10-11-2023, 09:11 AM
I think it's pretty funny that software engineers and lawyers are going to be some of the first
people greatly affected by the implementation of commercial applications of Large Language
Model AI's. There was some inexplicable magic that happened when they took very large neural
networks and simply, it seemed, had them get very good at predicting the next word. There was
something in the complex neural networks with zillions of weights that seemed to produce rational
thinking when it couldn't possibly have done so. It is yet to be explained and it's almost impossible
to analyze at the moment because the networks are so immense. There's some inherent memory
in the current models, but not that much of it. They're introducing significantly more memory,
and when they do that effectively, there's probably going to be further magic. Probably... There
may be some limit to what can be done using this model, but that limit seems to be very far away
at the moment and may not exist until well past the level of self-awareness. But who knows. It is,
for the moment, unexplainable magic; we have no idea what we're doing. But since it's profitable
for large corporations and nation states like the United States and China think it's essential to
maintaining their power, there is no stopping it. So, for better or worse, we're going to see.
Robots, especially their manual dexterity, are not very advanced and it's going to take a considerable
while to get to the level of humans. So, haha, it was the brains that got automated first. Not what
anyone expected. This means dog groomers, electronic technicians who repair tiny little things
(my long-time occupation), plumbers, and any occupation requiring lots of manual dexterity are
going to have a bit of job security for the foreseeable future. Haha, lawyers, maybe there is karma.
people greatly affected by the implementation of commercial applications of Large Language
Model AI's. There was some inexplicable magic that happened when they took very large neural
networks and simply, it seemed, had them get very good at predicting the next word. There was
something in the complex neural networks with zillions of weights that seemed to produce rational
thinking when it couldn't possibly have done so. It is yet to be explained and it's almost impossible
to analyze at the moment because the networks are so immense. There's some inherent memory
in the current models, but not that much of it. They're introducing significantly more memory,
and when they do that effectively, there's probably going to be further magic. Probably... There
may be some limit to what can be done using this model, but that limit seems to be very far away
at the moment and may not exist until well past the level of self-awareness. But who knows. It is,
for the moment, unexplainable magic; we have no idea what we're doing. But since it's profitable
for large corporations and nation states like the United States and China think it's essential to
maintaining their power, there is no stopping it. So, for better or worse, we're going to see.
Robots, especially their manual dexterity, are not very advanced and it's going to take a considerable
while to get to the level of humans. So, haha, it was the brains that got automated first. Not what
anyone expected. This means dog groomers, electronic technicians who repair tiny little things
(my long-time occupation), plumbers, and any occupation requiring lots of manual dexterity are
going to have a bit of job security for the foreseeable future. Haha, lawyers, maybe there is karma.
i used to know a lotta stuff, but i still have eight cats
