A Skeptical Reaction To IBM's Cat Brain Simulation Claims

kreyszig writes "The recent story of a cat brain simulation from IBM had me wondering if this was really possible as described. Now a senior researcher in the same field has publicly denounced IBM's claims." More optimisticaly, dontmakemethink points out an "astounding article about new 'Neurogrid' computer chips which offer brain-like computing with extremely low power consumption. In a simulation of 55 million neurons on a traditional supercomputer, 320,000 watts of power was required, while a 1-million neuron Neurogrid chip array is expected to consume less than one watt."

Comments

nonlinear

[Garble Snarky]

Wouldn't power consumption grow more than linearly with neuron count? I would think the number of connections is the dominant factor - so the comparison of two data points of power consumption vs neuron count is meaningless.

Re:Does anyone really know what a cat thinks?

[marqs]

"If a lion could talk, we could not understand him."
Ludwig Witgenstein - tractatus logico-philosophicus

long ways to go yet

[Anonymous Coward]

From the original FA: "The simulation, which runs 100 times slower than an actual cat's brain, is more about watching how thoughts are formed in the brain and how the roughly 1 billion neurons and 10 trillion synapses in a cat's brain work together."

So the most bad-ass computer simulation, assuming it worked, which this guy is saying it probably didn't, was still 100 times slower than a real cat's brain. A real cat's brain also fits inside a tiny furry space the size of a baseball... and it runs on a once-daily small bowl of cat food. We have a long ways to go.

Re:long ways to go yet

[Zackbass]

Considering how little we know about the emergence of intelligence from networks how is it possible to claim outright that an ANN can't approach the capabilities of a human brain? Real neurons are vastly more complex and aren't accurately modeled with such simple systems, but we don't have any idea what those complexities have to do with intelligence, so it seems to be quite the leap of faith to make claims on the topic.

Re:long ways to go yet

[Xest]

It basically just seem to be a case of the same old AI arguments we've always heard even since Turing's days.

The problem is, we don't actually know what the limits of ANNs are, there is no proof that suggests that they can't, given ever greater amounts of computing power allow for the emergence of (at least seemingly) truly intelligent response to an event.

So on one hand we have the IBM guys overstating what they've achieved, and on the other we have a guy spouting out a view of the limits of ANNs without actually putting any effort into providing evidence for their limitations.

I don't know why but the AI field has always been horifically polarised, the kind of arguments you get in that field are just so immature it's beyond belief. You have people in the AI field following their viewpoint religiously, completely unwilling to consider the other viewpoint. To see what I mean just look up some of the discussions on Searle's chinese room argument.

If AI scientists spent as much time on research as they did bitching at each others experiments and theories we'd have a walking talking robo-jesus by now that could build worlds.

Re:long ways to go yet

[Rod Frey]

Isn't there value in moving to a higher level of abstraction than a single neuron though? Or simplifying the basic elements for the sake of a tractable broader model?

Simulating a single atom, for example, is reasonably complex: it would be impossible with current computational resources to simulate the electromagnetic properties of a metal if we required accurate simulations of individual atoms. Yet despite ignoring what we know about the atomic models, the higher-level models are very predictive.

Not that we have such predictive, higher-level models for the brain. That's what some researchers are searching for: I'm just suggesting that those models hopefully won't require accurate simulation of individual neurons. That seems to be the pattern in other domains.

Brute force neurons...

[xtracto]

So according to this guy rant letter, the "cat-brain simulation" was nothing more than the simulation of a ANN wiht X number of neurons with X equal to the average number of neurons in a cat.

However, it seems the /complexity/ of the simulated neurons is not remotely similar to that of the neurons of a real cat.

With that view, yes it seems less breakthrough. The experiment reminds me of AI researchers that thought that we could get intelligent machines using a brute-force kind of approach; this by adding /enough/ knowledge-rules, /enough/ processing power, etc...

Re:Brain Power

[Yvan256]

In a simulation of 55 million neurons on a traditional supercomputer, 320,000 watts of power was required, while a 1-million neuron Neurogrid chip array is expected to consume less than one watt.

320kW / 55 = 5.818kW per million of neuro with a traditional supercomputer.
One watt per million of neuro with a Neurogrid chip array.

So if a cat's brain is 1 BILLION neurons, that would require 5818.182kW with a supercomputer and 1kW with the Neurogrid chip array.

A reduction of 5817.182kW.