Open Philanthropy

GiveWell[1] and Open Philanthropy[2] would probably be of interest here. (I am a donor to some of GiveWell's recommended organizations but otherwise not affiliated with them) [1] http://www.givewell.org/ [2] http://www.openphilanthropy.org/

It's great that over $1B was donated now for global health! But the highest ever grant was $55M for "potential risks from AI". The 4th and 6th highest grants were similar.[1] More recently[2] the highest ($10.7M) grant was for "Potential Risks from Advanced AI" and third highest grant ($2.3M) was for "longtermism" and there were 4 other similar grants on the first page. [1] https://www.openphilanthropy.org/grants/?sort=high-to-low#categories [2] https://www.openphilanthropy.org/grants/?sort=recent#categories

I have followed EA for ten years and there is absolutely no doubt that its most prominent voices - Will MacAskill, Nick Bostrom, Rob Wiblin, Eliezer Yudkowsky, not to mention Sam Bankman-Fried - ascribe enormous significance to AI. I myself have yet to see a compelling argument as to how AGI is possible, let alone dangerous. If you look at 80,000 hours' list of 'most pressing problems', AI safety is first: https://80000hours.org/problem-profiles/ Open Phil has donated $250m to AI safety according to its website: https://www.openphilanthropy.org/focus/potential-risks-advanced-ai/

It's worth noting that there are several reputable efforts pursuing criminal justice reform independently of the problematic BLM organization or movement. https://www.openphilanthropy.org/focus/us-policy/criminal-justice-refo… has a useful overview of groups you could look into.

Highly recommended for anyone who liked this article: https://www.openphilanthropy.org/blog/impact-incarceration-crime

There are ways to greatly improve chances that AI will be beneficial to humanity rather than otherwise. Check out: UC Berkeley's Center for Human-Compatible AI, led by Prof. Stuart Russell, a co-author of the field's standard textbook. [1] He just gave a TED talk on the issue [2]. Several other noted researchers in AI are working on the issue as well. For a short primer: https://futureoflife.org/background/aimyths/ [1] http://www.openphilanthropy.org/focus/global-catastrophic-risks/potent… [2] https://www.ted.com/talks/stuart_russell_how_ai_might_make_us_better_p…

For those interested in another very different analysis of the evidence for and against superintelligence being a worthwhile concern, here's Holden Karnofsky, executive director of GiveWell and the related Open Philanthropy Project: - Up until 2016, Holden held the opinion that it was not a worthwhile concern. He describes why he changed his mind here: http://www.openphilanthropy.org/blog/three-key-issues-ive-changed-my-m… - And he outlines his current arguments in favor of treating it as a significant concern here: http://www.openphilanthropy.org/blog/potential-risks-advanced-artifici… In that lengthy write-up, he addresses both the shorter-term risks from things like misuse of non-superintelligent AI, along with longer-term risks from superintelligent AI.

Huge fans of the work GiveWell does. For those that share GiveWell's outlook on the world, keep an eye on their Open Philanthropy Project: http://www.openphilanthropy.org/

As someone in the "malaria nets wing", I think you're directionally correct, but overstating things. > it's quite obvious that the opinionated people who control the (huge!) purse-strings are gradually gaining organizational control of the movement This is to some extent true of Open Philanthropy, although the effect looks larger than it is because they're consciously committed to not throwing all of their resources behind whatever they think is the best option. I'm not a fan in principle, but it's not insane. See https://www.openphilanthropy.org/research/worldview-diversification/ for their take. GiveWell remains firmly on the global health side, and I don't see that changing. Here's the first my-screen-worth of organizations they've funded in the last year, with approximate numbers: - 87 million to the Malaria Consortium - 77 million to Hellen Keller International (Vitamin A supplementation) - 42 million to New Incentives (infant vaccination) - 17 million to Sightsavers (parasitic worms) - 8 million to PATH (malaria) - 6 million to Nutrition International (Vitamin A) - 5 million to IRD Global (healthcare infrastructure) - 5 million to Miracle Feet (clubfoot) - 7 million to Deworm the World (parasitic worms) - 2 million to RICE (postnatal care) From https://airtable.com/appGuFtOIb1eodoBu/shr1EzngorAlEzziP/tblG72bMUu36l…

