The issue is not a moral one concerning ai in research, the issue is that I simply do not believe current machine learning technology is capable of creating a functioning full brain simulator which can accurately model the interaction of the entire cns and its interaction with theoretical drugs.
That simply sounds like Theranos to me. AIVCs are at a “That might be possible in future” stage, and this proposes to model billions of them and their incredibly complex interactions.
Well, that might be possible in the future never is if no work is done to make it so,and if we learn things we can use now then the research is still valuable. I think what they’re doing is neat and would absolutely want my brain used like this once I no longer need it.
The problem with your argument here is that these people are not working on improving machine intelligence. They’re working on a model that requires the tech to be more advanced than it is (At least as far as I understand it).
Well, that might be possible in the future never is if no work is done to make it so
you’re smuggling in an assumption that this work done today will be of value to the future people, which is not automatically the case. we already have examples of that with experiments prior to 1950 that were already not up to the standards of data collection by the 1980s.
we’re really bad at brain stuff in particular and i think it’s more likely the more capable future people will, at best, have to re-do all of this work from scratch.
graduate students spent decades taking photos of thin slices of zebra fish brains only to have computerized scanning technology do all that work faster and better. it’s perhaps not fair to expect someone in the 1970s to have predicted that technological leap but i think we know enough now to say that the specific area you’re talking about is relatively likely to be obsoleted by other advancements in neuroscience
The issue is not a moral one concerning ai in research, the issue is that I simply do not believe current machine learning technology is capable of creating a functioning full brain simulator which can accurately model the interaction of the entire cns and its interaction with theoretical drugs.
That simply sounds like Theranos to me. AIVCs are at a “That might be possible in future” stage, and this proposes to model billions of them and their incredibly complex interactions.
Well, that might be possible in the future never is if no work is done to make it so,and if we learn things we can use now then the research is still valuable. I think what they’re doing is neat and would absolutely want my brain used like this once I no longer need it.
The problem with your argument here is that these people are not working on improving machine intelligence. They’re working on a model that requires the tech to be more advanced than it is (At least as far as I understand it).
It smells like Theranos to me, I’m sorry.
you’re smuggling in an assumption that this work done today will be of value to the future people, which is not automatically the case. we already have examples of that with experiments prior to 1950 that were already not up to the standards of data collection by the 1980s.
we’re really bad at brain stuff in particular and i think it’s more likely the more capable future people will, at best, have to re-do all of this work from scratch.
Do nothing! Wait for the future people! Horseshit
graduate students spent decades taking photos of thin slices of zebra fish brains only to have computerized scanning technology do all that work faster and better. it’s perhaps not fair to expect someone in the 1970s to have predicted that technological leap but i think we know enough now to say that the specific area you’re talking about is relatively likely to be obsoleted by other advancements in neuroscience