Wiki Interview With Eliezer/Donations And Funding

Donations and Funding

Why should someone donate to SIAI?

The fundamental reason someone donates to SIAI is that they believe that humanity ought to be doing something about the Singularity - not just talking about it, but actually doing something about it. Right now, SIAI is the only center of gravity for people who want to take action that directly affects the Singularity - not just doing something else that might eventually have a trickle-down impact on the Singularity, but actions whose intent and first cause is to affect the Singularity.

What benefits do you provide to donors?

The benefit we provide to donors is the Singularity. We may be able to put out a newsletter for donors at the end of the year, and we certainly provide a lot of material online, but the real benefit of supporting the Singularity Institute is that you support the Singularity. That's what it's about. Getting T-Shirts is not the point.

Aside from the tax deduction, we do not have any near-term visible and physical benefits for donors, or at least not at this time. Maybe we will at a future date, but it's never going to be our major priority as an organization. The Foresight Institute provides a number of benefits for donors that are pragmatic and helpful in obtaining funding, but the Institute for Molecular Manufacturing provides much fewer benefits, and we are more closely analogous to IMM than Foresight. I'm not saying that it's *bad* to spend resources providing immediate benefits, just that so far we haven't gotten around to doing it yet. It could provide enough end-result benefit to justify doing it, but it would take time and resources.

What type of donations will SIAI accept?

Right now we don't have a huge amount of organizational infrastructure, as it were, so donating us your used cars and so on is not really an option. Checks and Paypal is mostly it, although a donation of appreciated stock would be okay too. The Institute doesn't really have any preliminary plans on what type of pragmatic benefits would be offered to potential donors, but the most likely benefits would be access - members-only mailing lists, newsletters, and so on. We're very much open to the possibility of volunteer work assisting with such plans.

In PtS, you wrote, "Funding from individuals is the only way of moving into the "long-term" stage (4). While it may be possible to work with minimal funding, I believe the primary strategy should be to find at least one donor wealthy enough that we simply don't have to worry. Ideally, we should become the Silicon Crusade, the heart and ideal of Silicon Valley, the charity of choice for every technomillionaire. Why not? The Singularity deserves to be a crusade, and the meme is powerful enough. We're trying to massively alter the fate of the human species; as I've remarked elsewhere in this document, trying to do it on a shoestring is silly. Our ideas are on the grand scale, our goals are on the grand scale, and there's no reason to think small." Has SIAI attempted to acquire funding from one wealthy donor? Does SIAI have near-term interest in trying to become the "charity of choice for every technomillionaire"? Do you have any ideas on how the Institute, purportedly under the direction of an Executive Director, would attempt this? Do you believe it's a likely possibility?

Unfortunately, you can't just go up to billionaires and ask for funding, *they* find *you*. If there's any way around this, none of us have enough fundraising experience to know it, but my impression is that it is universal. Everyone wants a slice of the billionaire's time, not to mention their wallets, so the billionaire filters it all out and just surfs the web or whatever until s/he runs across something that s/he thinks is interesting. Millionaires are also typically pretty hard to contact, read Eric S. Raymond's Slashdot post when VALinux went public. To become the charity of choice for Silicon Valley would require a memetic shift. One would have to start with influential but nonwealthy individuals, who are harder to contact, but not impossible, and try to get the memes spreading outward from those points. This is again the sort of thing that probably requires an Executive Director, because I don't necessarily *know* how to do these things, or even whether it is the most practical way of getting to the point of sustaining an AI project. Hopefully the Executive Director will know more about it than I do.

Is the Institute presently a shoestring operation?

Not quite as shoestring as it was when we were just starting out, but in a word, yes.

To date, how much funding have you received?

We've gotten, I would say, around $12K in outside donations [(not from Brian Atkins) since our founding. [01/16/03: To my knowledge, the Institute's present outside funding is around $45,000.]

What type of efforts have been made to acquire funding?

As I'm the only full-time person and I've been concentrating on the AI research so far, there is relatively little effort available for fundraising - mostly we get funding from people who run across us on the Web, but I do attend conferences like the Foresight Gathering, Extros, and so on, and look for funding there.

