Biological evolution produces “messy” models. They are needlessly complex and difficult to understand. And yet they are alive!
Here are my notes on the topic of evolution and artificial life. The section “Sparsity & Modularity” discusses what I mean by “messy” models. https://coldcoffee.neocities.org/evolution_review
Forget about the model’s weights, the revolution will be published in an academic journal. The underlying principles of AGI are going to be talked about, even if the exact methods are secret. As a scientist, to discover something important and then carry it to the grave is an absurd proposition. That is not asking people for discretion or secrecy, that is asking them for their resignation and early permanent retirement.
The nuclear bomb industry is in a similar state: we all know the basic science but we dont publish schematics! This minor omission from the scientific literature only slowed down nuclear armament, but it did not prevent the cold war nuclear arms race. However, unlike bombs, AGI might actually be useful and so most people are much more motivated to persue this tech.
Sorry to get your hopes up but I was being facetious and provocative. Instead of a glass jar, our horse’s brain is going to live inside of a computer simulation. Nonetheless, I think my argument still holds true.
Neuroscientists scoff at the thought of whole brain simulation. They’re incredulous and as a result they’re unambitious. They want it but they know they can’t have it; they’ve got sour grapes. Despite these bad vibes, they have been working diligently and I think we’re not too far off from making simulations which are genuinely useful.
On a wacky side note, IMO, if we did have a horses brain in a jar, then interacting with it would be the easy part. There have been some really neat advances in how we interact with brains.
We can make neurons light up when they activate, see GCaMP
And here is a video of GCaMP in action:
We can activate synapses with light, see Optogenetics
The hard part would be keeping it alive for its 25-30 year lifespan even though it’s missing important internal organs like the heart, lungs, liver, and adaptive immune system.
Your introduction describes how I feel about my area of expertise too!
Working in the field of *neuroscience* is a bizarre experience. No one seems to be interested in the most interesting applications of their research. Neuroscience has significantly advanced in the past few decades. I keep telling people that soon we’re all going to have self driving cars powerd by a horse’s brain in a glass jar. Most people just laugh in disbelief, others make frightened noises and mention “ethical issues” or change the subject. The smart money is off chasing the deep learning pipe dream, and as a direct consequence, there is low-hanging fruit absolutely everywhere.
Great stuff!
To expand on a few points:
The brain is less complex than people make it out to be. It is complicated, but there is clearly a logic to it, which we only need to discover. Often cited statistics about the numbers of cells and synapses include the cerebellum, which isnt really necessary for intelligence.
Evolution produces some beautiful designs, but if left to its own devices it takes longer than G.R.R. Martin takes to write a novel. According to the theory of punctuated equilibria: evolution spends long periods of time stuck in a stasis, when evolution essentially gets stuck in a local optimum.
Meow Meow,
I’d like to introduce myself. My name is David and I am an AGI enthusiast. My goal is to reverse engineer the brain in order to create AGI and to this end I’ve spent years studying neuroscience. I look forward to talking with you all about neuroscience and AGI.
Now I must admit: I disagree with this community’s prevailing opinions on the topic of AI-Doom. New technology is almost always “a good thing”. I think we all daydream about AGI, but whereas your fantasies may be dark and grim, mine are bright and utopian.
I’m also optimistic about my ability to succeed. Nature has provided us with intelligent lifeforms which we can probe and disect until we understand both life and intelligence. Technology has advanced to the point where this is within our reach. Here is a blog post I wrote in suport of this point.
As a final note I’d like express my distain for deep learning. It’s not biologically plausible. It does not operate on the same basic prinicples as intelligent life. Maybe with sufficient effort you could use deep learning to create AGI, but I suspect that in doing so you’d rediscover the same principles that are behind biological intelligence.
Im curious how you think animal training works. It seems at odds with your ideas.