Show Notes
What happens if we build something smarter than us and we can't control what it does next? In this episode, Grace breaks down the alignment problem: one of the most important and least talked about risks in AI. This episode, like all the others is for beginners. Talking through the risks in a clear, honest explanation. This episode will discuss why the people who built this technology are really worried and why it concerns to all of us right now. Covering the paperclip maximiser thought experiment, Geoffrey Hinton's warnings, and why being intelligent doesn't automatically mean being safe. Hosted on Acast. See acast.com/privacy for more information.
More from The AI Girls Club
Long before AI existed, the world treated the male body as the default. Crash test dummies were built on it, which is part of why a woman wearing a seatbelt is more likely to be seriously injured in a car crash than a man. Medical research was built around it too.
What actually happens when AI becomes capable of improving itself... Not the science fiction version, but the scenario that researchers at the world's leading AI labs are actively planning for right now.
In this episode Grace gets into the part of the AI story that needs to be spoken about more - the physical infrastructure behind every ChatGPT prompt, every Instagram scroll, every AI tool you have ever used. What these buildings actually are, how much water and energy they consume, what they are doing to the communities built around them and why the people making these decisions were never elected by anyone.
In 2023, an AI was caught making an illegal trade in a research experiment.
In this episode Grace breaks down what AI safety regulation means, what is already being done and what an independent assessment of the biggest AI companies in the world found when it looked at their safety planning.