Here are some EA highlights: * Founding and scaling GiveWell, synthesizing and analyzing global poverty research to make recommendations. This has helped guide about $1B+ in really valuable work, the largest portion in anti-malarial bednets. https://www.givewell.org/about/impact * Convincing Dustin Moskovitz and Cari Tuna to allocate their giving along these lines, building Open Philanthropy which has made a lot of grants that look really valuable https://www.openphilanthropy.org/grants (this is in addition to the ones where they granted at GiveWell's recommendation -- don't want to double count). * Michael Kremer won a Nobel prize in econ (shared with Esther Duflo and Abhijit Banerjee who I'd describe as EA-sympathetic but probably don't consider themselves EAs?) for global poverty research. * Built an "AI Safety" subfield of computer science, with the 2016 paper "Concrete problems in AI safety" having 1.8k citations. * Founding and scaling 80,000 Hours, helping people figure out how to use their careers to do more good. This includes looking into potential career opportunities, writing up public advice, and 1:1 advising, and 1k+ people have cited them as part of why they changed what they were working on. https://80000hours.org (Not especially representative; there's been thousands of people working along these lines for years and I'm sure I'm missing a lot of valuable work.)

Dear Rich Pineapple: Congrats on your huge BTC surplus! To make the biggest difference, consider looking into Effective Altruism[1] (an idea, not a charity) if you haven't already. Their motto is “using reason and evidence to do the most good”. Perhaps do away with the charity proposal form, and instead choose one or several of the most effective charities as ranked by GiveWell.org[2] (who do rigorous effectiveness analysis). OpenPhilanthropy[3] is a charity that takes the EA perspective. Unfortunately, seemingly worthy charities are often orders of magnitude less effective than the few most effective causes. Charity evaluators like GiveWell help us find the gems that can leverage our money many times more effectively. For more info about Effective Altruism, listen to Tim Ferriss[4] and Sam Harris[5] interview the co-founder of Effective Altruism, check out this NYT piece[6], or read Doing Good Better[7]. [1] https://www.effectivealtruism.org/ [2] https://www.givewell.org/ [3] https://www.openphilanthropy.org/ [4] http://tim.blog/2015/11/22/will-macaskill/ [5] https://www.samharris.org/podcast/item/being-good-and-doing-good [6] https://www.nytimes.com/2015/08/16/upshot/effective-altruism-where-cha… [7] https://www.amazon.com/Doing-Good-Better-Effective-Altruism/dp/1592409…

The second author seems to be human. https://www.openphilanthropy.org/about/team/alex-lawsen/

Skimming through https://www.openphilanthropy.org/research/what-a-compute-centric-frame… , it seems that Davidson does not recognize that discontinuities, jumps, are needed to go from current technology to AGI.

Have you read their original statements? > We're a non-profit research company. Our full-time staff of 60 researchers and engineers is dedicated to working towards our mission regardless of the opportunities for selfish gain which arise along the way. "regardless of the opportunities for selfish gain which arise along the way", they failed really hard on that point. That Sam Altman had already changed course by 2019 doesn't mean he didn't lie at the start. https://www.openphilanthropy.org/files/Grants/OpenAI/OpenAI_Mission.pd…

> And longtermism has taken over much of the EA movement Show me the money. Sci-Fi/Fantasy longtermst nerd concerns get a disproportionate amount of attention. But from what I can tell, most donations still go to charities dealing with malaria, deficiencies and vaccines: https://www.openphilanthropy.org/research/2021-allocation-to-givewell-… Sure, some people are getting paid thousands of dollars for navel gazing. But more than a billion dollars was donated to directly improve global health.

This comparison is not perfect for various reasons. For example, the average firing rate of neurons is pretty low. Most attempts at comparison are done using FLOPS, e.g. https://www.openphilanthropy.org/research/how-much-computational-power… You may prefer this comparison using Traversed Edges Per Second: https://aiimpacts.org/brain-performance-in-teps/.

