Co-Director @ENAIS, connecting researchers and policy-makers for safe AI
Formerly director of EA Germany, EA Berlin and EAGxBerlin 2022
Happy to connect with people with shared interests. Message me with ideas, proposals, feedback, connections or just random thoughts!
I’m excited you’re doing this! This seems helpful to fill this gap in AIS field building.
(I’m late to the party, just saw this now in a newsletter)
tldr: Do high school still neglect ethics / moral philosophy in their curriculums? Mine did (year 2012). Are there tractable ways to improve the situation, through national/state education policy or reaching out to schools and teachers? Has this been researched / tried before?
The public high school I went to in Rottweil (rural Southern Germany) was overall pretty good, probably top 2-10% globally, except for one thing: Moral philosophy. 90min/week “Christian Religion” was the default for everyone, in which we spent most of the time interpreting stories from the bible, most of which to me felt pretty irrelevant to the present. This was in 2012 in Germany, a country with more atheists than Christians as of 2023, and even in 2012 my best guess is that <20% of my classmates were practicing a religion.
Only in grade 10, we got the option to switch to secular Ethics classes instead, which only <10% of the students did (Religion was considered less work).
Ethics class quickly became one of my favorite classes. For the first time in my life I had a regular group of people equally interested in discussing Vegetarianism and other such questions (almost everyone in my school ate meat, and vegetarians were sometimes made fun of). Still, the curriculum wasn’t great, we spent too much time with ancient Greek philosophers and very little time discussing moral philosophy topics relevant to the present.
How have your experiences been in high school? I’m especially curious about more recent experiences.
Are there tractable ways to improve the situation? Has anyone researched this?
1) Could we get ethics classes in the mandatory/default curriculum in more schools? Which countries or states seem best for that? In Germany, education is state-regulated—which German state might be most open to this? Hamburg? Berlin?
2) Is there a shortage in ethics teachers (compared to religion teachers)? Can we get teachers more interested in teaching ethics?
3) Are any teachers here teaching ethics? Would you like to connect more with other (EA/ethics) teachers? We could open a whatsapp group, if there’s not already one.
Forwarding private comment from a friend: Interoperability was part of Digital Markets Act, so EVP Ribera will be main enforcer, and was asked about her stance in her EU parliament confirmation hearing yesterday. You could watch that / write her team abt the underrated cybersecurity benefits of interoperability esp. given it would upgrade WhatsApp’s encryption.
Curious to get your thoughts on this too!
Thanks for looking up their funding situations, appreciate it!
I meant neglected as in “they don’t seem to prioritize it for whichever reason”, not necessarily funding- or capacity-constraints.
I see how they might not want to implement some of these features, though even in the case of “show message history” to new members, there could be more elegant solutions like giving members the option to opt-in to sharing their messages with new people in the group.
Other features like “enabling community chats / supergroup” or “better chat archiving & sorting chats in folders” seem not in conflict with privacy, at least not obviously. Generally they do seem to copy many features from other messengers (they recently launched stories, similar to Whatsapp status), they just seem a lot slower than Whatsapp and Telegram to adopt these things and far behind.
Forwarding private comment from a friend: Interoperability was part of Digital Markets Act, so EVP Ribera will be main enforcer, and was asked about her stance in her EU parliament confirmation hearing yesterday. You could watch that / write her team abt the underrated cybersecurity benefits of interoperability esp. given it would upgrade WhatsApp’s encryption
Signal (similar to Whatsapp) is the only truly privacy-friendly popular messenger I know. Whatsapp and Telegram also offer end-to-end encryption (Telegram only in “secret chats”) but they still collect metadata like your contacts, and many people I meet strongly prefer Signal for various reasons: Some people work in cybersecurity and have strong privacy preferences, others dislike Telegram (bad rep, popular among conspiratists, spam) and Meta (Whatsapp owner). For some vulnerable people such activists in authoritarian regimes or whistleblowers in powerful organizations, secure messaging seems essential, and Signal seems to be the best tool we have.
While Signal is improving, I still often find it annoying to use compared to Telegram. Here just some examples:
1) it’s easily overwhelming: No sorting chats in folders, archiving group chats doesn’t really work (they keep popping back to ‘unarchived’ whenever someone writes a new message), lots of notifications I don’t care about like “user xyz changed their security number” and no way to turn them off
2) no option to make chat history visible to new group members, which is really annoying for some use cases
3) no poll feature, no live location sharing
4) no “community”/supergroup feature, people need to find and manually join all different groups in a community
5) no threads (in Telegram that’s possible in announcement channels)
I wouldn’t be surprised if we’re collectively losing thousands or even millions of productive hours and valuable attention on Signal (I would strongly recommend Slack over Signal, but for some use cases or some users Slack doesn’t work). This seems high in scope, neglected and tractable to me.
