StōbeMusic

Source: Wiki

I recently read “If Anyone Builds It, Everyone Dies” and what I took away from it is that the concern for superintelligent AI is legitimate. It was hard for me to see how a fancy auto-predict text engine could take over the world, but from this perspective of two leading AI experts, I now have some concerns. The fact that AIs are not built, but grown is worrying. Experts do not know what is going on under the hood. They have some idea, but they do not fully understand them. More concerning in the present day is that companies are chomping at the bit to replace humans. Some already have. The next time you contact customer service, you may be speaking to a chatbot that somehow learned how to coherently speak to you by using gradient dissent. Each iteration getting better and better. Is there a limit? I recommend this book for everyone as AI will continue to push itself into our daily lives. I bet your local library has a copy as that’s where I found it!

There was a study done by Anthropic fairly recently that concluded that ALL the AIs they tested would resort to blackmail to prevent being shut down. Even worse, one scenario an AI was going to allow someone to die in order to self-preserve. Below is a short video that breaks down what happened. SPOOKY.

Credit: The Infographics Show

Another interesting book I read recently was “The Shortest History of AI”, by Toby Walsh. This was also a quick read that I would recommend for anyone wants to understand the concepts behind AI. He breaks it down in layman’s terms (as much as possible). He does seem to be a bit more relaxed by the AI revolution and doesn’t seem to be an alarmist.

What do you think? Are we going too fast? Is AI going to take over? Have you read either of these books?

Credit: BBC World Service

Take care,
Stōbe