Ray Solomonoff on Speed of AI Takeoff

In 1985, Ray Solomonoff offered his thoughts on six milestones in AI and the economic and technological growth that might be expected when generally intelligent AI is developed. The paper is called “The Time Scale of Artificial Intelligence: Reflections on Social E ffects”.

Here is the abstract:

Six future milestones in AI are discussed. These range from the development of a very general theory of problem solving to the creation of machines with capacities well beyond those of a single human. Estimates are made for when these milestones will occur, followed by some suggestions for the more e ffective utilization of the extremely rapid technological growth that is expected.

When I read lines like that last sentence, what I see nowadays is “extremely scary technological growth”. Rapid growth is scary when that growth is controlled by systems that may not optimize reality in ways that we explicitly value. (See “The Future of Human Evolution” for an explanation.)

A select milestone:

Milestone C. A critical point in AI development would be a machine that could usefully work on the …

Read More

Ray Solomonoff, 1926-2009

Ray Solomonoff, the father of algorithmic probability theory and one of the founding fathers of Artificial Intelligence, died December 7th after a brief illness.

Solomonoff was a pioneer of probabilistic thinking in AI, and in general. It is my own view that the value of probabilistic thinking is the single most important insight about reality that humanity has ever had, and Solomonoff helped add to that great edifice with his idea of Algorithmic Probability.

Solomonoff was the founder of universal inductive inference, which gives a mathematically optimal method of predicting the next bit of sensory information in a sequence based on prior information. (Unfortunately, it is incomputable, though computable approximations have been used throughout the field of AI.) As far as I know, Solomonoff made the first mathematically rigorous attempt at automated sequence prediction.

Solomonoff’s work is being carried forward by theorists such as Marcus Hutter, Juergen Schmidhuber, and Shane Legg, among many others.

Just last week I posted on AIXI, which is essentially a marriage between …

Read More

Winter Edition of h+ Magazine Available

The Winter edition of h+ magazine is out. There is strong representation from the Singularity Institute among the article contributors, with pieces by myself, Tom McCabe, and Ben Goertzel. Definitely check out McCabe and Goertzel’s interesting articles.

On page 12, I talk about Ned Seeman’s latest totally-awesome robotic nanomanipulating arm. It places atoms and molecules with 100% accuracy.

The theme of the issue is DIY, which is great. It makes use of leverage. If the goal of h+ magazine is to promote scientific research into human enhancement, then promoting DIY technology is an effective use of their money.

In the DIY realm, I’ve lately been following 3D fabbing with interest, as have many mainstream news sources.

I notice that the website has a thoughtful article on the significance of 4chan. Jason Louv writes:

Yet what the media has failed to grasp is what 4chan can tell us about where we’re headed. The Chans aren’t the freak sideshow of the Internet. They are the heart and soul of the Internet. And they are the …

Read More

The Uncertain Future: Now in Beta

A webapp that I worked on with Steve Rayhawk, Anna Salamon, Tom McCabe, and Rolf Nelson, during the Singularity Institute for Artificial Intelligence Summer 2008 Research Program, with helpful discussions with a few others, is now in beta and ready for public announcement. It is called The Uncertain Future.

The Uncertain Future represents a new kind of futurism — futurism with heavy-tailed, high-dimension probability distributions. In fact, that’s the name of the paper presented at the European Conference on Computing and Philosophy that unveiled the project: “Changing the frame of AI futurism: From storytelling to heavy-tailed, high-dimensional probability distributions”.

Most futurism is about telling a story — more like marketing than an honest attempt at uncovering the possible range of what the future may hold. Better than creating a single story is scenario building — but this falls short as well. Scenario building is human nature, but it leaves us susceptible to anchoring effects where we overestimate the probability of vivid scenarios. To quote “Cognitive Biases Potentially Affecting Judgment of Global Risks”, page 6:

The conjunction …

Read More

New(-ish) ScienceBlogs Blog Focusing on AI?

2009 saw a lot of mainstreaming of “transhumanist” ideas, foci, and emphases. As I recently pointed out, Foreign Policy magazine gave this phenomenon a nod by including two transhumanists on their list of 100 global thinkers.

I am particularly interested in any possible mainstreaming of AGI and Friendly AI ideas, for obvious reasons. These ideas are not mainstreaming as fast as “wow-tech” like life-extension or cybernetics, so watching for it is even more challenging and interesting. That’s why this ad on the ScienceBlogs network caught my eye:

It links to Collective Imagination, a relatively new blog on the ScienceBlogs network with an about page that doesn’t mention AI at all. But, click the ad and you go to their front page, which currently is all about AI. On November 19th, their head blogger, Greg Laden, bought into the IBM “cat brain” deliberately deceptive news item, but then did a double-take a week later. What is interesting about his double-take is that he takes the time to …

Read More

IQ: “Lonely Ice Floe” or Consensus Science?