Have you applied or are you waiting to apply the funding you have already received?

The problem is that for most of the things we'd like to do, there's a threshold where we have to accumulate a certain amount of funding before we can get started. We're currently accumulating toward a level of funding that might be usable to hire a full-time Executive Director to do fundraising and so on, if we can find someone who meets our exacting specifications and who is willing to work for a very low salary, but we may not be able to find an Executive Director. In that case we would go on accumulating funding toward the initiation of the primary AI project. There would also be the possibility of organizing a Singularity conference, but I'm not sure we presently have enough attendees to support that, even if we have enough funding.

Please state some of the specifications you're seeking in an Executive Director.

Ideally, we'd like a deep understanding of Singularity issues, dedication to the Singularity, previous experience with nonprofit fundraising, or failing that previous business experience, and strong writing ability. Unfortunately, unless you're a really good fundraiser, it may be some time before we can afford to pay you more than an insultingly small fraction of what you're worth. But the Singularity is worth it, right?

You probably realize the large filters that exist within your requirements. Acquiring such an individual or even one less ideal within one to two years may prove very difficult. Please state less ideal specifications.

At our current level of funding we can afford to hire an ideal Executive Director, but not a non-ideal one - it wouldn't be the best use of funds.

Why should someone fund the Institute's AI work instead of other AI work?

Well, the only real answer to this question is "Because the Institute's technical approach is more likely to succeed than any current competing approach", but to appreciate this answer on its own terms requires Cognitive Science expertise. (note that I don't just say "AI expertise") However, there are several things about the Institute that are immediately visible, the most important of which is that we are not a project started to do something commercial that wants to try for General Intelligence and seed AI as a side goal, we are here to climb this enormous mountain however long it takes. We are here to try for the Singularity *only*. Anything with commercial benefits or philanthropic benefits is something we do while trying for the Singularity, not the other way around. General Intelligence is such an enormous challenge that you cannot afford to be distracted while doing it, nor is it realistic to expect to achieve it while trying for something else, nor is it something where the researchers can promise a specific timeframe in advance, which is why there is a Singularity Institute and not a Singularity Inc.

(copied to Why Donate To SIAI Instead Of Another AI Project)

Do you or does anyone else have a running list of commercial or philanthropic benefits that could be achieved while achieving the Singularity?

Oh, sure, there are quite a few. My favorite one would be an AI that could teach reading, writing, and arithmetic, maybe more advanced subjects, reduce America's education problem to a hardware manufacturing problem, at least for the lower grade levels, and maybe even some of the higher ones. That would be a good example of an absolutely pure benefit. The question is whether you can do it more than six months in advance of the Singularity, but there's a serious probability that it might. More mundane commercial applications would be things like data-mining, but you can easily think of dozens of commercial and philanthropic benefits, everything that requires a computer to display a level of intelligence short of human, but substantially in advance of what there is now. Another example would be low-grade understanding of an enormous amount of material, where I am thinking specifically of the Literature - that is, the scientific literature. You don't necessarily need human-level understanding to read *more of* the Literature than any one human being nowadays can manage. Also, being able to suggest to paper authors that they talk to each other, and being able to respond to a query like "what computational architectures have been proposed for the cerebellum", maybe even going beyond providing abstracts to contrast some of the views, if that degree of semantic understanding can be achieved (earlier than six months before the Singularity). That could be an enormous benefit to the scientific community. Another of these is that we are talking about seed AI - Recursive Self-Improvement - and Friendly AI. Since we started talking about these things other projects have also started talking about them, but we are the ones who have fully developed theories of it. Similarly, we are past saying "But we need to discuss the ethical implications of AI" and are now saying "Here is our current theory of the ethical implications of AI, and here is what we are doing about it." These are properties that you would expect to see in any project that was taking the Singularity seriously, or that you would expect to see for any project trying for Human-Equivalence, which, even *without* Hard Takeoff and Singularity, would have a cosmos-shattering impact. AI projects that have thrown the words "human equivalence" around like it was trivia clearly do not really expect to build a person. We do and we act like it.

What projects would SIAI presently like to initiate, and how much funding will be required? Please provide present-day cost estimations for projects to be initiated at different levels of funding.

