3 ms·
Whole Brain Emulation: No Progress On C. Elgans After 10 Years
- asplake 5y agoVery interesting: > The best we can do is modeling a generic worm - pretraining and running the neural network with fixed weights. Thus, no worm is "uploaded" because we can't read the weights, and these simulations are far from realistic because they are not capable of learning. Hence, it's merely a boring artificial neural network, not a brain emulation.
- MuffinFlavored 5y agois Tesla’s self driving stuff also just a boring neural network?
- dunefox 5y agoFundamentally, yes.
- ctchocula 5y agoTypo in the title. It should be "Whole Brain Emulation: No Progress On C. Elegans", not "C. Elgans".
- codewiz 5y agoThe OpenWorm "playable" demo: https://openworm.org/ https://openworm.org/
- dunefox 5y agoI wonder if there's a way for me to go from computational linguistics more towards neuroscience. I'm pretty disillusioned with the neural network approach to ai/nlp/. It seems to become more of a engineering tool than intelligence
- nynx 5y agoI think this is because there’s an obvious series of steps that the openworm team was unable or unwilling to do: teach a C. Elegans a habitual memory, preserve it using ASC, scan it slice by slice using an electron microscope, try to predict neuron weights based on chemical content or synapse size, and then feed that back into their model. One of the comments on the lesswrong post made a good point: a team without access to significant money or expertise was unable to make progress. That doesn’t say too much about the actual problem.