2. Seeing people at meetups with powerful, personal grievances against AI get talked over and then pushed away by folks laser-focused on x-risk.
3. There is a Ban Superintelligence bill. Full text isn't out yet, but the summary is spot-on in addressing the problem and Sanders has had direct conversations with Yudkosky, Kokotajlo, etc. Have you checked whether people whose primary grievance against AI isn't x-risk support this bill? I have (on an anecdotal level) and the answer is yes.
4. There is an incredibly obvious way to balance authenticity and meeting people where they are at, once one recognizes these are the interests in balance. In full, it sounds like: "There are many reasons to be worried about AI, including A, B, and C. Notice that these keep getting worse, and more numerous. The logical conclusion, and what motivates me personally, is existential risk. What ties everything together is the fact that the AI companies have too much power and are abusing that power by overriding the public interest. This is why we need a proactive solution, stopping the frontier, so that we're not endlessly playing whac-a-mole with new problems, while also taking increasing risk of permanent loss of control."
Yes, there is a failure more of too-deeply burying one's true point (example: https://secureainow.org/mission/), but pointing this out doesn't require swinging to an opposite extreme. Specifically, if 4 is essentially what you are saying already, that's not at all the impression I get from reading this post, which seems much closer to the characterization from point 2.
I recently attended a first aid training course (in part to be an effective steward for the recent PauseAI protest in London), and it was hammered into us that we need to triage problems against our instinct, in order of urgency:
- Unconscious casualties, because you don't know what's wrong, could be a life threatening bleed
- Bleeds
- Burns
- Broken bones
And if that person is in a building on fire or threatening to collapse, you carry or drag them out. It doesn't matter that they might be injured or bleeding and the dragging might harm them because the risk of death of leaving them there is just too high.
As much as I want to believe this, it really doesn't match up with a lot of what I have seen in reality:
1. https://www.socialchangelab.org/research/which-ai-harms-and-risks.html like it or disagree, to not cite this given the subject of this post is wild.
2. Seeing people at meetups with powerful, personal grievances against AI get talked over and then pushed away by folks laser-focused on x-risk.
3. There is a Ban Superintelligence bill. Full text isn't out yet, but the summary is spot-on in addressing the problem and Sanders has had direct conversations with Yudkosky, Kokotajlo, etc. Have you checked whether people whose primary grievance against AI isn't x-risk support this bill? I have (on an anecdotal level) and the answer is yes.
4. There is an incredibly obvious way to balance authenticity and meeting people where they are at, once one recognizes these are the interests in balance. In full, it sounds like: "There are many reasons to be worried about AI, including A, B, and C. Notice that these keep getting worse, and more numerous. The logical conclusion, and what motivates me personally, is existential risk. What ties everything together is the fact that the AI companies have too much power and are abusing that power by overriding the public interest. This is why we need a proactive solution, stopping the frontier, so that we're not endlessly playing whac-a-mole with new problems, while also taking increasing risk of permanent loss of control."
Yes, there is a failure more of too-deeply burying one's true point (example: https://secureainow.org/mission/), but pointing this out doesn't require swinging to an opposite extreme. Specifically, if 4 is essentially what you are saying already, that's not at all the impression I get from reading this post, which seems much closer to the characterization from point 2.
Very well put.
I recently attended a first aid training course (in part to be an effective steward for the recent PauseAI protest in London), and it was hammered into us that we need to triage problems against our instinct, in order of urgency:
- Unconscious casualties, because you don't know what's wrong, could be a life threatening bleed
- Bleeds
- Burns
- Broken bones
And if that person is in a building on fire or threatening to collapse, you carry or drag them out. It doesn't matter that they might be injured or bleeding and the dragging might harm them because the risk of death of leaving them there is just too high.
There is power in directness. It is dissipated with maneuvering. Be accurate. Be concise. Be direct.