Peter Favaloro

AI safety grantmaking at @Open_Phil. (Any opinions are mine, not necessarily my employer's.)

[Peter Favaloro @PeterFavaloro]

@MaxNadeau_ Well put, Max. An additional reason we went big: we want safety researchers to be ahead of the curve on automating their research, and Geoffrey shares that vision.

I'm excited to see what Resolution can accomplish with these resources!

[Peter Favaloro @PeterFavaloro]

@notanastronomer @eni_iljazi @indevmag 😎 very cool

[Peter Favaloro @PeterFavaloro]

@notanastronomer @eni_iljazi Wait Lauren how do you know Eni?

[Peter Favaloro @PeterFavaloro]

@coeff_giving @JakeMendel99 @MaxNadeau_ We've pushed the deadline back to Dec 1 to give you more time to apply. Apply here, and please forward to anyone who could be a good fit! https://t.co/5NRX7BSjFM

[Peter Favaloro @PeterFavaloro]

Come work on my team at @coeff_giving ! We fund lots of cool technical research on AI safety, and we're growing like crazy. We're spending ~$130m in grants this year, and there's a lot more fantastic work that needs funding. To make those grants happen, we need you!

[Coefficient Giving @coeff_giving]

(1/8) Open Philanthropy’s Technical AI Safety team is recruiting grantmakers to support research aimed at reducing catastrophic risks from advanced AI. We’re hiring at all levels of seniority.

[Peter Favaloro @PeterFavaloro]

@coeff_giving Don't just take my word for it: check out what grantmakers @jakemendel99 and @maxnadeau_ have to say about what these jobs are like: https://t.co/J3IttNieo8

[Jake Mendel @JakeMendel99]

Work with me to turn hundreds of millions of dollars into awesome technical work to reduce AI risk (interp, adv robustness, scalable oversight, DL theory, AI control, etc)! There really is lots to do, lots of money to move, and (maybe) not much time.

[Peter Favaloro @PeterFavaloro]

Super excited to see what applications we get to our RFP! Launching today:

[Max Nadeau @MaxNadeau_]

🧵 Announcing @open_phil's Technical AI Safety RFP!

We're seeking proposals across 21 research areas to help make AI systems more trustworthy, rule-following, and aligned, even as they become more capable.

[Peter Favaloro @PeterFavaloro]

@eigenrobot Obligatory note that Stephenson was talking about cryptocurrencies a decade before Nakamoto published. https://t.co/UnnY2IKNzq

[Peter Favaloro @PeterFavaloro]

@ciphergoth Finance billionaires (hedge fund titans etc) often have go-to employees for this stuff. That’s why e.g. Ray Dalio produces so much non-finance content.

[Peter Favaloro @PeterFavaloro]

@StefanFSchubert @trevposts Thomas Friedman references the wisdom of taxi drivers enough that it has become its own trope: https://t.co/n5LndzVsPi

[Peter Favaloro @PeterFavaloro]

@exgenesis Didn’t know we were there already. Deets?

[Peter Favaloro @PeterFavaloro]

@otis_reid Good telling of that story: https://t.co/9cY2S09QU3

[Peter Favaloro @PeterFavaloro]

@TaliaRinger Huh, interesting. If we somehow developed artificial neurons to replace the damaged brain tissue in a stroke patient (and managed to replicate the person's previous neurons exactly), do you think they'd have less subjective experience?

[Peter Favaloro @PeterFavaloro]

@liron @Noahpinion Some more realistic analogies:
- You can't put a human in your fighter-jet drone, if you want it to be able to accelerate at >10 Gs.
- In automated warehouses, when there's a human present all the robot forklifts have to run at like 1/5 their normal speeds.

[Peter Favaloro @PeterFavaloro]

@liron @Noahpinion I agree it doesn't hold in the likeliest near-term scenarios. But it's useful to show the limits of the comparative-advantage logic.

[Peter Favaloro @PeterFavaloro]

@diviacaroline @liron @Noahpinion Yeah that's where Katja goes in her essay on ants, right?

[Peter Favaloro @PeterFavaloro]

@liron @Noahpinion Yeah but they can't, is the point. They're good at functioning in a certain context, and that context is not the modern human economy.

A chimp employee would cause more headaches than would be worth his output, no matter how cheap his wages.

[Peter Favaloro @PeterFavaloro]

@liron @Noahpinion Classic response to this: why don't we hire chimpanzees to do some jobs? Comparative advantage would say that there should be *some* job where they can achieve the outcome cheaper than a human, even if a human is better at it.

[Peter Favaloro @PeterFavaloro]

@davidad @liron @beffjezos Awesome chart. Is there a writeup that goes with it, or any more detail on the analysis?

[Peter Favaloro @PeterFavaloro]

@moorehn Huh, I feel like Californians value (and have) autonomy a lot more than New Yorkers. When I lived in NYC I felt like the city had a sense of camaraderie and everyone felt justified to be in each other’s business. In CA it feels more atomized; living as individuals.