4 ms·

There was a very good paper in Nature showing this definitively: https://news.ycombinator.com/item?id=41437933 https://news.ycombinator.com/item?id=41437933 Mo

by aithrowaway1987 2y ago

There was a very good paper in Nature showing this definitively: https://news.ycombinator.com/item?id=41437933 https://news.ycombinator.com/item?id=41437933 Modern ANN architectures are not actually capable of long-term learning in the same way animals are, even stodgy old dogs that don't learn new tricks. ANNs are not a plausible model for the brain, even if they emulate certain parts of the brain (the cerebellum, but not the cortex) I will add that transformers are not capable of recursion, so it's impossible for them to realistically emulate a pigeon's brain. (you would need millions of layers that "unlink chains of thought" purely by exhaustion)

whimsicalism 2y ago
this paper is far from “showing this definitively” even if we bought this negative result as somehow “proving impossibility”, i’m not convinced plasticity is necessary for intelligence huge respect for richard sutton though
- aithrowaway1987 2y ago
  Isn't "plasticity is not necessary for intelligence" just defining intelligence downwards? It seems like you want to restrict "intelligence" to static knowledge and (apparent) short-term cleverness, but being able to make long-term observation and judgements about a changing world is a necessary component of intelligence in vertebrates. Why exclude that from consideration? More specifically: it is highly implausible that an AI system could learn to improve itself beyond human capability if it does not have long-term plasticity: how would it be able to reflect upon and extend its discoveries if it's not able to learn new things during its operation?
  - scarmig 2y ago
    Anterograde amnesia is a significant disruption of plasticity, and yet people who have it are still intelligent. (That said, I agree plasticity is key to the most powerful systems. A human race with anterograde amnesia would have long ago gone extinct.)
  - sebastiennight 2y ago
    Let's not forget that software has one significant advantage over humans: versioning. If I'm a human tasked with editing video (which is the field my startup[0] is in) and a completely new video format comes in, I need the long term plasticity to learn how to use it so I can perform my work. If a sufficiently intelligent version of our AI model is tasked with editing these videos, and a completely new video format comes in, it does not need to learn to handle it. Not if this model is smart enough to iterate a new model that can handle it. The new skills and knowledge do not need to be encoded in "the self" when you are a bunch of bytes that can build your successor out of more bytes. Or, in popular culture terms, the last 30 seconds of this Age of Ultron clip[1]. [0]: https://www.onetake.ai https://www.onetake.ai [1]: https://www.youtube.com/watch?v=qA5wYcybkCM&t=25 https://www.youtube.com/watch?v=qA5wYcybkCM&t=25
theGnuMe 2y ago
You can always convert a recursive function call to a loop.
calf 2y ago
You've read the abstract wrong. The authors argue that neural networks can learn online and a necessary condition is random information. That's the thesis, their thesis is not that neural networks are the wrong paradigm.
davedx 2y ago
> Modern ANN architectures are not actually capable of long-term learning What do you think training (and fine-tuning) does?
- lxgr 2y ago
  That's not how we (today) practically interact with LLMs, though. No LLM currently adapts to the tasks its given with an iteration cycle shorter than on the order of months (assuming your conversations serve as future training data; otherwise not at all). No current LLM can digest its "experiences", form hypotheses (at least outside of being queried), run thought experiments, then actual experiments, and then update based on the outcome. Not because it's fundamentally impossible (it might or might not be), but because we practically haven't built anything even remotely approaching that type of architecture.
slashdave 2y ago
So use a new NN architecture. I mean, that is the point, isn't it?

whimsicalism 2y ago

this paper is far from “showing this definitively” even if we bought this negative result as somehow “proving impossibility”, i’m not convinced plasticity is necessary for intelligence huge respect for richard sutton though

aithrowaway1987 2y ago
Isn't "plasticity is not necessary for intelligence" just defining intelligence downwards? It seems like you want to restrict "intelligence" to static knowledge and (apparent) short-term cleverness, but being able to make long-term observation and judgements about a changing world is a necessary component of intelligence in vertebrates. Why exclude that from consideration? More specifically: it is highly implausible that an AI system could learn to improve itself beyond human capability if it does not have long-term plasticity: how would it be able to reflect upon and extend its discoveries if it's not able to learn new things during its operation?
- scarmig 2y ago
  Anterograde amnesia is a significant disruption of plasticity, and yet people who have it are still intelligent. (That said, I agree plasticity is key to the most powerful systems. A human race with anterograde amnesia would have long ago gone extinct.)
- sebastiennight 2y ago
  Let's not forget that software has one significant advantage over humans: versioning. If I'm a human tasked with editing video (which is the field my startup[0] is in) and a completely new video format comes in, I need the long term plasticity to learn how to use it so I can perform my work. If a sufficiently intelligent version of our AI model is tasked with editing these videos, and a completely new video format comes in, it does not need to learn to handle it. Not if this model is smart enough to iterate a new model that can handle it. The new skills and knowledge do not need to be encoded in "the self" when you are a bunch of bytes that can build your successor out of more bytes. Or, in popular culture terms, the last 30 seconds of this Age of Ultron clip[1]. [0]: https://www.onetake.ai https://www.onetake.ai [1]: https://www.youtube.com/watch?v=qA5wYcybkCM&t=25 https://www.youtube.com/watch?v=qA5wYcybkCM&t=25

theGnuMe 2y ago

You can always convert a recursive function call to a loop.

calf 2y ago

You've read the abstract wrong. The authors argue that neural networks can learn online and a necessary condition is random information. That's the thesis, their thesis is not that neural networks are the wrong paradigm.

davedx 2y ago

> Modern ANN architectures are not actually capable of long-term learning What do you think training (and fine-tuning) does?

lxgr 2y ago
That's not how we (today) practically interact with LLMs, though. No LLM currently adapts to the tasks its given with an iteration cycle shorter than on the order of months (assuming your conversations serve as future training data; otherwise not at all). No current LLM can digest its "experiences", form hypotheses (at least outside of being queried), run thought experiments, then actual experiments, and then update based on the outcome. Not because it's fundamentally impossible (it might or might not be), but because we practically haven't built anything even remotely approaching that type of architecture.

slashdave 2y ago

So use a new NN architecture. I mean, that is the point, isn't it?