Here’s How AI Fuels Your Friends’ Racist Propaganda

[ad_1]

As if we didn’t have enough in-person racism to deal with, it turns out that chatbots could be racist too. Disturbing reports from the Washington Post have revealed that one of the most widely used datasets to train AI chatbots contains right-wing content.

Project 1619 is our Black TV Pick of the Week

For people unfamiliar with AI, AI programs like ChatGPT literally can’t think for themselves. Instead, companies are feeding AI programs with massive amounts of data pulled from all over the internet. AI uses this dataset to mimic human thinking. So if your robot friend starts trying to share 9/11 conspiracy theories, chances are the dataset has a little too much Alex Jones in it.

And that, my friends, is precisely where the problem begins. According to the Washington Post survey, the news websites used in one of the most widely used AI datasets include a ton of far-right and untrustworthy sources. The dataset in question is Google’s C4 dataset, which powers some of the largest AI models in the world, including Facebook and Google’s AI models.

So where exactly do they get their news from? Well, Breitbart is definitely on the list. Russian state propaganda website RT.com also has a presence there, alongside anti-immigration group Vdare.com.

40% off

Ninja Mega Kitchen System

A bit of everything This system has a huge 8 cup capacity and can blend, chop, pulverize and even make dough, which is more than we can do.

You don’t have to take our word for it to find out if Breitbart is pushing racism. In 2016, right-wing commentator Ben Shapiro expressed his contempt for the website, saying it pushed white ethno-nationalism content. And if you’re too right wing for Ben Shapiro… you might want to start asking some tough questions. A major concern is that AI programs don’t always cite their sources, meaning you could ask an AI a question and not know the answer came from a right-wing, hate-spewing site.

Sarah Posner of MSNBC, which covers the right, drew attention to the danger of having these entries in the algorithm:

Anyone who has searched the web for information on a topic knows that it can sometimes land them on a site spewing bigoted content or misinformation. The building blocks of chatbots were pulled from the same internet. An offended user may leave a toxic site in disgust. But because data collection for LLMs is automated, this content is included in the instructions for them. So if an LLM includes information from sites like Breitbart and VDare, which publish transphobic, anti-immigrant, and racist content, that information or misinformation could be incorporated into chatbots’ responses to your questions or requests for help.

The problem with AI (other than the inevitable day it takes over the world) is that it is the product of our own biases and judgments. And until we get a much better handle on this or at least put better guardrails in place, racist chatbots might be here to stay.

[ad_2]