5 ms·

> “Humans are actually reasoning. LLMs are not.” Again, how would you measure such a thing? I would posit that reasoning is the ability to construct new, previ

> “Humans are actually reasoning. LLMs are not.” Again, how would you measure such a thing? I would posit that reasoning is the ability to construct new, previously-unexpressed information from prior information. If ChatGPT existed 110 years ago and fed all the then-known relevant experimental data regarding subatomic particles, it would not have been able to arrive at the new notion of quantum mechanics. If it existed in the medieval era and was trained on the output and influence of Galen, it would not be able to advance beyond the theory of humours to create germ theory. It's only because quantum mechanics is a known concept that has been talked about in literature that ChatGPT is able to connect that concept to other ones (physics, the biography of Niels Bohr, whatever). So the test for actual reasoning would be a test of the ability to generate new knowledge.

transcoderx 4y ago
Very few humans ever had the ability to derive Quantum Mechanics.
rvnx 4y ago
Asking: "Describe to me a completely new and credible explanation on why space-time is shrinking" It gives me: Title: Isotropic Space-Time Contraction: A Novel Hypothesis for Shrinking Space-Time Abstract: This paper introduces a new and credible explanation for the phenomenon of shrinking space-time, which we call "Isotropic Space-Time Contraction" (ISTC). ISTC postulates that space-time contracts uniformly in all directions due to the continuous creation of dark energy in the quantum vacuum. This process results from the interaction between dark energy and the cosmic fabric, leading to a constant reduction in the scale of space-time. I think it can create very very very interesting ideas or concepts.
- nikodotio 4y ago
  a quick google reveals that all of the words in the "new" title already exist from human producers, and this is mix and matched together.
  - scrollaway 4y ago
    A quick google reveals the same of your comment. Are you an AI?
    - freehorse 4y ago
      The comment answers to how GPT supposedly produces "new interesting concepts and ideas". The comment you are answering is not supposed or perceived to present anything novel. So both your argument and the comment you reply to are technically correct, neither produces any novel idea; only difference is your reply is out of place.
      - maxdoop 4y ago
        What exactly is a novel idea? Is there any true “novelty” that just suddenly arises in a human brain without having prerequisite experience / knowledge?
        freehorse 4y ago
        I mean, indeed most "novel ideas" are about making novel connections between existing concepts, and more rarely about creating novel concepts. However, bunching together concepts, even if in a novel way, does not necessarily constitute "novel idea" as the connections have to be in some ways meaningful and interesting, in a certain context. In the previous case, which is actually quite typical of LLMs, a "novel idea" was produced by bunching together certain terms, in a way that it looks syntactically correct and even conceptually solid in a superficial manner. But from a scientific point of view, it is non-sensical and, basically, junk. Bunching together concepts as words does not make a scientific idea or theory. This is a complex process in science, that involves different scales of argumentation, none of which is provided by LLMs. But this is as far as merely manipulating language can get one.
      - rvnx 4y ago
        I'm pretty sure space-time expands according to scientific literature, doesn't it ? (this is why I picked this weird concept of a shrinking space-time)
- tyfon 4y ago
  To me it actually looks like it just took the dark energy theory and s/expanding/shrinking/g on it. It's even called "isotropic space-time expansion" in many papers.
maxdoop 4y ago
I think there’s a huge misunderstanding of what “original ideas” are and what constitutes novelty. Mark Twain quote on originality: “ There is no such thing as a new idea. It is impossible. We simply take a lot of old ideas and put them into a sort of mental kaleidoscope. We give them a turn and they make new and curious combinations. We keep on turning and making new combinations indefinitely; but they are the same old pieces of colored glass that have been in use through all the ages.” I am not sure how humans “come up with new ideas” themselves. It does seem to be that creativity is simply combining information in new ways.
- dunefox 4y ago
  I don't think I would take Mark Twains word on such topics.
scrollaway 4y ago
If you’ve pair programmed with GPT you can absolutely tell it has the ability to reason. I’m sure you could get it to reason about new physics. You’re underestimating how much work went into discovering these new concepts; it’s not just a dude having a eureka moment and writing down an equation.
killerstorm 4y ago
> I would posit that reasoning is the ability to construct new, previously-unexpressed information from prior information. We should test it on a small scale, with synthetic examples. Not "invent Quantum Mechanics please". And yes, people already tested it on reasonable-sized examples, and it does work, indeed. E.g. ability to do programming indicates that. Unless you believe that all programming is just rehash of what was before, it is sufficient. Examples in the "Sparks of AGI" paper demonstrate ability to construct new, previously-unexpressed information from prior information. "It's not intelligent unless it is as smart as our top minds" is not useful. When it reaches that level you with your questions will be completely irrelevant. So you gotta come up with "as intelligent as a typical human", not "as intelligent as Einstein" criterion.
dudeinhawaii 4y ago
I was having a discussion with a colleague about how all knowledge that is "new" is necessarily derived from previous knowledge and a chance interaction with either unexpected consequences or unexpected ideas. I don't think our brains aren't magical devices that can "new up" concepts into existence that hadn't existed in some manner in which we could iterate on. Of course, there's no way to prove this at the moment. Would Einstein have invented relativity if instead he had become an art student and worked at a Bakery?

