news feed on Facebook on what you ought to see. And they have a lot of societal impacts. But they’re very difficult to audit.