My guess is that 200K-400K would be needed to get over the threshold for starting the AI project, $500K would be better, and in the long term SIAI will probably need to expand to $3M-$10M per year if not more. Hopefully the initial AI project will be able to get to the point of either providing a really impressive demo, or of providing an application that can be leased, without requiring those very high levels of funding. Hopefully.

$10M/year to achieve it by 2010? 2015? 2020?

Trying to project a date is a wide enough range in itself, trying to project the effect of funding on the date just throws me off entirely. I would say something along the lines of $500K/year might easily turn out to take until 2020, $5M/year might put us in range of 2010 or 2015, and $50M/year might let us do it in 2008, but I came up with that estimate over the course of 30 seconds because I'd honestly never thought about it before. I just think in terms of getting as much funding as possible so we can do it as soon as possible. It's not like there's any better use for money.

Based on present understanding and projected success, do you have a time estimate when you will have an AI capable of achieving an impressive demo?

I would guess one to five years after the project starts, with two to four years being more likely, bearing in mind that I tend to be very conservative about what I would call an "impressive demo". I guess if you're just talking about demonstrating something in advance of current AI capabilities, it might be more like one to three years, which is the time to the "functioning supersystem" milestone, but all such time estimates are fundamentally uncertain. I really do want to emphasize that on basic research problems there is no way of knowing in advance how long it will take.

(copied to Singularity Questions/MiscellaneousQuestion1)

If AI is the primary project, what other secondary projects would you like to initiate once the appropriate level of funding is achieved?

If we have funding that we can't spend on AI, then there are a *lot* of other things that can be done to support the Singularity: public education, running conferences, supporting research on the Cognitive Science priorities I talked about earlier, initiating other AI projects under the SIAI umbrella, supporting the Singularitarian community, supporting non-AI paths to the Singularity (which are much harder to accelerate, hence the focus on AI) such as brain-computer interfaces, supporting research on the Singularity proper - theoretical analysis of Hard Takeoffs, Existential Risks, and so on, etc.

How would SIAI spend $10K? $100K? $500K? $1M? $10M? $100M? $1B? Please provide approx. spending uses.

$100K would put us in range of hiring an Executive Director. $500K would enable us to rent offices, hire programmers, buy equipment, and start the AI project. $1M would enable us to hire more programmers. $10M would probably enable us to hire as many programmers as we can safely use, pay them reasonably, and have enough money left over for Singularity memetics, Flare, maybe a few experiments in Cognitive Science, major annual conferences, and so on. $100M would mean that it was time to start forking off many AI projects, bring Goertzel and Voss into the fold, support ExI / Foresight / WTA, run advertising campaigns, and fork off Earthweb projects and Cognitive Science studies. $1B would be as above, only more so, I guess. Actually, $1B might be enough to get started on human intelligence enhancement if there's any (moral, socially acceptable) way to get around the FDA. Of course, there is always the Singularity smartness effect: you hire an experienced philanthropic director, and s/he has a better idea than you of how to spend money effectively. I also made a post to Extropians once, as part of some debate started by David Brin (I think), on eight good uses for a billion dollars.

Why has the Institute had difficulty achieving the mandatory level of funding to initiate their next project(s)? Are there not enough individuals or groups explaining SIAI's reasoning and work in an attempt to acquire funding or gifts for the organization? What needs to be done to change this situation?

Part of the problem, as you point out, is the chicken-and-egg problem of not having enough funding to hire the people to get the funding and so on. The deeper problem is that modern culture encourages a passive attitude toward the future: Changes are supposed to happen as the result of someone else's efforts. "Scientists" discover things, but no one particular scientist that you can go up to and support. People argue over what *will* happen, it's a sport, like politics, but unlike politics, nobody is willing to invest any actual effort into changing the future, even though they may be willing to invest time, and sometimes even money, in fighting for their *view* of the future as a meme. The leaders of futurism defend particular views, but they don't actually defend particular futures in the sense of *working* toward them, they just argue that it *will* happen, or occasionally that it *should* happen, or that someone *else* should change their behavior - the "government" should fund this, "corporations" should stop doing that, but not actually becoming involved themselves. It seems to be a flaw in the way people are built to view the world. Bill Joy is still working on peer-to-peer distributed computing, whose sole apparent purpose is to make it impossible to shut down an Internet AI, even though he can't possibly need the money, and even after publishing that article in Wired it hasn't occurred to him to change *his personal* behavior patterns as a result of his beliefs, much less take actions that might result in some personal inconvenience to him. Someone *else* is the problem. It's always "we need to do X" and never "I am hereby announcing that I will do X". That's the flaw. Humanity easily has enough resources to confront the Singularity firmly, if the individual humans can stop waiting for someone else to do it.

