No doubt a stupid question, but I am masterful in the asking of them. I can be the guy everyone sniggers at whenever I post - every forum should have one.
So: if I understand it correctly, whole brain emulation is one approach to building a platform for Friendly AGI. If we assume that cognitive functionalism holds true as a theory — the idea that our brains are a kind of computer — there are two very promising approaches worth pursuing. One strategy is the rules-based approach as, for example, promoted by Ben Goertzel. This involves hard-coding to figure out the algorithms of intelligence and the ways that they’re intricately intertwined - cognitive science.
The second approach seems to be focussed on reverse engineering of the human brain itself. Neuroscientists say there’s no reason to believe that we can’t model this structure ourselves. Here, whole brain emulation implies the re-creation of all the brain’s properties in an alternative substrate, namely a computer system, creating a 1-to-1 model where all relevant properties of a system exist.
My question, then, is - would the latter method require reverse engineering a single brain and if so, whose? Having bipolar disorder, the last thing you would want would be me holding the fate of the world in my hands. Hell, I can barely use my iPhone and my mood swings would, I feel, make for a very dangerous Singularity.
So, is this a dumb question? If reverse engineering the brain in order to build the platform for AGI wins out of the two approaches stated above, who exactly will the AGI ‘be’ - at least to begin with? Or is it (as I suspect) somewhat more complicated than that?
The Idee is to reverse engineer the (learning) structure of a brain and let it learn and experience on its own. So it will actually develop its own brain structure which will be similar to a human brain. Kinda like that my brain isn’t looking like yours because of our different paths in life, in gaining experience, profession etc…
Thanks for the reply, I think I’m getting there slowly. Surely though, right at the very beginning, there will need to be ‘a brain’ to reverse engineer its learning structure? My question then would still stand - who’s do we begin with? Or am I still way of course?
I think people tend to label questions as stupid prematurely. The simplest questions are often the most profound ones.
If that’s a role you really want to play, please go ahead and have fun with that.
That’s a really tough question. I think you would need some kind of “generic” human brain. Perhaps even a “blank slate” brain that is trained in various successive steps to think and behave as “generic” humans behave. Possibly, it might be possible to get a “blank slate”-like brain by “averaging” the connectomes of many different human brains. If people wanted to go down the route of the freshly learning brain, they might desire to emulate the brains of infants, because those are extremely plastic and can learn almost everything – though that requires some time, of course.
Anyway, I think we have to ask a deeper question here: For what purpose do we want to reverse engineer the brain?
If we wanted perfect worker AGIs, we would reverse engineer the brains of the best workers
If we wanted perfect soldier AGIs, we would emulate the brains of the best soldiers
In general, if we want the perfect X AGI, we would simply emulate the brain of the human who can do X best (out of those who are available for scanning and emulation).
After all, we are doing something really similar with astronauts. We don’t choose astronauts arbitrarily. They are chosen according to very strict criteria and then get really tough and thorough training. I don’t see why we wouldn’t choose a similar path when it comes to whole brain emulations.
Now who defines the criteria? This sounds very much like a question of politics – and money, of course.
I used to suffer from unipolar depression, and had my phases in which I wished the world to burn! So, I can kinda relate to your reluctance. However, I think that we will be able to “recalibrate” our brains, once we understand how they really work, or at least can emulate them to a sufficient degree of accuracy.
I had an epiphany last night. Unfortunately nothing to do with this question
So what seems to be the case - we’re not seeking to transfer ‘personality’ to the brain, merely a building block for the AGI to develop its own?
I still can’t see my way around the fact that even the best brains at X (and how do we quantify that?) could have all kinds of (hidden) imperfections that could lead to the AGI being good at X while being crap at Y.
If we could scan human brains in sufficient detail, we could transfer a human personality to an AGI. But that’s optional.
There is no reason for any object or person to be good for everything. I doubt that that would even be possible.
And yes, most people are crap at most things. It mostly doesn’t matter, because people just don’t do most things. It’s enough if people are good at thousands of different tasks, even though there are probably billions of tasks that any person could be good at.
So, I can kinda relate to your reluctance. However, I think that we will be able to “recalibrate” our brains, once we understand how they really work, or at least can emulate them to a sufficient degree of accuracy.