Monthly Archives: May 2008

A Premature Word on AI

By Eliezer Yudkowsky · May 31, 2008 1:48 pm · Comments (68)

Followup to:  A.I. Old-Timers, Do Scientists Already Know This Stuff?

In response to Robin Hanson’s post on the disillusionment of old-time AI researchers such as Roger Schank, I thought I’d post a few premature words on AI, even though I’m not really ready to do so:

Anyway:

I never expected AI to be easy.  I went into the AI field because I thought it was world-crackingly important, and I was willing to work on it if it took the rest of my whole life, even though it looked incredibly difficult.

I’ve noticed that folks who actively work on Artificial General Intelligence, seem to have started out thinking the problem was much easier than it first appeared to me.

In retrospect, if I had not thought that the AGI problem was worth a hundred and fifty thousand human lives per day – that’s what I thought in the beginning – then I would not have challenged it; I would have run away and hid like a scared rabbit.  Everything I now know about how to not panic in the face of difficult problems, I learned from tackling AGI, and later, the superproblem of Friendly AI, because running away wasn’t an option.

Try telling one of these AGI folks about Friendly AI, and they reel back, surprised, and immediately say, "But that would be too difficult!"  In short, they have the same run-away reflex as anyone else, but AGI has not activated it.  (FAI does.)

Roger Schank is not necessarily in this class, please note.  Most of the people currently wandering around in the AGI Dungeon are those too blind to see the warning signs, the skulls on spikes, the flaming pits.  But e.g. John McCarthy is a warrior of a different sort; he ventured into the AI Dungeon before it was known to be difficult.  I find that in terms of raw formidability, the warriors who first stumbled across the Dungeon, impress me rather more than most of the modern explorers – the first explorers were not self-selected for folly.  But alas, their weapons tend to be extremely obsolete.

Continue reading "A Premature Word on AI" »

GD Star Rating
loading...

A.I. Old-Timers

By Robin Hanson · May 31, 2008 6:00 am · Comments (47)

Artificial Intelligence pioneer Roger Schank at the Edge:

When reporters interviewed me in the 70’s and 80’s about the possibilities for Artificial Intelligence I would always say that we would have machines that are as smart as we are within my lifetime. It seemed a safe answer since no one could ever tell me I was wrong. But I no longer believe that will happen. One reason is that I am a lot older and we are barely closer to creating smart machines. 

I have not soured on AI. I still believe that we can create very intelligent machines. But I no longer believe that those machines will be like us….

What AI can and should build are intelligent special purpose entities. (We can call them Specialized Intelligences or SI’s.)  Smart computers will indeed be created. But they will arrive in the form of SI’s, ones that make lousy companions but know every shipping accident that ever happened and why (the shipping industry’s SI) or as an expert on sales (a business world SI.) … So AI in the traditional sense, will not happen in my lifetime nor in my grandson’s lifetime. Perhaps a new kind of machine intelligence will one day evolve and be smarter than us, but we are a really long way from that.

This was close to my view after nine years of A.I. research, at least regarding the non-upload A.I. path Schank has in mind.  I recently met Rodney Brooks and Peter Norvig at Google Foo Camp, and Rodney told me the two of them tried without much success to politely explain this standard "old-timers" view at a recent Singularity summit.  Greg Egan recently expressed himself more harshly:

Continue reading "A.I. Old-Timers" »

GD Star Rating
loading...
Tagged as: ,

Empty Space

By Robin Hanson · May 30, 2008 9:00 pm · Comments (21)

Yesterday I talked at an "International Space Development Conference" on interstellar colonization, alongside Air Force futurist Peter Garretson and polymath, L5 cofounder, and Scientology-thorn Keith Henson.  Both Keith and Peter described solar power satellites as our last hope against a peak-oil disaster, where world population would fall in proportion to oil production.  In the exhibit hall, a space elevator guy said they are feasible now and would cost only five billion dollars to let us put many tons a day into orbit for a few dollars a pound. 

