Artificial intelligence (AI) is bringing more dangers as it continues to grow. An author argues we aren’t keeping up, but society needs to.
Darren McKee, ArtSci ’03 and MPA ’08, published his book Uncontrollable: The Threat of Artificial Superintelligence and the Race to Save the World on November 17, 2023. In the beginner-friendly book, he explores the dire consequences of AI while offering ways to navigate the evolving technology.
“[The book is a] look at the promise and peril of artificial intelligence. There’s a lot going on in the news about AI and the risks and capabilities it may have. This is a book that’s trying to put it all together in one spot in a very easy to read manner about what’s going on, how AI works, and why it might be a risk,” McKee said in an interview with The Journal.
McKee started writing when he became alarmed about AI’s rapidly developing capabilities and lack of public awareness. Not only was there a gap in understanding, but there was also no current literature that spoke to AI’s development and need for navigation.
“These systems are getting more and more capable and there isn’t currently an up-to-date book about AI safety and risk that’s built for everyone. No technical background, no science background,” he said.
According to McKee, deepfake pornography, racist and sexist algorithmic biases, and voice cloning software are pressing concerns that merit immediate attention.
“AI systems themselves are already very sophisticated. They can demonstrate things like deception, or manipulation and it is possible that the systems themselves will engage in what we call power seeking behavior,” McKee said.
“It doesn’t have to be intentional at this stage. It’s just a way of solving a problem. The system isn’t initially motivated by greed. It’s just trying to achieve a goal and it doesn’t have normal human constraints in mind.”
A large issue with AI tools is the potential lack of reliability while displaying confidence in their answer output, regardless of whether the information is true.
McKee offered an example of someone asking ChatGPT about sexual harassment cases against American law professors, and for the AI to include citations. The chat drew information about a professor being accused of sexual harassment while on a school-sponsored trip to Alaska, followed by a citation from The Washington Post.
“The issue is none of it’s true. The professor wasn’t associated with that university. He never went to Alaska. There’s no charge against him, and there’s no article, but the response the system gave made it seem like this was a very legitimate, real thing,” he said.
Becoming more involved and informed about AI is crucial to safeguarding against ill effects. McKee said talking to friends, family, and political representatives about mitigation strategies are specific ways to protect oneself. Broader suggestions include liability implementation for harmful AI practices and increased transparency regarding when, where, and how AI is being used.
It’s inevitable people will use AI, and they should, according to McKee—it’s simply about learning how to navigate the systems and avoid the associated risks.
“At a certain age we don’t want kids to use calculators because they should learn how to do math. That’s a good idea. But later in life, if we’re not having them use calculators, they’re at a huge disadvantage. Similarly, while AI systems used in the wrong way could reduce one’s learning or growth, it’s also true these are tools younger generations will have to use,” he said.
Tags
All final editorial decisions are made by the Editor(s) in Chief and/or the Managing Editor. Authors should not be ed, targeted, or harassed under any circumstances. If you have any grievances with this article, please direct your comments to [email protected].