It’s commonplace these days to say that AI – artificial intelligence – can be a powerful force for good, but it can also do great damage if it isn’t very carefully watched.
We at the EU Agency for Fundamental Rights dug deep in there to see what actually, in practice, is needed to make sure that AI avoids the pitfalls and does us harm. We’re publishing now a new report on bias in algorithms.
The results, frankly, are startling – in terms of the extent to which AI makes mistakes, putting humans at risk – depending on the application, of course.
What have we learned?
Well, we’ve learned for instance that we need strong regulation. But we also need to ensure that the human stays in the story: humans must supervise very closely the application of AI.
And ultimately, maybe the most abiding lesson for us is that we must never cease to test, test, test AI in every possible application and use.
Have a look at our report – it’s just one further contribution from our agency towards unravelling what it takes for AI to be that force for good.