jump to navigation

IBM Produces Computer Chip That Acts Like A Brain August 19, 2011

Posted by Metabiological in Synthetic Intelligence.
Tags: , , , ,
add a comment

One of the more interesting aspects of the human brain is its ability to rewire itself.  Indeed the ability of our neurons to form new connections with each other is the basis of both our memory and our ability to learn new skills.  Its also something that separates us from current machine intelligence.  Well IBM may have taken a step closer to bridging that gap with the development of a microprocessor which behaves like a human brain.

In humans and animals, synaptic connections between brain cells physically connect themselves depending on our experience of the world. The process of learning is essentially the forming and strengthening of connections.

A machine cannot solder and de-solder its electrical tracks. However, it can simulate such a system by “turning up the volume” on important input signals, and paying less attention to others…

Instead of stronger and weaker links, such a system would simply remember how much “attention” to pay to each signal and alter that depending on new experiences.

Interesting certainly, but let’s keep things in perspective.  This does not signal the development of a “machine brain”.  While coming closer to the hardware is a nice step the real challenge is going to be figuring out the software (i.e. consciousness) which still presents a host of both engineering and philosophical problems.

The other issue of course is that not everyone agrees that building a machine in the likeness of the human brain is the best way to achieve synthetic intelligence.  More than a few actually think that reverse engineering the brain is precisely the wrong way to go about, being difficult, time-consuming and in their view unnecessary.  I admit to being only an interested observer in the realm of machine intelligence and therefore have to rely on what the experts in the field tell me.  That being said it strikes me as intuitive that an intelligence that is built on the same foundations as our own, with thoughts and information exchanged in a similar thought obviously not identical manner, might be easier for us to relate to and understand (and perhaps more importantly predict) than an intelligence whose very structure is totally alien to our own.

That’s merely a personal opinion but the question may be an important one.  A similar situation can be found in the realm of animal intelligence, specifically the intelligence of creatures like cephalopods whose evolution and brain structure is vastly different from our own.  Whether or not such things matter in how w being perceives and acts is an open question, one I’m sure will be answered sooner than we think.

Advertisements

Scientists Discover How Brain Recognizes Faces June 1, 2011

Posted by Metabiological in Synthetic Intelligence.
Tags: , , ,
add a comment

Interesting news coming out of the PNAS.  A group of researchers have pinpointed the areas of the brain responsible for that most human of talents; recognizing another person’s face.

“Faces are among the most compelling visual stimulation that we encounter, and recognizing faces taxes our visual perception system to the hilt. Carnegie Mellon has a longstanding history for embracing a full-system account of the brain. We have the computational tools and technology to push further into looking past one single brain region. And, that is what we did here to discover that there are multiple cortical areas working together to recognize faces.”

While this is certainly cool in and of itself and will have great implications for our understanding of the brain and conditions like prospagnosia (the inability to recognize faces) what excites me most about this is the implications it could have for the field of synthetic intelligence.

One of the paradoxes of SI research has been that tasks we perform quite easily, such as face recognition or folding the laundry, has consistently given machines difficulties.  The reason for this as far as I can tell (not being a SI person) has to do with the underlying structure of the human brain.  The brain is more or less a giant pattern recognizing device, designed through evolution to allow us to tease apart the facts that allowed us to survive in a prehistoric world.  Facts like whether or not a certain colored fruit is okay to eat or whether there was a dangerous predators hiding in the grass or, you guessed it, whether or not that person standing in front of us was someone we already know.  Machines on the other hand, for all their great speed, are still little more than calculators performing single calculations one after another.  They lack the massive parallel processing abilities of the brain because their not built like it.

There’s a fair amount of debate within the SI community as to whether or not achieving machine sentience requires the construction of an artificial brain first.  Not being a computer scientist I’m not entirely sure where I stand on that debate, though I can say that plenty of computer scientists do believe so, but sooner or later we will create a machine in the model of the human brain.  Research like this may help us get their sooner.