As we all know too well, we have heard of stories across any media form about artificial intelligence performing some task it should not, or was not, programmed to do. But we are now starting to hear about AI making people think about things they should not, and weren’t intending to look for.
This all came in to fruition when YouTube implemented artificial intelligence to curate people’s advertisements towards what they had been clicking on or watching. Sounds like a great idea, right? If the web is going to track and sell ads to us, they might as well be ones we are interested in. Well, YouTube’s AI took it a bit too far.
If someone was watching a Trump rally on YouTube, advertisements for white supremacist rallies, holocaust denials, and other videos of that nature came up as advertisements or on the “Watch Next” tab on the other side of the screen. People watching videos about jogging then returned results and suggestions about running multiple marathons.
The problem here is that the AI machine was encouraging viewers to the most radical standpoint. In today’s world, we know this is exactly what we are trying to move away from – we do not need AI to lead us back in that direction even deeper than we already are today.
Julia Riccio – jmr17e – https://www.quantamagazine.org/artificial-intelligence-will-do-what-we-ask-thats-a-problem-20200130/
LIS 4022 Spring 2020 