Rocketforge

This is a panel on the Prospects for AI. Panel members are Neil Jacobstein, Patrick Lincoln, Peter Norvig, and Bruno Olshausen. First up is Neil Jacobstein talking about the current state of AI and what’s real and what isn’t. In general the terms have changed though: In other words, if it works, then it isn’t “AI”. AI has become a real aspect of IT in a much more incremental and ’stealth’ way rather than the old hyped announcements of super-brain computers that can out think humans.

Now he’s reviewing the knowledge, systems, business and cultural issues still to be solved. To many to blog well, though. He’s covering stuff from the Semantic Web to nanotech and reverse brain engineering. So this is really a review instead of something really new.

Now Patrick Lincoln is reviewing why you need AI and a review of the various flavors of AI in increasing difficulty (i.e. SAT, Baysian filters, Byzantine fault tolerant systems, etc). “It is more interesting to explore ALL of the behaviors of an abstract system than SOME of the behaviors of a complete system”.

“What is the worlds most urgent and important problem?” Collectively getting better at solving urgent and important problems….

Next up is Peter Norvig from Google: his first slide is “AI in the middle”. AI as a mediating force between people. His point is that humans have actually done most of the work building the knowledge base that an AI needs to infer over, its just a matter of making it accessible and searchable (ala Google). A good example is statistical machine translation that uses human translations to build an AI engine that can take over once it has learned how to translate.

Next up is Bruno Olshausen on brain modelling and applying that to real world AI. He is starting a company called Numenta to commercialize it. Title slide is “Neuroscience and Future Prospects for Intelligent Systems”. He makes the point that much of what they’re learning isn’t from human brains. Jumping spiders for example don’t have compound eyes and have a very advanced visual system that runs on 30,000 neurons. The current state of neuroscience is still limited to the point where they still don’t know what kind of computing device a neuron actually is. Theoretical neuroscience is a combination of experimental psychology, neurobiology and math/computer science.

Its clear from all of the speakers that the old “top down” model of building AI systems has lost the fight. Especially since its gotten rather obvious that building large complex systems “top down” generally doesn’t work. Its certainly not the way nature has learned to do it.

Up next is Ray Kurzweil: Re: hard vs soft takeoff or “how to protect ourselves from pathologically strong AI” is probably one of the most important questions.

If you accept the power law aspects of all of the metrics of technological development then you probably have a very radical view of what the future may look like. But people in general have a very linear view of it as its happening. or “if you take any exponential curve and just look at a very small segment it looks like a straight line”.

He came to all of this trying to determine which startups would succeed and which ones wouldn’t. The models they creates have actually held up but all predict a period at which the model falls apart. I.e. all the Excell cells show “#######”.

Right now he’s showing several very historical models of technology development and the exponential nature of evolution and change. or “when you stand on the shoulders of giants you get really good at climbing giants”.

This slide is the 6 epochs of evolution: chemical, biological, Brains, Technology, Merger of technology and intelligence, The Universe Wakes Up.

This slide shows the same stacked S curves that Vernor showed. its not that a single technology is exponential, but the subsequence generations of technologies are.

(Hmm….. we’re spending a lot of time on this stuff which should be common knowledge to this crowd.)

The point Ray is getting to is that the models suggest that the beginning of the phase shift will probably start in the next 20 years.

I’m at Accelerating Change 2005 and will be live blogging as much of it as I can. Much of the conference is made up of ad hoc BOF groups so I’m not sure if I can capture much beyond what I’m directly involved in. The gist of the conference is how the rate of technological change is accelerating and how that rate itself will be accelerating due to specific force multipliers such as AI, nanotech, and life extension. More to come when the sessions startup:

John Smart, Founder and President of Acceleration Studies Foundation, is giving the introduction to the conference and a status of where the Foundation is at and where its going.

(grrr… my camera’s flash has stopped working so I may not have pictures)

First up is Vernor Vinge:

“Exponential growth” is one of the simplest and most pervasive patterns in nature. Some of these patterns are “exponential growth with catastrophic collapse” and “Exponential growth with saturation”. Particular technologies have “exponential growth with saturation” but when considered in aggregate the base technology curve stays exponential.

If this continues for a few more decades we get to the “killer app” of exponential improvement in computational ability: The development of creativity and intellect that surpass present-day humans.
The question of the future isn’t can we re-create human intellect, we can already do that in the form of new humans. The real question is when we create something that goes beyond it. This is a good point to declare a Singularity.

Vinge defines singularity as “a place where some regularity property is lost”. Or analogous to the physics terms connotation of “a place where the rules profoundly change”. A working definition is that while you could bring Mark Twain to today’s time and pretty much explain the state of the world in an afternoon, you could not do the same thing with a goldfish. Its also been described as technical progress that is incomprehensibly complex and rapid.

What would it be like if the Singularity didn’t happen?

Maybe Murphy’s Law trumps Moore’s Law: “The maximum possible effectiveness of a software system increases in direct proportion to the log of hte effectiveness (i.e. speed, bandwidth, memory .

Or maybe catastrophe intervenes. The more we learn about the cosmos, the more we learn how dangerous it is.

Vernor’s conclusion: while the Technological Singularity is not at all a sure thing, it is the most likely non-catostrophic scenario on the horizon. Of course the Singularity itself could be catastrophic. What can we do to make the bad versions less likely?

What if: AI succeeds
What if: The internet itself attains unquestioned life and intelligence [ed: IMHO, not likely]
What if: Fine grained distributed systems are aggressively successful?
(or what happens when all of the embedded systems network and ‘wake up’)
What if: Intelligence Amplification occurs: as the radical endpoint of human/computer interface or as the outcome of bioscience research?

Soft takeoffs vs hard takeoffs:

Soft takeoff – the transition takes years, perhaps even with the exact begininning and end states a matter of debate
Hard takeoff: the transition takes place in a very short period of time, perhaps less than 100 hours and without obvious precursors. Hard takeoff is generally considered a bad thing due to its catastrophic nature. Vinge thinks that the best way to plan for a hard takeoff is by using Intelligence Amplification in order to ensure that humans can adapt with the phase shift instead of being subject to it. This even brings up the point that if you are amplifying your human intelligence at the same time as the rest of the technological rate change is happening you actually don’t see a Singuarity happen. You ride it.

Questions: with our limited intelligence compared to things post Singularity, can we even tell if things start being much smarter than we are. Things such as corporations made up of AIs, etc….

Question: how can I make money on the takeoff: from the back of the room: “I’ll sell you hard takeoff insurance“. Vinge: if its a soft takeoff then everyone will get rich off of it. You may be able to get rich by simply sitting on your couch.

Question: will the biology end up trumping the silicon for what that super intelligence post singularity state may be based on.