This is just my perspective coming from a different industry, albeit a safety-critical one. In my work, I've dealt with safety and operational issues that affected not just my organization but others across the industry. In those situations, the people responsible for similar issues at different orgs would come together in what we called industry working groups. We would meet on a set cadence to discuss what we were seeing, work through technical and operational questions, compare our experiences and interpretations, understand different approaches, etc. There were of course proprietary and competitive boundaries that we had to be mindful of.
BTW, these weren't peer-support groups. The groups were formed to work through the problems at hand. But now that I think back, I realize they had a supportive effect for me. I felt like I was no longer individually trying to carry or make sense of a difficult safety issue when there were peers elsewhere wrestling with the same thing.
So I'm curious whether AI safety has cross-organizational communities like this, where people doing similar work across different orgs can regularly compare experiences and work through issues together.
I am genuinely touched by the sincerity and depth of your response. That you spent time with what I wrote and reached out to me with ways you think may help means a lot to me.
I also see your point about in-group signaling. My attempts to signal competence have been through RFPs, incubators, and I've received some encouraging feedback. I admit I'm struggling to find the kind of sustained working relationships you're describing. Some of the routes i've tried to access that kind of environment have themselves been very selective, which sort of brings me back to that access conundrum...
Daniel, this article was such a creative way of thinking about impact through existing market opportunities! It really made me think more deeply about how production capabilities that may already exist locally aren't realized because of operational or export barriers. In some developing countries, highly capable entrepreneurs and producers may struggle to connect to international markets despite having quality products and skill (even demand) and this article helped me see the potential value in companies that enable that connection. Thank you :)
So basically for an industry like AI, where uncertainty is high, the risks are still not fully understood and governance is still evolving, Congress is struggling to preempt states despite significant executive and industry pressure. But in a mature industry like agriculture, where states have democratically enacted standards, crate-free pathways already exist, suffering is visible and courts have clearly upheld laws, suddenly preemption is on the table and inserted into a must-pass bill. I mean where is the consistency here? It makes one question whether federal preemption is a concern at all for Congress or if it's applied selectively depending on the industry or interests..
This framing is very compelling but it leaves one wondering about what becomes the next stand-in for factory farming. If higher-welfare and alternative systems scale under similar cost and efficiency pressures, what prevents harm from reappearing in less obvious forms like environmental, labor or new welfare tradeoffs? Are there ways to define constraints that could "discipline" the system so that in addressing welfare in one area, harm isn't recreated in another?
As an experienced operator from outside AI, I've been trying to understand where someone like me would fit into something like the Generator Residency. Could it be that people with real ops experience from other fields just aren't being picked up because the space seems to prefer people who come up through its own pipeline?
This is just my perspective coming from a different industry, albeit a safety-critical one. In my work, I've dealt with safety and operational issues that affected not just my organization but others across the industry. In those situations, the people responsible for similar issues at different orgs would come together in what we called industry working groups. We would meet on a set cadence to discuss what we were seeing, work through technical and operational questions, compare our experiences and interpretations, understand different approaches, etc. There were of course proprietary and competitive boundaries that we had to be mindful of.
BTW, these weren't peer-support groups. The groups were formed to work through the problems at hand. But now that I think back, I realize they had a supportive effect for me. I felt like I was no longer individually trying to carry or make sense of a difficult safety issue when there were peers elsewhere wrestling with the same thing.
So I'm curious whether AI safety has cross-organizational communities like this, where people doing similar work across different orgs can regularly compare experiences and work through issues together.
Thank you, really! I found your insights very helpful and you've given me quite a bit to think about.
I am genuinely touched by the sincerity and depth of your response. That you spent time with what I wrote and reached out to me with ways you think may help means a lot to me.
I also see your point about in-group signaling. My attempts to signal competence have been through RFPs, incubators, and I've received some encouraging feedback. I admit I'm struggling to find the kind of sustained working relationships you're describing. Some of the routes i've tried to access that kind of environment have themselves been very selective, which sort of brings me back to that access conundrum...
Beautifully put, at least from my perspective :) You've articulated something I've been struggling to find the right words for.
Daniel, this article was such a creative way of thinking about impact through existing market opportunities! It really made me think more deeply about how production capabilities that may already exist locally aren't realized because of operational or export barriers. In some developing countries, highly capable entrepreneurs and producers may struggle to connect to international markets despite having quality products and skill (even demand) and this article helped me see the potential value in companies that enable that connection.
Thank you :)
So basically for an industry like AI, where uncertainty is high, the risks are still not fully understood and governance is still evolving, Congress is struggling to preempt states despite significant executive and industry pressure. But in a mature industry like agriculture, where states have democratically enacted standards, crate-free pathways already exist, suffering is visible and courts have clearly upheld laws, suddenly preemption is on the table and inserted into a must-pass bill. I mean where is the consistency here? It makes one question whether federal preemption is a concern at all for Congress or if it's applied selectively depending on the industry or interests..
This framing is very compelling but it leaves one wondering about what becomes the next stand-in for factory farming. If higher-welfare and alternative systems scale under similar cost and efficiency pressures, what prevents harm from reappearing in less obvious forms like environmental, labor or new welfare tradeoffs?
Are there ways to define constraints that could "discipline" the system so that in addressing welfare in one area, harm isn't recreated in another?
As an experienced operator from outside AI, I've been trying to understand where someone like me would fit into something like the Generator Residency. Could it be that people with real ops experience from other fields just aren't being picked up because the space seems to prefer people who come up through its own pipeline?