Readers may also be interested in EA Opportunities Board

Cool initiative! Heads up that I think one of the links in the TLDR is to a private forum draft (or at least it doesn't work for me)

See this post 

I'd be very curious if there are historical case studies of how much private corporations stuck to voluntary commitments they made, and how long it took for more binding regulation to replace voluntary commitments

Pulling out highlights from PDF of the voluntary commitments that AI safety orgs agreed to:

The following is a list of commitments that companies are making to promote the safe, secure, and transparent development and use of AI technology. These voluntary commitments are consistent with existing laws and regulations, and designed to advance a generative AI legal and policy regime. Companies intend these voluntary commitments to remain in effect until regulations covering substantially the same issues come into force. Individual companies may make additional commitments beyond those included here.

Scope: Where commitments mention particular models, they apply only to generative models that are overall more powerful than the current industry frontier (e.g. models that are overall more powerful than any currently released models, including GPT-4, Claude 2, PaLM 2, Titan and, in the case of image generation, DALL-E 2).

1) Commit to internal and external red-teaming of models or systems in areas including misuse, societal risks, and national security concerns, such as bio, cyber, and other safety areas.

2) Work toward information sharing among companies and governments regarding trust and safety risks, dangerous or emergent capabilities, and attempts to circumvent safeguards

3) Invest in cybersecurity and insider threat safeguards to protect proprietary and unreleased model weights

4) Incent third-party discovery and reporting of issues and vulnerabilities

5) Develop and deploy mechanisms that enable users to understand if audio or visual content is AI-generated, including robust provenance, watermarking, or both, for AI-generated audio or visual content

6) Publicly report model or system capabilities, limitations, and domains of appropriate and inappropriate use, including discussion of societal risks, such as effects on fairness and bias

7) Prioritize research on societal risks posed by AI systems, including on avoiding harmful bias and discrimination, and protecting privacy

8) Develop and deploy frontier AI systems to help address society’s greatest challenges

Cool initiative! I'm worried about a failure mode where these just stay in the EA blogosphere and don't reach the target audiences who we'd most like to engage with these ideas (either because they're written with language and style that isn't well-received elsewhere, or no active effort is made to share them to people who may be receptive).

Do you share this concern, and do you have a sense of mitigate it if you share the concern? 

Maybe check out VSFS internships

College students interested in this may also want to check out remote government internships via VSFS! Here are the open positions

This might be a too loose of a criteria for 'power-seeking', or at least the version of power-seeking that has the negative connotations this post alludes to. By this criteria, a movement like Students for Sensible Drug Policy would be power seeking.

1. they try to seed student groups and provide support for them 
2. They have multiple buttons to donate on their webpage
3. They have things like a US policy council and explicitly mention policy change in their title.

Maybe it's just being successful at these things that makes the difference between generic power-seeking  and power-seeking that is perceived as alarming? 

But if I had to guess (of the top of my head) the negative associations with longtermism's alleged power seeking come more from 1) longtermism being an ideology that makes pretty sweeping, unintutive moral claims and 2) longtermism reaching for power among people society labels as 'elites' (e.g., ivory tower academics, politicians, and tech industry people). 

Agree. Something that clarified my thinking on this (still feel pretty confused!) is Katja Grace's counterarguments to basic AI x-risk case. In particular the section on "Different calls to ‘goal-directedness’ don’t necessarily mean the same concept" and discussions about "pseduo-agents" clarified how there are other ways for agents to take actions than purely optimizing a utility functions (which humans don't do).

I really agree with the thesis that EA orgs should make it clear what they are and aren't doing.

But I think specifically for EA orgs that try to be public facing (e.g., an org focused on high net worth donors like Longview) there's a way that very publicly clarifying a transparent scope can cut against their branding/ how they sell themselves. These orgs need to sell themselves at some level to be as effective as possible (it's what their competition is doing), and selling yourself and being radically transparent do seem like they trade off at times (e.g., how much you prioritise certain sections of your website). 

Thankfully, we have a lovely EA forum that can probably solve this issue :)