Thanks for this great first post. I agree with your main point "We are not getting through". There's a weird disconnect with the public knowing more about AI safety and getting more concerned, while people in power and AI industry leaders seem less worried than ever. Claude is really nice to talk to so its fine right alignment solved?

I like your bent towards action too.

I'm not sure the correct answer is that we should "ALL" be out there lobbying, but yes there's no reason why someone shouldn't talk to every MP in their office hours. Unfortunately my brief experience has been that the personality of the average EA (especially in AI safety) isn't inclined towards getting out there and having awkward conversations with people who don't think like them.

I wonder sometimes whether AI safety people could "hire" a team from the Humane league  or something who have proved themselves to be good at this kind of thing.

I just wish we could know for sure.... As much as I try for my love to work in sentience probabilities it ain't easy. And I'm more of a math guy than most!

I feel like even if this is largely true, it doesn't negate the part of the OPs point of which is something like there a mismatch with communicating "the world needs you working on AI" and "there don't seem to be enough jobs for half of good people that want to work on AI"

On your second point "We need more people working on these neglected issues" doesn't necessarily mean that orgs have the "management capacity to absorb more people". 

if that's true then in practical sense do we actually need more people working on these neglected issues? Or do we need more jobs first before we push for more people? Or like the OP suggested could there be more junior hires then effort building people up through the system?

And are orgs like 80,000 hours being honest enough about the job market in their communication?

Me: "Well at least this study shows no association beteween painted houses and kids' blood lead levels. That's encouraging!"

Wife: "Nothing you have said this morning is encouraging NIck. Everything that I've heard tells me that our pots, our containers and half of our hut are slowly poisoning our baby"

Yikes touche...

(Context we live in Northern Uganda)

Thanks @Lead Research for Action (LeRA) for this unsettling but excellently written report. Our house is full of aluminium pots and green plastic food containers. Now to figure out what to do about it!

https://drive.google.com/file/d/1pqRUeejiRCX2bXekeZnL0zGi34zbK23w/view

Wow this is brilliant thank you!

To add to this problem, I think that more uncertain causes are more likely to have overestimated point-estimates even errors non-withstanding. This is for a few reasons, one because we are by nature more optimistic than reality about small probabilities, and that lower quality studies such as cross sectional studies and cohort tend to overestimate effects. When better studies like RCTs or even larger cohorts are done effects often become smaller or disappear.

Unfortunately for many interventions the best data we have is low quality data, and we anchor on those likely overestimated values.

Thanks for the meeting, appreciate it a lot! One observation I have is there seemed to be 2 groups of people in the meeting, that could speak the same language with each other but not so well in between. I still gained a bit from 

1. AI Coders: People who were focused on coding and building applications. They use Codex and Claude code (maybe down Clawdbot...), have multiple windows open and use words like "skills". They utilise AI to the max 25 hours a day.

2: Everyday optimisers People who use mainly the browser AI for a wide range of research / writing / image / organisation tasks. Maybe they a $20 a month subscription, and might use it a little every day or a few times a week. 

I'm very much in group 2, and I still gained a bit from talking to group 1 but I think the use cases are as much of a difference as ability here.

Thanks again for organising if theres more I will come again :)

There is however rising appetite for 1v1s. I just went to an online meeting about Skoll world forum, which is probably the biggest NGO and Funder conference in the world. Both speakers emphasized 1v1s being the most important aspect, and advised only going to other sessions at times when 1v1s weren't books.

So maybe the GHD ecosystem at least is wakign up a bit...

I've said this before and got it completely wrong, but this feels like an LLM wrote a lot of it.