On Monday I was interviewed on BBC Radio about what if any economic value we get out of NASA.  I said the benefits were mostly like the pyramids – national prestige and being part of history.  I was not told I would be debating a NASA fan, who said gave the tang/teflon argument that all research is good, and also said we don’t want to be like the Chinese who pulled their advanced ships back just before the Europeans took over the world.  There was not time to respond.

Sigh.  The US government spends more on space research than on NIH and NSF combined, which most scientists consider far out of proportion to its science value.  Most any ambitious tech project, like floating cities, 3DTV, or robot mules, gives similar indirect tech spinoffs per dollar spent, and surely we can find other projects with larger direct payoffs.  Sure the Chinese might have colonized the Americas, but we can see now there are no similarly lush gardens accessible in space – we’ll colonize Antarctica and the Earth oceans long before, as these are far less harsh environments with plenty of the sunlight and materials which are mainly what space has to offer.  Yes someday we’ll run out of stuff here on Earth, but that day is far off.  We’ll probably use kite power before solar satellites, world population is not proportional to oil production, and hopes for more than a tiny space tourism market anytime soon are pure fantasy. 

What we have is a strong and growing demand for satellites, especially by the military.  But that demand curve slopes down a lot, leading me to doubt if there is enough demand to cover the (large) fixed cost of a (militarily vulnerable) space elevator soon.  I could imagine the Chinese building one mainly to get the respect they crave, but like the US public they would probably not want to publicly admit that was their main motivation.

Added 31May: The demand for protecting Earth from asteroids is much smaller than for satellites, and deep mine colonies would better protect humanity from extinction.

GD Star Rating
loading...
Tagged as:

Class Project

By Eliezer Yudkowsky · May 30, 2008 8:23 pm · Comments (21)

Followup to:  The Failures of Eld Science, Einstein's Superpowers

"Do as well as Einstein?" Jeffreyssai said, incredulously.  "Just as well as Einstein?  Albert Einstein was a great scientist of his era, but that was his era, not this one!  Einstein did not comprehend the Bayesian methods; he lived before the cognitive biases were discovered; he had no scientific grasp of his own thought processes.  Einstein spoke nonsense of an impersonal God – which tells you how well he understood the rhythm of reason, to discard it outside his own field! He was too caught up in the drama of rejecting his era's quantum mechanics to actually fix it.  And while I grant that Einstein reasoned cleanly in the matter of General Relativity – barring that matter of the cosmological constant – he took ten years to do it.  Too slow!"

"Too slow?" repeated Taji incredulously.

"Too slow!  If Einstein were in this classroom now, rather than Earth of the negative first century, I would rap his knuckles!  You will not try to do as well as Einstein!  You will aspire to do BETTER than Einstein or you may as well not bother!"

Jeffreyssai shook his head.  "Well, I've given you enough hints.  It is time to test your skills.  Now, I know that the other beisutsukai don't think much of my class projects…"  Jeffreyssai paused significantly.

Brennan inwardly sighed.  He'd heard this line many times before, in the Bardic Conspiracy, the Competitive Conspiracy:  The other teachers think my assignments are too easy, you should be grateful, followed by some ridiculously difficult task – 

Continue reading "Class Project" »

GD Star Rating
loading...

Einstein’s Superpowers

By Eliezer Yudkowsky · May 30, 2008 2:40 am · Comments (56)

Followup to:  Einstein’s Speed, My Childhood Role Model, Timeless Physics

There is a widespread tendency to talk (and think) as if Einstein, Newton, and similar historical figures had superpowers – something magical, something sacred, something beyond the mundane.  (Remember, there are many more ways to worship a thing than lighting candles around its altar.)

Once I unthinkingly thought this way too, with respect to Einstein in particular, until reading Julian Barbour’s The End of Time cured me of it.

