Comments
65
This is a crosspost from the new Animal Welfare Alignment Newsletter by Anima International. You can subscribe on Substack if you are interested in following these efforts. Audio reading also available on Substack.
The goals of this post are to:
1. Raise a question I see as crucially important to the goal of aligning AI to animal welfare...
In their current form these question are rather broad which makes them a bit hard to effectively/insightfully answer. That being said, my thinking is:
Depending on how broad your definition of “defense industry” is, most likely there are some positions in the field that make sense to pursue if someone already has relevant connections and/or experience, especially some things related to AI safety or biosafety/biosecurity (while my cursory impression is that things related to nuclear weapons or missile defense are less promising due to effort saturation (less neglectedness), inflexibility of long-standing systems, the fact that asymmetric improvements in such weapons/defenses could be destabilizing, etc.). However, aside from those areas nothing else immediately comes to mind, and overall I suspect that the defense industry is probably not the best career path for most people—even for someone interested in international security there are other options like focusing on security policy. (If your parents are arms magnates, I’d say it might be a good idea, but otherwise I’m fairly skeptical)
I can’t really provide a good answer to this, in part because I’m not even sure what issues to address/focus on, but also in part because I haven’t seen it as a impactful question to answer: we can’t just wish away a (bad?) industry by snapping our fingers, much like we can’t just wish away all nuclear weapons in the world. Should America have a defense industry? Absolutely. Does America spend too much on its defense industry? Most likely in some areas yes, but overall I don’t know. One plain justification is “without an American defense industry, other countries (e.g., China, Russia) will gain relative military advantage and bully/invade surrounding countries, prop up brutal dictators, etc. Additionally, subnational groups will still acquire weapons and commit similar acts of violence against populations…”
I should have narrowed down the question by excluding the AI and nuclear part which I fully agree that are non-neglected topics. However, the rest of the industry: Aerospace (transport, fighters, satellites...), armament, naval, electronic war suppliers, etc... are composed by powerful companies that are not usually "EA targets" in the literature, I mean, not identified as relevant career paths. This seems strange to me since this engineering sector (and even the military itself) seem to have great impact. Hence, it should be a good candidate to somehow be an "effective" sector to work in. Actually, if this sector turns out to be a relevant career path this could be a way for engineers to have a great impact. This sounds really inspiring, but on the other hand, not a lot of ideas come to my mind...
I agree on the current existence of defense industry basically because of same rationale you mentioned on point 2. Hence, given the fact that almost for sure we need Defense (both industry and military), should we encourage EA/RAT people to aim for strategic positions in this sectors?
I think that even if there are no clear and concrete actions to implement in this sector (by the moment), I would tentivevely say that I prefer to have EA/RAT community performing at least some potential strategic roles in the insdustry. But of course, I don't really know which roles or how effective this could be.
Sorry for the non-concrete apporach to the topic but I just feel like this sector could be somehow a neglected path that could be relevant and it also seems odd to me that there is no concrete literature in EA for this (maybe I have not found it or maybe because there are no good reasons for considering this sector as an effective career path). This is why this question has this brainstorming/intuition apporach. So if someone has any thought or interesting article about this, please let me know.
I would be interested to see an 80K Hours profile on this field (among many others) if there isn't one already, but regarding the points you mentioned:
On some level, I agree that defense engineering and the broader defense industry can be impactful, but even if something is impactful overall it may not be a good candidate for cause prioritization. For example, it may not be neglected in the important areas (e.g., many thoughtful and/or skilled people are already working in the field), so having an extra person (or a more ethical and/or skilled person than whoever they are replacing) may not be as valuable as one might expect. Additionally, it may not be a very tractable field for someone who wants to have an impact by being more ethical, since institutional momentum/precedent and corporate incentive may be very strong, and such individuals will often just be a small part in a large bureaucracy/organization. This latter point is especially significant given that improvements in the industry are not so unambiguously good--in fact, in some instances they might be bad if, for example, a new technology leads to destabilizing dynamics. (80K actually does briefly touch on this idea in point 6 of this article) Additionally, my understanding is that there are already some non-EA efforts to promote better norms regarding weapons research (such as the anti-nuke campaigns like ICAN and the anti lethal autonomous weapons (LAWs) campaign, although I know both groups/movements have met with mixed reactions from EAs)
Ultimately, I would still like to hear others' opinions or see any research people may have found, but on the surface I don't see it as clearly neglected by the EA community relative to its potential value. Furthermore, I feel that even if someone does see this field as very valuable, they should probably see defense/security policy (e.g., in executive branches of government, as policy researchers in think tanks) as even more promising, since that work seems less saturated and less constrained.