plex

Karma: 364

Anti-squatted AI x-risk domains index

plex12 Aug 2022 12:00 UTC

53 points

9 comments1 min readEA link

ea.domains—Domains Free to a Good Home

plex12 Jan 2023 13:32 UTC

48 points

8 comments4 min readEA link

plex 30 Jul 2023 13:35 UTC
41 points
4 ∶ 0
on: Shutting down AI Safety Support
AI Safety Support has been for a long time a remarkably active in-the-trenches group patching the many otherwise gaping holes in the ecosystem (someone who’s available to talk and help people get a basic understanding of the lie of the land from a friendly face, resources to keep people informed in ways which were otherwise neglected, support around fiscal sponsorship and coaching), especially for people trying to join the effort who don’t have a close connection to the inner circles where it’s less obvious that these are needed.
I’m sad to see the supporters not having been adequately supported to keep up this part of the mission, but excited by JJ’s new project: Ashgro.
I’m also excited by AI Safety Quest stepping up as a distributed, scalable, grassroots version of several of the main duties of AI Safety Support, which are ever more keenly needed with the flood of people who want to help as awareness spreads.

plex 31 Aug 2022 12:45 UTC
38 points
0 ∶ 0
in reply to: Trev Prew’s comment on: How might we align transformative AI if it’s developed very soon?
Hi Trev!
Very briefly on your points:
1. We don’t think AI needs to break thermodynamics to be dangerous.
2. We don’t think all human-specified goals are safe, and we don’t know how to give a safe one to an extremely powerful AI.
3. We are not worried about self-awareness or consciousness in particular.
4. Turning off highly capable systems is likely to be extremely challenging, unless the stop-button research problem is solved.
Consider familiarizing yourself with some of the basic arguments, for example using this playlist, “The Road to Superintelligence” and “Our Immortality or Extinction” posts on WaitBuyWhy for a fun, accessible introduction, and Vox’s “The case for taking AI seriously as a threat to humanity” as a high-quality mainstream explainer piece.
The free online Cambridge course on AGI Safety Fundamentals provides a strong grounding in much of the field and a cohort + mentor to learn with.^[1]
1. ^
  Links borrowed from Stampy, your one-stop-shop for answering questions about AI Safety.

plex 28 Dec 2022 20:17 UTC
28 points
5 ∶ 0
in reply to: Jeff Kaufman’s comment on: CEA Disambiguation
It’s pretty weird that the charity page has been edited since this comment to hide the board members. I’d like to see reflexive transparency, not reflexive hiding of information.
https://ev.org/charity/
https://web.archive.org/web/20221221123432/https://ev.org/charity/

plex 29 Dec 2022 16:30 UTC
17 points
5 ∶ 0
in reply to: pseudonym’s comment on: CEA Disambiguation
It does seem plausible that it was a miscommunication actually, the original was:
Effective Ventures Foundation (formerly named “Centre for Effective Altruism”) is a charitable company limited by guarantee in England and Wales (company number 07962181), and registered as a charity with the Charity Commission of England and Wales (charity number 1149828).
Effective Ventures Foundation is governed by a board of five trustees (Will MacAskill, Nick Beckstead, Tasha McCauley, Owen Cotton-Barratt, and Claire Zabel) (the “Board”). The Board is responsible for overall management and oversight of the charity, and where appropriate it delegates some of its functions to sub-committees and directors within the charity.
The Centre for Effective Altruism USA Inc. is a tax exempt public charity under section 501(c)(3) of the Internal Revenue Code. ~~Effective Ventures Foundation is the sole member of the Centre for Effective Altruism USA Inc.~~ This group of entities will be collectively referred to as “Effective Ventures” (“EV”). The EV group may expand in future to include more non-profit and philanthropic organisations.
(crossed out is the intentionally removed bit) And the new is
Effective Ventures Foundation (formerly named “Centre for Effective Altruism”) is a charitable company limited by guarantee in England and Wales (company number 07962181), and registered as a charity with the Charity Commission of England and Wales (charity number 1149828).
The Centre for Effective Altruism USA Inc. is a tax exempt public charity under section 501(c)(3) of the Internal Revenue Code. This group of entities will be collectively referred to as “Effective Ventures” (“EV”). The EV group may expand in future to include more non-profit and philanthropic organisations.
I’d guess someone handed someone who was rushing and not paying attention the task of removing the sentence about Effective Ventures in the middle of the last paragraph, and it got misunderstood as removing both that and the middle paragraph. Delegation is pretty failure prone.
I’m also quite keen to assume good faith, and would prefer that central EA nodes didn’t feel like they have to focus undue effort and brain-cycles on watching their backs, and were free to optimize more important things.

plex 11 Dec 2022 12:20 UTC
15 points
4 ∶ 0
on: Personal Finance for EAs
I have been thinking along similar lines. My current angle is to try and connect EA with the FIRE (Financial Independence Retire Early) community, as I think having as many EAs as possible free to work on whatever they find most valuable would be extremely beneficial.
The FIRE movement seems philosophically somewhat EA-adjacent, in that they’re explicitly optimizing finances. Some of them have, once they’re personally set for life, moved to philanthropy. They also have lots of free time on their hands, and some might be happy to consult with EAs who want to optimize their finances.
The name I’ve been thinking of this under is “Efficient Altruism”, and I think it gets back to EA’s roots.
Happy to have a call if you’re interested in bouncing ideas around.
What links here?

plex 9 Sep 2022 2:38 UTC
13 points
3 ∶ 0
on: AI alignment with humans… but with which humans?
The closest thing that comes to mind is Critch’s work on multi-user alignment, e.g. What Multipolar Failure Looks Like, and Robust Agent-Agnostic Processes (RAAPs).

