[girfan]

Curious to know what people's thoughts are on this "Stop The AI Race" movement?

I have not been able to find a clear objective---beyond slowing the progress on various AI fronts---and how that achieves an end-goal towards safer AI.

They mention things like "limits through compute thresholds (FLOP caps)" which seems extremely hand-wavy to me. What does that achieve and how do you enforce only "negative" use-cases of AI being a target of such policies? The same compute is used for non-AI workloads, along with plenty of useful advancements as a result of AI. The lines between general scientific work in numerous tangential fields (e.g., computer architecture) and how it impacts AI is blurring fast, but that does not mean (at least to me) that those fields should also be subject to an outright blockade.

I completely understand and support the need for researchers and practitioners to come together, ensure net-positive outcomes from AI progress and minimize the steep cost (e.g., environmental). I want to understand how this movement in particular (i.e., their demands, vision, plan) brings us closer to that end-goal?