Curious to get your thoughts on:
a) Disagree with my argument? Am I missing anything?
b) What’s the bottleneck of Signal? Is it a matter of prioritization, funding/talent (edit: probably not, see harfe’s comment), or something else? Does anyone have insights?
c) Can we help? How?
9.7/10 average rating* is extremely good. Well done!
Also cool to see the Nigerian and African EA community growing. Keep us updated!
(*I don’t think average rating is the most important metric, the expected impact from the conference seems more important, but it’s a useful metric nevertheless)
Are you still active? The website seems offline.
This seems pretty bad news from an AI safety perspective :/
Any chance to override his veto, or get a similar bill passed soon?
More context:
TechCrunch article (not paywalled like the WSJ article)
Response from Senator Wiener on Twitter + discussion
Thanks for sharing!
@Jeff Kaufman Would you like to respond to this? Do you feel like this addresses your concerns sufficiently? Any updates in either direction?
I just skimmed it due to time constraints, but from what I read and from the reactions this looks like a very thoughtful response, and at least a short reply seems appropriate.
If anyone here also would like more context on this, I found @Garrison’s reporting from 16 August quite insightful:
The Tech Industry is the Biggest Blocker to Meaningful AI Safety Regulations
Thank you for the comprehensive research! California state policy as a lever for AI regulation hasn’t been much on my radar yet, and as a European concerned about AI risk, I found this very insightful. Curious if you (or anyone here) have thoughts on the following:
1) Is there anything we can and should do right now? Any thoughts on Holly’s “tell your reps to vote yes on SB 1047” post from last week? Anything else we can do?
2) How do you see the potential for California state regulation in the next few years. Should we invest more resources in this, relative to US AI policy?
I understand your concern, thanks for flagging this!
To add a perspective: As a former EA movement builder who thought about this a bunch, the reputation risks of accepting donations from a platform by an organization that also organizes events that some people found too “edgy” seem very low to me. I’d encourage EA community organizers to apply, if the money would help them do more good, and if they’re concerned about risks, ask CEA or other senior EA community organizers for advice.
Generally, I feel many EAs (me included) lean towards being too risk-averse and hesitant to take action than too risk-seeking (see omission bias). I’m also a bit worried about increasing pressure on community organizers not to take risks and worry about reputation more than they need to. This is just a general impression, I might be wrong, and I still think there are many organizers who might be not sufficiently aware of the risks, so thanks for pointing this out!
Cool that you’re doing this! I could share two failures, one in my career plans and one in job applications. I could do that in 1-4min, depending on how many other people want to share and how much time we have. Looking forward! :)
Nice map! Do you want to upload this on some website, so people can share and find it easier? (similar to aisafety.world) Could be worth investing a tiny bit of money to buy such a domain?
One additional reason:
If you get your (initial) training from a neutral-ish impact organisation, like some management consulting or tech companies, and then move on to a high impact job, you can add value right away with less ‘training costs’ for the high impact org = more impact.
All else equal, an EA org with staff with 1-3 years (non-EA) job experience can achieve more impact quicker than one with partly inexperienced staff.
That said, some things such as good epistemics or high moral integrity may be easier to learn at EA orgs (though they can definitely also be learned elsewhere).
I’ve supported >100 people in their career plans, and this seems pretty solid but underappreciated advice. Thanks for writing it up!
I think I made that mistake too. I went for EA jobs early in my career (running EA Berlin and then EA Germany 2019-22, funded by CEA grants). There were some good reasons: This work seemed particularly neglected in 2019-21, it seemed a good fit and three senior people I had in-depth career 1-1s with all recommended it. I learned a lot, met many inspiring people and I think I did had some significant positive impact as well, on the community overall (it grew and professionalized) and on some individual member’s careers.
However, I made a lot of mistakes too, had slow feedback loops (no manager, little mentorship), and I’m pretty sure I would have learned many (soft) skills faster and built overall better career capital (both in- and outside of EA), if I had first spent 1-2 years in management consulting or in a fast growing (non-EA) tech company with good management, and then went on to direct EA work.
I agree that it would be good to have citations. In case neither Ozzie nor anyone else here finds it a good use of their time to do it—I’ve been following OpenAIs and Sam Altman’s messaging specifically for a while and Ozzie’s summary of their (conflicting) messaging seems roughly accurate to me. It’s easy to notice the inconsistencies in Sam Altman’s messaging, especially when it comes to safety.
Another commenter (whose name I forgot, I think he was from CLTR) put it nicely: It feels like Altman does not have one consistent set of beliefs (like an ethics/safety researcher would) but tends to say different things that are useful for achieving his goals (like many CEOs do), and he seems to do that more than other AI lab executives at Anthropic or Deepmind.
I voted for promising early-stage orgs, as they seem often more funding-constrained and more cost-effective on the margin than established organizations (who have a much easier time fundraising). More speculative, but seemed worth the risk. I like this format, thanks for organizing this!