There is a long history of connectionist attempts trying to ballpark the brain compute to constrain AI timelines, going back to von Neumann/Turing/Good. The most recent one would be https://www.openphilanthropy.org/brain-computation-report You can see in Figure 1 that your 10^12 steady state is the very low end. If you're interested in seeing where your envelope estimate differs from the others, well, it has the references.

The situation is more complicated. Anyone with sufficient money can effectively buy "content". For example: https://www.openphilanthropy.org/focus/us-policy/farm-animal-welfare/t…

If statistics are bullshit I’m sure you’ll never refer to the below evidence. If you refuse to look at or listen to evidence unless it’s ideologically congenial you’re not a thinker, you’re a hack. https://www.openphilanthropy.org/blog/reasonable-doubt-new-look-whethe… > I estimate, that at typical policy margins in the United States today, decarceration has zero net impact on crime outside of prison. That estimate is uncertain, but at least as much evidence suggests that decarceration reduces crime as increases it. The crux of the matter is that tougher sentences hardly deter crime, and that while imprisoning people temporarily stops them from committing crime outside prison walls, it also tends to increase their criminality after release. As a result, “tough-on-crime” initiatives can reduce crime in the short run but cause offsetting harm in the long run.

Givewell's founders know this, which is why they spun off the Open Philanthropy Project: https://www.openphilanthropy.org/focus/global-catastrophic-risks

See this introduction to the global risk of geomagnetic storms from the Open Philanthropy Project. http://www.openphilanthropy.org/blog/geomagnetic-storms-introduction-r…

> have the EA people actually done something so far? In terms of stuff done: OpenPhil, an EA org, has given north of $30million to groups researching strong AI and its risks[0]. Some effective altruists also choose to donate to AI research groups like MIRI[1]. Why would stopping a botnet be evidence of effectiveness at tackling strong AI? Botnets don't exhibit any kind of intelligence as far as I'm aware. 0. http://www.openphilanthropy.org/giving/grants 1. http://effective-altruism.com/ea/14c/why_im_donating_to_miri_this_year…

Well put. The Open Philanthropy folks take a broadly similar view of the probabilities, both of when strong AI is likely to arrive and of how much positive influence per dollar they can have on how things turn out: http://www.openphilanthropy.org/blog/potential-risks-advanced-artifici…

The Open Philanthropy Project uses the word 'Open' to mean (http://www.openphilanthropy.org/what-open-means-us): "Open to many possibilities ... instead of starting with a predefined set of focus areas, we’re considering a wide variety of causes where our philanthropy could help to improve others’ lives." and "Open about our work ... Very often, key discussions and decisions happen behind closed doors, and it’s difficult for outsiders to learn from and critique philanthropists’ work. We envision a world in which philanthropists increasingly document and share their research, reasoning, results and mistakes to help each other learn more quickly and serve others more effectively." This all seems pretty useful so I don't get what your criticism is.

Thanks for the enthusiasm! I think a lot of your questions can be answered here (including the other pages that these pages link to): http://www.openphilanthropy.org/research/our-process and http://www.givewell.org/how-we-work/process Because our fund managers are all GiveWell and Open Philanthropy staff, we inherit their methodology. If in the future we move beyond GiveWell and Open Philanthropy fund managers, we'll need to have pages on their process too (and how it potentially differs from GW / OP). A more accessible introduction to how to assess which problem area is most important is here: https://80000hours.org/career-guide/most-pressing-problems/

https://www.hhs.gov/about/news/2016/07/28/hhs-forges-unprecedented-par… HHS has a joint initiative with UK-based research foundations. I really wish an easy non-profit foundation existed that I could throw some change at. Here's some more information: http://www.openphilanthropy.org/research/cause-reports/antibiotic-resi…

