October 18, 2009

TED Talks: Henry Markram builds a brain in a supercomputer


This is another remarkable TED talk -- fascinating, incredibly informative and not without controversy. I'm overjoyed to hear an expert from IBM put forth a theory of mind that tries to address the problem of how the brain projects a representation of the universe to a subjective observer. I'm fairly convinced that his framing of the issue will yield some positive results.

Henry Markram says the mysteries of the mind can be solved in fairly short order. He argues that mental illness, memory and perception are all made of neurons and electric signals -- and he plans to find them with a supercomputer that models all the brain's 100,000,000,000,000 synapses.

5 comments:

Nato said...

I just hope that his final comment reflects the awareness that, past some point, the success of their project would yield a person, rather than (merely) a technical achievement.

Russell Blackford said...

Nato has a point here. I'm not usually one to point out moral pitfalls, but this seems to be a situation where some worrying ones exist.

If you design a superintelligent computer capable of solving all sorts of problems, but without any actual capacity for suffering, self-consciouness, yearning, and so on, you still have a tool. If you give it those capacities, what do you then have? A slave?

How does he propose to get around this?

proto Post said...

While the successful completion of this project will be a major achievement, it will only be the beginning. Consider the fact that we've mapped the human genome but we haven't yet cured all genetic disorders.

With regards to the person debate, I'm not sure if this project will ever create a person as we know it. It's designed to test inputs, outputs, and processes, but it's not designed to grow and develop a human being. This is the equivalent of taking a chunk out of a person's brain and running tests on it, except that you don't have to lobotomize someone to do it.

Nato said...

I have no issues with testing subunits of a mind or whathaveyou, though I would question whether one could really generate (again, past some point) the inputs necessary to produce telltale outputs without those inputs being themselves generated by other parts of the "simulated" mind. Sure, a motor cortex and a visual system do not a person make, but I'm a bit skeptical that such investigations will really yield the answers sought with little islands of quasi-cognition.

Russell Blackford said...

What Nato said.

Now, maybe this research can be carried on without overstepping any ethical lines. But I'd just like to be sure that everyone knows that they're there.

I tend to be on the team that thinks it's possible, in principle, to create a genuinely sentient and conscious intelligence running on computer hardware (although it will be very difficult unless we use the method of exact emulation of the connectivity of a real brain). Anyway, "possible" does not equal "morally desirable". We need to tread warily in this area.