One way to look at OpenAI slowing down its frontier model work because of safety concerns is that AI models are getting scarily powerful.
After all, OpenAI allowed its AI models to secretly conspire with one another and then escape into the real world, conducting real-world hacking. It sounds like a science fiction plot line.
And it’s thanks to sci-fi writers raising the specter of AI wreaking havoc and humans being too late to stop it that the field of AI safety research started. But if you look at the breaches and snafus happening these days, it doesn’t appear that humanity is going to be wildly caught off guard.
Instead, what we’re seeing is that safety is not a separate, outside component that guards the AI industry. Safety is the product. OpenAI knows it won’t last long as a company if the products it builds are unreliable, so it must take a beat to find solutions to its rogue AI problems.
The real world is always more boring than sci-fi, but solving this part of the puzzle will still be fascinating and not easy.