Thanks so much for this blog post. As you know, I’ve been attempting to understand Cooperative AI a bit better over the past weeks. More concretely, I found it helpful as a conceptual exploration of what Cooperative AI is. Including how it relates to other adjacent (sub)fields—including the diagram! I also appreciated you flagging the potential dual-use aspect of cooperative intelligence—especially given the fact that you’re working in this field and therefore be prone to wishful thinking.
That said, I would’ve appreciated if you covered a bit more about: - Why Cooperative AI is important. I personally think Cooperative AI (at least as I currently understand it) is undervalued on the margin and that we need more focus on potential multi-agent scenarios and complex human interactions. - What people in the field of Cooperative AI are actually doing—including how they navigate the dual-use considerations.
Thanks so much for this blog post. As you know, I’ve been attempting to understand Cooperative AI a bit better over the past weeks.
More concretely, I found it helpful as a conceptual exploration of what Cooperative AI is. Including how it relates to other adjacent (sub)fields—including the diagram! I also appreciated you flagging the potential dual-use aspect of cooperative intelligence—especially given the fact that you’re working in this field and therefore be prone to wishful thinking.
That said, I would’ve appreciated if you covered a bit more about:
- Why Cooperative AI is important. I personally think Cooperative AI (at least as I currently understand it) is undervalued on the margin and that we need more focus on potential multi-agent scenarios and complex human interactions.
- What people in the field of Cooperative AI are actually doing—including how they navigate the dual-use considerations.