Malcolm Gladwell calls those who accept the Mainstream Science on Intelligence statement “IQ fundamentalists”, but the reality of g and the predictive validity of intelligence tests are widely accepted as consensus science by intelligence researchers, with some caveats. Reading Eurekalert and PhysOrg, I see press releases practically every day that analyze the correlation of intelligence with a variety of genetic and environmental factors. Here’s one from yesterday:

Fit teenage boys are smarter But muscle strength isn’t the secret, study shows In the first study to demonstrate a clear positive association between adolescent fitness and adult cognitive performance, Nancy Pedersen of the University of Southern California and colleagues in Sweden find that better cardiovascular health among teenage boys correlates to higher scores on a range of intelligence tests — and more education and income later in life.

“During early adolescence and adulthood, the central nervous system displays considerable plasticity,” said Pedersen, research professor of psychology at the USC College of Letters, Arts & Sciences. “Yet, the effect of exercise on cognition remains poorly understood.”

Read More

CNN Casually Mentions Human Enhancement in a Positive Light

From the “Top 10 Scientific Discoveries of 2009″ on CNN.com:

3. Gene therapy cures color blindness

Modern science already offers ways to enhance your mood, sex drive, athletic performance, concentration levels and overall health, but a discovery in September suggests that truly revolutionary human enhancement may soon move from science fiction to reality. A study in Nature reported that a team of ophthalmologists had injected genes that produce color-detecting proteins into the eyes of two color-blind monkeys, allowing the animals to see red and green for the first time. The results were shocking to most — “We said it was possible, but every single person I talked to said, ‘Absolutely not,’ ” said study co-author Jay Neitz of the University of Washington — and raised the possibility that a range of vision defects could someday be cured. That’s a transformative prospect in itself, but the discovery further suggests that it may be possible to enhance senses in “healthy” people too, truly revolutionizing the way we see the world.

The only people who bother to object to human enhancement …

Read More

Hit and Run: “We can all agree that Ron Bailey defecates better transhumanism coverage than I can ever hope to produce”

Coverage of the recent H+ Summit is available at Reason’s Hit and Run blog. Here is a funny bit:

Futurist John Smart is wrapping up the Humanity + Summit by noting that human enhancement believers are too focused on pie-in-the sky visions. Instead of making weird flying-car predictions about the far future, transhumanoids should be pointing to contemporary advances.

John Smart is known for predicting that the Earth will be artificially collapsed into an engineered black hole in an effort to compress matter and energy to more efficiently run uploads. I think he is right.

There is coverage of Aubrey and Todd Huffman’s “Rasputin beards”. An informal poll found that three out of three women found Todd’s finger magnet implant hot.

More reporting:

I don’t know enough about transhumanism to say whether the movement is at any kind of crossroads, but I was struck by how modest the claims were at this event — in addition to all the calls for empathy, which I referred to yesterday. Toe …

Read More

MIT News: “Rethinking Artificial Intelligence”

Here is the article. It contains coverage of Ed Boyden’s brain-computer interfacing efforts, along with commentary by Dennett, Pinker, Minsky, Gershenfeld, and Boyden. The important paragraph is here:

The new project, launched with an initial $5 million grant and a five-year timetable, is called the Mind Machine Project, or MMP, a loosely bound collaboration of about two dozen professors, researchers, students and postdocs. According to Neil Gershenfeld, one of the leaders of MMP and director of MIT’s Center for Bits and Atoms, one of the project’s goals is to create intelligent machines — “whatever that means.”

You can read up on the Mind Machine Project here.

Read More

The World’s Smallest Snowman — So What?

Various futurist and transhumanists are abuzz about the world’s smallest snowman. Just like IBM’s recent deliberately misleading “cat brain” announcement, I consider this non-news. As far as I can tell, this doesn’t represent any sort of interesting technological advance. Microscale tin beads are not new. Focused ion beams are not new. Ion beam deposited metals are not new. This is just a gimmick.

I am not a nanoscientist. I am just a guy who reads news feeds like Nanowerk/CRN/Foresight and skims papers once in a while. But the way that the transhumanist and futurist community is reacting to this at all makes me roll my eyes. The majority of futurists lack scientific knowledge of any depth because they are too busy flying around, attending meetings, giving interviews, and running scenario sessions. Paying someone to sit around and read papers is not a common practice outside of academia.

Some portions of the press release are especially banal:

The snowman is mounted on a silicon cantilever from an atomic force microscope whose sharp tip ‘feels’ surfaces creating topographic surveys …

Read More

Computable AIXI — Should We Be Afraid?

An interesting point of dispute in the field of Artificial General Intelligence concerns the relevance/irrelevance of optimal formal models of inference to creating computationally feasible AI. On one side we have figures like Marcus Hutter and Jürgen Schmidhuber, the creators of the formal models AIXI and the Gödel machine respectively. What is AIXI? From the source:

Decision theory formally solves the problem of rational agents in uncertain worlds if the true environmental prior probability distribution is known. Solomonoff’s theory of universal induction formally solves the problem of sequence prediction for unknown prior distribution. We combine both ideas and get a parameterless theory of universal Artificial Intelligence. We give strong arguments that the resulting AIXI model is the most intelligent unbiased agent possible.

What is a Gödel machine?

We present the first class of mathematically rigorous, general, fully self-referential, self-improving, optimally efficient problem solvers. Inspired by Kurt Gödel’s celebrated self-referential formulas (1931), a Gödel machine (or `Goedel machine’ but not `Godel machine’) rewrites any part of its own code as soon as it has found a …

Read More