Now, made curious by a post on a blog that I read from time to time, I checked out the website of the Machine Intelligence Research Institute (MIRI), a kind of Singularitarian think tank. Remember, Singularitarians are futurists who are convinced that progress is accelerating and will soon - that is, within the next few decades - produce a "technological singularity" beyond which everything will be unbelievably different.
The particular MIRI brand of singularitarian appears to be a computer nerd who further believes the way to achieve singularity is to build a self-improving artificial intelligence (AI), because it will then get even more intelligent in a fraction of the time it took to develop it, and the next iteration will improve itself even more, and one day later it has turned into Robot Jesus and there will be immortality, space flight and ponies for all. Because obviously a very intelligent computer must only do some super-quick armchair thinking and it will have solved all social, technological and scientific issues ever. Without having to do painstaking empirical testing of its ideas because it will just be that intelligent. And resource limits won't matter either because, hey, it will just be that intelligent.
(Yes, one does get the feeling that a certain kind of futurist computer nerd considers intelligence, which is sometimes even equated with computing speed, to be magic pixie dust.)
Conversely, they believe that the single greatest danger humanity faces is not resource depletion, antibiotics becoming useless, soil erosion, water shortages, biodiversity loss, or global change induced mass starvation, but instead that Robot Jesus may decide he'd be better off without us useless humans. So the mission of MIRI is to "ensure that the creation of smarter-than-human intelligence has a positive impact" as opposed to Terminator style genocide against humans. Which is obviously a very important goal, much more important than avoiding those other things I mentioned two sentences ago, and so we must all give them lots of donations to finance their, ahem, "research".
Before I looked at their website, I had unavoidably developed some preconceptions about what they might be like. Now that I have seen it, I find it depressing how accurate my preconceptions turned out to be.
Judging from their website, they seem to spend their time approximately as follows:
5% designing websites that look superficially sleek and professional but weird me out once I look closer. (Seriously, "apply to research this math"? "As featured in The New York Times"? Yes, that totally sounds like a legitimate research institute. Please go on.)
10% taking super pretentious staff portraits that they will later find embarrassing
25% standing in front of white boards trying to look intelligent. Whether they succeed is in the eye of the beholder I guess.
30% writing "papers" that use a lot of big words in their titles to distract from the fact that no actual empirical research or AI development appears to be happening at MIRI (at least as far as I can tell). It is basically as if a biologist claimed that we are going to figure out how to repair telomers and thus achieve eternal youth and then spent their entire career merely writing letters to the editor and review articles summarising other people's work instead of, well, developing an enzyme that repairs our telomers.
20% calling for donations.
10% being unable to believe what a sweet gig they managed to land.
As far as their staff goes, this is what the page "our team" shows as of 7 July 2014. There are seven guys who look so young that unless told otherwise I would assume they'd be university students doing an internship but no, apparently one of them is the director; one guy who has made himself look a bit more senior by growing a beard; and one young woman - that's a full 11%! Yay for diversity and inclusiveness!
Their "research associates" are broadly similar in profile, only this time there is actually an emeritus professor among them. On the downside, there are zero women among them, which kind of reduces the percentage of women playing any role at MIRI to near 0%.
Looking at the titles of their publications:
- Problems of Self-Reference in Self-Improving Space-Time Embedded Intelligence
- Definability of Truth in Probabilistic Logic
- Robust Cooperation on the Prisoner's Dilemma: Program Equilibrium via Provability Logic
- A Comparison of Decision Algorithms on Newcomblike Problems
- Ontological Crises in Artificial Agents' Value Systems
- Intelligence Explosion and Machine Ethics
- Intelligence Explosion Microeconomics
- How We're Predicting AI--or Failing To
- Intelligence Explosion: Evidence and Import
Ye gods but this must be awesome. Imagine spending your days fantasising about the potential implications of inventions that will never be made and actually being paid for it. And the sweet thing is, if they play their cards right they can ride that gravy train forever because despite the fact that all we have programmed to date is pretty stupid and that the only self-aware intelligences we have any experience with are biological, nobody has proof positive that a self-aware, highly intelligent and self-improving AI cannot be built. Prove me a negative, won't you?
As long as they don't set a date and perennially keep the moment Robot Jesus will come and make us immortal vaguely in the future (kind of like Christian Jesus has been said to come back any time now for the last two thousand years) MIRI can still ask for our donations in 80 or 200 years. And this being humans we are talking about, they will find takers.
Wishful thinking is a powerful force.