Does SIAI currently have any concentrated recruiting or fundraising efforts in place (i.e., is anyone devoting a substantial amount of time to fundraising)? If not, does SIAI have plans to begin concentrated recruiting or fundraising in '02 or '03? What needs to happen for this to occur? What methods will be used?

Unless we find a perfect Executive Director - which *could* happen, but is not something we can rely on - then recruiting or fundraising would have to be done either by volunteers or by me. I would like to try and set up a volunteer structure for recruiting and fundraising if I can find the time, and if I can find the volunteers. I would also like to try to hand off some more SIAI functions. In other words, the main question is whether I can free up the time to get that done.

Could you elaborate on what kind of volunteer structure you have in mind? Any models for references?

I'm not sure a volunteer structure would be needed aside from a mailing list, unless there are a lot of volunteers. The one main thing you need is a volunteer Volunteer Coordinator [01/11/03: Christian Rovner is SIAI's Volunteer Coordinator.]

What would a Volunteer Coordinator do?

The volunteer coordinator would check the progress of people who volunteered to do something specific, try to assign tasks to people who volunteer but don't have anything in particular in mind, maintain a web page with open tasks, and give progress updates to the rest of the universe. I'm not a very good volunteer coordinator myself, so obviously one of the requisites for this task is: "Doesn't need Eliezer Yudkowsky to figure out for him/her what a Volunteer Coordinator does" i.e., someone with independent drive.

I think there are probably many individuals who could spend a portion of their time, in their local state, attempting to explain the reasoning behind the singularity to wealthy individuals or groups capable of supporting singularity-related efforts. A collaborative effort across the United States and other countries would undoubtedly be a worthwhile effort. In the United States alone, a ridiculous amount of resources are wasted or inappropriately allocated without increasing our net-positive affect on the world. If even a very small fraction of those resources were filtered into the efforts of the Extropy Institute, Foresight Institute, the Singularity Institute, or the World Transhumanist Association, then the allocated resources would actually have a high probability of future-relevance. What do you think?

If you think that you can get these individuals organized, and that you can get them access to wealthy individuals, then organize them and start the project, but we aren't there yet. Humanity can easily afford to support the Singularity Institute, as I said. The question is getting a tiny but sufficient fraction of humanity to realize this. Just because something is a good idea doesn't mean that it will happen. Just because something is true doesn't make it easy to explain. Just because a lot of money is being wasted doesn't mean that it's easy to make it accomplish something useful.

What is SIAI's present policy on allowing an individual or group access to its non-published work? For example: If an individual or group is interested in donating to SIAI but would first like access to the Institute's non-published work, what would be the Institute's response? Would it depend on how much the individual or group is offering for donation? If IBM, for example, is interested in acquiring SIAI's off-line AI work, would they be allowed access to it? If not, why not?

Right now we don't have a major body of off-line work. When we do, it will probably be the AI design. I can see offering a sufficiently major donor the ability to look at that, but I can't see selling it or licensing it, even as a design - that's an action we would take only if it were independently deemed a good thing for the Singularity, which is possible, but is conceptually separate from donations. Incidentally, we already have one heck of a body of work online, so it's hard to see what specific question someone would have that couldn't be answered either by that or in a couple of paragraphs.

(copied to Singularity Questions/SIAIq1)

If a potential donor wanted access to your AI design, would he or she be required to sign a NDA?

I would expect so, assuming that the AI design was available to potential donors, which would probably depend on my estimated level of "how dangerous is this in the hands of someone with $10M who doesn't believe in FAI?"