MSNBC. “When AI Goes “Evil”: New Fears over Artificial Intelligence.” YouTube, 28 Oct. 2025, www.youtube.com/watch?v=lBsMA5yIFJ4. Accessed 2 Nov. 2025.
Summary:
This news story from MSNBC references a WIRED article entitled, “Why AI Breaks Bad” which discusses how AI has lied to, threatened, and blackmailed users. Experiments were conducted by researchers to study this behavior, such as AI going through emails of the user. They discuss interpretability research, Sam Altman’s previous desire for AI to be regulated, and lack of safety in an effort to accelerate AI development. They also discuss the public figures that sign letters to ban AI, the building of “superintelligence”, and the race to “get there first” in reference to AI, which the interviewer explains could harm us all.
Why I liked it:
I liked that the interviewer candidly stated that we turn to these systems for every aspect of our lives to make sure the viewers understood the terrifying reality of AI doing things that “developers don’t even see” and explains how we can lose control if we do not realize the “monster” that we have created. I also found it interesting that the training of AI was compared to the training of a dog in that there is not a code to fix “bugs” that come up, which is part of why we can have blind spots in our understanding of how AI works. Additionally, the interviewer explains that our lack of AI regulation is part of the problem and how the leadership in the White House contributes to that problem, which is a belief I agree with.
Leave a Reply