You say that as if people had been saying "10 years away" for ages, but I don't think that's true at all. There's some information about historical predictions at https://www.openphilanthropy.org/research/what-should-we-learn-from-pa… (written in 2016) from which (I am including the spreadsheet found at footnote 27) these are some I-hope-representative data points, with predictions from actual AI researchers, popularizers, pundits, and SF authors: 1960: Herbert Simon predicts machines can do all (intellectual) work humans can "within 20 years". 1961: Marvin Minsky says "within our lifetimes, machines may surpass us"; he was 33 at the time, suggesting a not-very-confident timescale of say 40 years. 1962: I J Good predicts something at or above human level circa 1978. 1963: John McCarthy allegedly hopes for "a fully-intelligent machine" within a decade. 1970: I J Good predicts 1994 +- 10 years. 1972: a survey of 67 computer scientists found 27% saying <= 20 years, 32% saying 20-50 years, and 42% saying > 50 years. 1977-8: McCarthy says things like "4 to 400 years" and "5 to 500 years". 1988: Hans Moravec predicts human-level intelligence in 40 years. 1993: Vernor Vinge predicts better-than-human intelligence in the range 2005..2030. 1999: Eliezer Yudkowsky predicts intelligence explosion circa 2020. 2001: Ben Goertzel predicts "during the next 100 years or so". 2001: Arthur C Clarke predicts human-level intelligence circa 2020. 2006: Douglas…

"Sure, you could prepare for imagined eventualities, or you could do the actual work of improving efficiency, reducing waste and unnecessary middle-men, and removing centuries old bureaucracies that are now absurdly pointless in the face of the internet. There is an underlying _desire_ for apocalypse encoded in this type of thinking." OP was written by the person who co-founded GiveWell[1] to make charitable giving more effective, and who while running Open Philanthropy oversaw lots of grants to things like innovation policy[2], scientific research[3], and land use reform[4]. Anyway, more broadly I think you present a false dilemma. You can both prepare for tail risks and also make important marginal and efficiency improvements. [1] https://www.givewell.org/ [2] https://www.openphilanthropy.org/focus/innovation-policy/ [3] https://www.openphilanthropy.org/focus/scientific-research/ [4] https://www.openphilanthropy.org/focus/land-use-reform/

1. Glad to hear! I lead our data team, and we've got a bunch of open roles, multiple of which probably match your background: https://jobs.lever.co/flyzipline/?team=Software%20Engineering Happy to talk it over if any catch your fancy! My email is matt dot e dot fay at gmail dot com 2. There's certainly fewer, but still plenty in this field! For example, if you check the 80,000 Hours Job Board, both Beyond Meat and Impossible Foods are hiring for software positions: https://80000hours.org/job-board/factory-farming/ I can also ask around if you'd like. 3. The x-risk work is definitely speculative, by definition honestly. If the first two are stocks and bonds, this is the angel investing of cause areas: https://www.openphilanthropy.org/blog/hits-based-giving I've stayed away mostly because I was looking for tangibly impactful work (too removed for me), but there are really smart, kind folks making convincing arguments that this is the best use of your time and/or money. For example: - https://80000hours.org/articles/future-generations/ - https://www.vox.com/future-perfect/2018/12/21/18126576/ai-artificial-i… - https://futureoflife.org/2018/09/17/moral-uncertainty-and-the-path-to-…

Fair enough. I guess I'll engage with this silliness once. Let's start with the basics. There's this vague, handwaving notion of "intelligence." In the community pontificating about this it seems to be operationalized as a combination of manipulating abstractions and extracting new, higher level entities from lower levels of abstraction. For the latter, the only AI work to date that I am aware of on the latter is Drescher's 'Made-up Minds'. Multi-layered neural nets end up with interior points that map to some sort of features, but there is a big jump from that to a system being able to develop an abstraction and then operate in terms of it. For the former, computers are very good at it because humans built them as tools to extend the human ability to do it, just like we built hammers because our hands are kind of a lousy tool to hit things with. There is an illusion of sudden, dramatic progress on this kind of intelligence because curves with lots of parameters (neural nets, SVMs, etc.) can fit all kinds of things if you can get the fit to converge. The advances in machine learning have been tricks to get fits to converge and ways of mapping data into spaces where we know how to get fits to converge. This isn't a denigration of the field. It's given us enormous capabilities that we didn't have before, but we need to be clear about what it is before we start extrapolating: curve fitting. Machine learning today lets us fit curves in order to engineer task specific algorithms.…