Barbour laid out the history of anti-epiphenomenal physics and Mach’s Principle; he described the historical controversies that predated Mach – all this that stood behind Einstein and was known to Einstein, when Einstein tackled his problem…

And maybe I’m just imagining things – reading too much of myself into Barbour’s book – but I thought I heard Barbour very quietly shouting, coded between the polite lines:

What Einstein did isn’t magic, people!  If you all just looked at how he actually did it, instead of falling to your knees and worshiping him, maybe then you’d be able to do it too!

Maybe I’m mistaken, or extrapolating too far… but I kinda suspect that Barbour once tried to explain to people how you move further along Einstein’s direction to get timeless physics; and they sniffed scornfully and said, "Oh, you think you’re Einstein, do you?"

Continue reading "Einstein’s Superpowers" »

GD Star Rating
loading...

Intro to Innovation

By Robin Hanson · May 29, 2008 6:00 am · Comments (21)

The topic of innovation comes up often here, so let us review some basics:

  1. Systems are parts in a structure; innovations are better part or structure designs. An innovation embodies insights whose value depends on a context, and so changes with that context.
  2. Most net growth in the number or size of large systems has been due to collecting innovations.
  3. Wars, quakes, and diseases may be distributed so most impact comes from the few largest instances.  In contrast, in large systems most innovation value comes from many small innovations.  Even big innovations require many matching small innovations to be viable.
  4. Innovation rates increase when early innovations make it easier to pursue later innovations, and decrease when the most valuable easiest innovations tend to be pursued first.  Steady (exponential) growth suggests that these factors roughly cancel.  Since growth rates commonly increase then decrease, usually the second factor eventually wins.   
  5. Innovation in large systems comes mostly from part innovation, so system innovation is steadier than part innovation, and the largest systems grow steadiest.
  6. System structures vary in how well they encourage and test innovations locally and then distribute the best ones widely.  Better structures for this are meta-innovations.
    • Good modularity reduces the need to match innovations in differing parts. 
    • Good abstraction puts similar innovation problems within the same part.
  7. If a barrier isolates two systems, the faster growing one eventually dominates.  A system that better promotes innovation can lose to a system with a larger source of innovation. 
  8. In large innovation pools, similar innovations commonly arise from several semi-independent sources at nearly the same time.  No single source was essential.
  9. Current human society can give incentives to innovate too much, when innovation is used to signal, and to innovate too little, when innovators are not paid the full value gained by others.

I learned this stuff long ago so I have little idea how commonly known this all is. 

GD Star Rating
loading...
Tagged as:

Timeless Causality

By Eliezer Yudkowsky · May 29, 2008 2:45 am · Comments (40)

< ?xml version="1.0" standalone="yes"?> < !DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd">

This post is part of the Quantum Physics Sequence.
Followup to:  Timeless Physics

Julian Barbour believes that each configuration, each individual point in configuration space, corresponds individually to an experienced Now – that each instantaneous time-slice of a brain is the carrier of a subjective experience.

On this point, I take it upon myself to disagree with Barbour.

There is a timeless formulation of causality, known to Bayesians, which may glue configurations together even in a timeless universe.  Barbour may not have studied this; it is not widely studied.

Such causal links could be required for “computation” and “consciousness” – whatever those are.  If so, we would not be forced to conclude that a single configuration, encoding a brain frozen in time, can be the bearer of an instantaneous experience.  We could throw out time, and keep the concept of causal computation.

Continue reading "Timeless Causality" »

GD Star Rating
loading...

Overconfidence & Paternalism

By Robin Hanson · May 28, 2008 6:00 am · Comments (27)

Paul Graham tries to explain paternalism: 

Parents know they’ve concealed the facts about sex, and many at some point sit their kids down and explain more. But few tell their kids about the differences between the real world and the cocoon they grew up in. Combine this with the confidence parents try to instill in their kids, and every year you get a new crop of 18 year olds who think they know how to run the world.

