Our mid-2014 strategic plan

 |   |  MIRI Strategy


Events since MIRI’s April 2013 strategic plan have increased my confidence that we are “headed in the right direction.” During the rest of 2014 we will continue to:

  • Decrease our public outreach efforts, leaving most of that work to FHI at Oxford, CSER at Cambridge, FLI at MIT, Stuart Russell at UC Berkeley, and others (e.g. James Barrat).
  • Finish a few pending “strategic research” projects, then decrease our efforts on that front, again leaving most of that work to FHI, plus CSER and FLI if they hire researchers, plus some others.
  • Increase our investment in our Friendly AI (FAI) technical research agenda.

The reasons for continuing along this path remain largely the same, but I have more confidence in it now than I did before. This is because, since April 2013:

  • We produced much Friendly AI research progress on many different fronts, and do not remotely feel like we’ve exhausted the progress that could be made if we had more researchers, demonstrating that the FAI technical agenda is highly tractable.
  • FHI, CSER, and FLI have had substantial public outreach success, in part by leveraging their university affiliations and impressive advisory boards.
  • We’ve heard that as a result of this outreach success, and also because of Stuart Russell’s discussions with researchers at AI conferences, AI researchers are beginning to ask, “Okay, this looks important, but what is the technical research agenda? What could my students and I do about it?” Basically, they want to see an FAI technical agenda, and MIRI is is developing that technical agenda already (see below).

In short, I think we tested and validated MIRI’s new strategic focus, and now it is time to scale. Thus, our top goals for the next 6-12 months are to:

  1. Produce more Friendly AI research.
  2. Recruit more Friendly AI researchers.
  3. Fundraise heavily to support those activities.

Our low-level tactics for achieving these high-level goals will probably change quickly as we try things and learn, but I’ll sketch our current tactical plans below anyway.


1. More Friendly AI research

Our workshops have produced much FAI research progress, and they allowed us to recruit our two new FAI researchers of 2014, but we’ve learned that it’s difficult for “newcomers” (people who haven’t been following the research for a long time) to contribute to FAI research at our workshops. Newcomers need better tutorials and more time to think about the research problems before they can contribute much at the cutting edge (see the next section). Therefore our efforts toward novel research progress in 2014 will focus on:

  • Organizing research workshops attended mostly or solely by “veterans” (people who have been following the research for a long time), such as our May 2014 workshop.
  • Inviting individual researchers to visit MIRI for a few days at a time to work with us on very specific research problems with which they are already familiar.
  • Giving our staff FAI researchers time to make theoretical progress on their own and with each other.


2. Recruiting

Here I talk about “recruiting” rather than “creating a field,” but in fact most of the planned activities below accomplish both ends simultaneously, because they find — and help to activate — new FAI researchers. Still, where there’s a steep tradeoff between recruiting and helping to create a field, we focus on recruiting. FAI research is best done as a full-time career with minimal distractions, and right now MIRI is the only place offering such jobs.

We plan to try many different things to see what most helps for recruiting:

  • We plan to publish an overview of the FAI technical agenda as we see it so far. This should make it easier for potential FAI researchers to engage.
  • We plan to prepare, test, improve, and then deliver (in many different cities) a series of tutorials on different parts of the FAI technical agenda. Besides giving tutorial lectures, we may also organize one-day or two-day workshops that are just for tutorials and Q&A sessions rather than being aimed at making novel research progress.
  • We plan to release a book version of Yudkowsky’s Less Wrong Sequences, in part because we’ve noticed that those who have contributed most to FAI research progress have read — and been influenced by — those writings, and this seems to help when they’re doing FAI research.
  • We’ll continue to help fund SPARC, the lead organizer of which is a frequent MIRI workshop participant (Paul Christiano).
  • We’ll continue to fund independently-organized FAI workshops via our MIRIx program.
  • We may offer widely-advertised cash prizes for certain kinds of research progress, e.g. a winning decision algorithm submitted to a robust program equilibrium tournament, or a certain kind of solution to the Löbian obstacle.
  • We may advertise in venues such as Notices of the AMS (widely read by mathematicians) and Communications of the ACM (widely read by computer scientists).
  • We’re in discussion with development staff at UC Berkeley about a variety of potential MIRI-Berkeley collaborations that could make it easier for researchers to collaborate heavily with MIRI from within a leading academic institution.

We’ve also made our job offer to FAI researchers more competitive.


3. Fundraising

To support our growing Friendly AI research program, we’ve set a “stretch” goal to raise $1.7 million in 2014.

To reach toward that goal, we participated in SV Gives on May 6th and raised ~$110,000, and also won ~$61,000 in matching and prizes from sources that otherwise wouldn’t have donated to MIRI. Our thanks to everyone who donated!

As usual, we’ll also run our major summer and winter fundraising drives, starting in July and December respectively.

This year we’ll explore the possibility of corporate sponsorships, but we are not optimistic about that strategy, because MIRI’s work has little near-term commercial relevance, and we’ve been counseled that corporate sponsors often require more in return from a sponsored organization than the funds are worth.

We’re more optimistic about the potential returns from improving our donor stewardship and donor prospecting, which we’ve already begun, and we’re seeking to hire a full-time Director of Development.

Donor prospecting is unlikely to result in new donors in 2014, but will be important for future fundraising years. This is also the case for our recent efforts to find and apply for grants from both private and public grantmakers. We don’t expect to win much in grants in 2014, but we expect to learn in detail what else we need to do over the next couple years so that we can win large grants in the future, and thereby diversify our funding sources.