plex 5 Sep 2022 14:30 UTC
10 points
0 ∶ 0
in reply to: RoboTeddy’s comment on: Improving Karma: My median estimate is it’s worth $8mn
There is already a team working on the transitive pagerank-like karma over at EigenTrust.Net, with a functional prototype using Discord reacts. We’d love to integrate with EA Forum / LW / Alignment Forum. Feel free to drop by our dev channel!

plex 30 Jan 2023 13:05 UTC
9 points
2 ∶ 0
on: Time-stamping: An urgent, neglected AI safety measure
Seems worthwhile and good for an extra reason: It allows us to train AIs on only data from before there was significant AI generated content, which might mitigate some safety concerns around AIs influencing future AIs training.

plex 20 Oct 2022 19:05 UTC
8 points
1 ∶ 0
on: Hackathon on Mon, 12/5 to follow EAGxBerkeley
Alignment Ecosystem Development maintains a list of projects which volunteer-scale efforts could shift the needle on, and hosts monthly calls where people can pitch or join projects. Maybe someone could drop by one of those calls and see if anything fits the bill for this?

plex 15 Nov 2022 18:47 UTC
7 points
1 ∶ 0
on: Hackathon on Mon, 12/5 to follow EAGxBerkeley
Some approaches to solving alignment go through teaching ML systems about alignment and getting research assistance from them.^[1] Training ML systems needs data, but we might not have enough alignment research to sufficiently fine tune our models, and we might miss out on many concepts which have not been written up. Furthermore, training on the final outputs (AF posts, papers, etc) might be less good at capturing the thought processes which go into hashing out an idea or poking holes in proposals which would be the most useful for a research assistant to be skilled at.
It might be significantly beneficial to capture many of the conversations between researchers, and use them to expand our dataset of alignment content to train models on. Additionally, some researchers may be fine with having their some of their conversations available to the public, in case people want to do a deep dive into their models and research approaches.
The two parts of the system which I’m currently imagining addressing this are:
1. An email where audio files can be sent, automatically run through Whsiper, and added to the alignment dataset github.
2. Clear instructions for setting up a tool which captures audio from calls automatically (either a general tool or platform-specific advice), and makes it as easy as possible to send the right calls to the dataset platform.^[2]
1. ^
  Ought’s Elicit is the prime example.
2. ^
  I hear OBS might be a good tool for this.

Storyteller’s convention, 2223 A.D.

plex7 Apr 2023 11:54 UTC

6 points

0 comments1 min readEA link

plex 9 Feb 2023 0:21 UTC
6 points
1 ∶ 0
in reply to: Ben_West’s comment on: EigenKarma: trust at scale
I’m curious about what the thing you call EigenKarma is, is it the way people with more karma have more weighty votes? Or is it something with a global eigenvector?

plex 3 Nov 2022 13:37 UTC
6 points
0 ∶ 0
in reply to: Holly Morgan’s comment on: All AGI Safety questions welcome (especially basic ones) [~monthly thread]
Yes, this seems like a good idea, I’ll do that next time unless someone else is up for handling incoming questions.

plex 24 Sep 2022 13:51 UTC
6 points
1 ∶ 0
in reply to: Zach Stein-Perlman’s comment on: Announcing the Future Fund’s AI Worldview Prize
Agree, and add that code models won’t be data constrained as they can generate their own training data. It’s simple to write tests automatically, and you can run the code to see whether it passes the tests before adding it to your training dataset. As an unfortunate side effect, part of this process involves constantly and automatically running code output by a large model, and feeding it data which it generated so it can update its weights, both of which are not good safety-wise if the model is misaligned and power seeking.
I don’t know if this has been incorporated into a wider timelines analysis yet as it is quite recent, but this was a notable update for me given the latest scaling laws which indicate that data is the constraining factor, not parameter count. Much shorter timelines than 2043 seem like a live and strategically relevant possibility.

plex 8 Feb 2023 19:05 UTC
5 points
1 ∶ 0
on: EigenKarma: trust at scale
Topically, this might be a useful part of a strategy to help the EA forum to stay focused on the most valuable things, if people had the option to sync their vote own history with the EigenKarma Network and use EKN lookup scores to influence the display and prioritization of posts on the front page. We’d be keen to collaborate with the EAF team to make this happen, if the community is excited.

plex 12 Nov 2022 20:29 UTC
5 points
0 ∶ 0
on: aisafety.community—A living document of AI safety communities
Updates:
- Simplified columns (link is now in the name of each entry, rather than a separate column, and made the names bold)
- Reordered to put some of the entries added by volunteers nearer the top (unknown people are adding stuff! excellent!)
- Moved one now-inactive project to the bottom
- Added a note about EA Gather Town’s invite link breaking every month, and to go via the Discord instead.
- Updated icon
- Added note about who to talk to for one invite-only Slack
- Successfully applied for $1k credit for the software, which means we have edit histories enabled on this and all docs for about three years for free

plex 21 Oct 2022 9:39 UTC
5 points
0 ∶ 0
in reply to: Chris Leong’s comment on: aisafety.community—A living document of AI safety communities
Done, and done. Edit access should be open now too.

plex 21 Oct 2022 9:37 UTC
5 points
0 ∶ 0
in reply to: Chris Leong’s comment on: aisafety.community—A living document of AI safety communities
Good idea, done.

plex

Anti-squat­ted AI x-risk do­mains index

ea.do­mains—Do­mains Free to a Good Home

Sto­ry­tel­ler’s con­ven­tion, 2223 A.D.

Anti-squatted AI x-risk domains index

ea.domains—Domains Free to a Good Home

Storyteller’s convention, 2223 A.D.