Don’t all 18 year olds think they know how to run the world? Actually this seems to be a recent innovation, no more than about 100 years old. In preindustrial times teenage kids were junior members of the adult world and comparatively well aware of their shortcomings. They could see they weren’t as strong or skillful as the village smith. In past times people lied to kids about some things more than we do now, but the lies implicit in an artificial, protected environment are a recent invention. Like a lot of new inventions, the rich got this first. Children of kings and great magnates were the first to grow up out of touch with the world. Suburbia means half the population can live like kings in that respect.  …

Continue reading "Overconfidence & Paternalism" »

GD Star Rating
loading...
Tagged as: , ,

Timeless Beauty

By Eliezer Yudkowsky · May 28, 2008 12:32 am · Comments (54)

This post is part of the Quantum Physics Sequence.
Followup to:  Timeless Physics

One of the great surprises of humanity’s early study of physics was that there were universal laws, that the heavens were governed by the same order as the Earth:  Laws that hold in all times, in all places, without known exception. Sometimes we discover a seeming exception to the old law, like Mercury’s precession, but soon it turns out to perfectly obey a still deeper law, that once again is universal as far as the eye can see.

Every known law of fundamental physics is perfectly global. We know no law of fundamental physics that applies on Tuesdays but not Wednesdays, or that applies in the Northern hemisphere but not the Southern.

In classical physics, the laws are universal; but there are also other entities that are neither perfectly global nor perfectly local. Like the case I discussed yesterday, of an entity called "the lamp" where "the lamp" is OFF at 7:00am but ON at 7:02am; the lamp entity extends through time, and has different values at different times.  The little billiard balls are like that in classical physics; a classical billiard ball is (alleged to be) a fundamentally existent entity, but it has a world-line, not a world-point.

In timeless physics, everything that exists is either perfectly global or perfectly local.  The laws are perfectly global.  The configurations are perfectly local – every possible arrangement of particles has a single complex amplitude assigned to it, which never changes from one time to another.  Each configuration only affects, and is affected by, its immediate neighbors.  Each actually existent thing is perfectly unique, as a mathematical entity.

Newton, first to combine the Heavens and the Earth with a truly universal generalization, saw a clockwork universe of moving billiard balls and their world-lines, governed by perfect exceptionless laws. Newton was the first to look upon a greater beauty than any mere religion had ever dreamed.

But the beauty of classical physics doesn’t begin to compare to the beauty of timeless quantum physics.

Continue reading "Timeless Beauty" »

GD Star Rating
loading...

Lazy Lineup Study

By Robin Hanson · May 27, 2008 6:00 am · Comments (16)

Thursday’s Nature suggests standard police line-ups may not be so bad:

The traditional US procedure is familiar to any fan of television cop shows. Witnesses are presented with a line-up that includes both the suspect and a number of innocent people, or foils’, and are asked to identify the perpetrator. In the early 1990s … then attorney-general, Janet Reno, invited experts to form a working group to address how this method could be improved. … The working group’s most important recommendation was that line-ups should be conducted in a double-blind fashion, so that neither the witness nor the official overseeing the procedure would know who the suspect was. The group also recommended that the suspect and foils be presented sequentially rather than simultaneously, and that the witness be asked to make a decision after each one rather than waiting until the end. …

In 2003, the Illinois State Police commissioned its own study to test line-ups under real-world, field conditions …, with the cooperation of two psychologists and three of the state’s police departments … [they] spent a year conducting some 700 eyewitness identifications. Some of the procedures were non-blind and simultaneous; the rest were double-blind and sequential. Both conditions were a mix of live line-ups and photo arrays. The team found that the double-blind, sequential technique produced higher rates of foil picks – that is, clear errors – and lower rates of suspect picks than the traditional, nonblind line-up. …

Continue reading "Lazy Lineup Study" »

GD Star Rating
loading...
Tagged as: