Warnings and Motives
There’s been a lot of talk about AI safety this week following Jacob Coxon’s resignation from Anthropic.
One theme I’ve seen is people arguing about whether the “fear campaigns” (posts like his, and warnings from labs and leaders in the space) are genuine.
While I have my own opinions on this, what I keep thinking is that it doesn’t matter. We’re actively building something we can’t explain and potentially can’t stop, whether that’s the system itself or the race to build it.
We’re arguing about the motives of people warning us instead of asking whether we can actually understand what is being built and what harm it could do.
At the same time, I think it’s a good thing for this discourse to become more widespread. This is the first time it feels like more people are starting to realize the risks. It’s so important right now for people to understand the risks of much more capable AI, and the rate at which those capabilities are being developed compared to the rate of progress in alignment.