transcoderx 4y ago

Very few humans ever had the ability to derive Quantum Mechanics.

rvnx 4y ago

Asking: "Describe to me a completely new and credible explanation on why space-time is shrinking" It gives me: Title: Isotropic Space-Time Contraction: A Novel Hypothesis for Shrinking Space-Time Abstract: This paper introduces a new and credible explanation for the phenomenon of shrinking space-time, which we call "Isotropic Space-Time Contraction" (ISTC). ISTC postulates that space-time contracts uniformly in all directions due to the continuous creation of dark energy in the quantum vacuum. This process results from the interaction between dark energy and the cosmic fabric, leading to a constant reduction in the scale of space-time. I think it can create very very very interesting ideas or concepts.

nikodotio 4y ago
a quick google reveals that all of the words in the "new" title already exist from human producers, and this is mix and matched together.
- scrollaway 4y ago
  A quick google reveals the same of your comment. Are you an AI?
  - freehorse 4y ago
    The comment answers to how GPT supposedly produces "new interesting concepts and ideas". The comment you are answering is not supposed or perceived to present anything novel. So both your argument and the comment you reply to are technically correct, neither produces any novel idea; only difference is your reply is out of place.
    - maxdoop 4y ago
      What exactly is a novel idea? Is there any true “novelty” that just suddenly arises in a human brain without having prerequisite experience / knowledge?
      - freehorse 4y ago
        I mean, indeed most "novel ideas" are about making novel connections between existing concepts, and more rarely about creating novel concepts. However, bunching together concepts, even if in a novel way, does not necessarily constitute "novel idea" as the connections have to be in some ways meaningful and interesting, in a certain context. In the previous case, which is actually quite typical of LLMs, a "novel idea" was produced by bunching together certain terms, in a way that it looks syntactically correct and even conceptually solid in a superficial manner. But from a scientific point of view, it is non-sensical and, basically, junk. Bunching together concepts as words does not make a scientific idea or theory. This is a complex process in science, that involves different scales of argumentation, none of which is provided by LLMs. But this is as far as merely manipulating language can get one.
    - rvnx 4y ago
      I'm pretty sure space-time expands according to scientific literature, doesn't it ? (this is why I picked this weird concept of a shrinking space-time)
tyfon 4y ago
To me it actually looks like it just took the dark energy theory and s/expanding/shrinking/g on it. It's even called "isotropic space-time expansion" in many papers.

maxdoop 4y ago

I think there’s a huge misunderstanding of what “original ideas” are and what constitutes novelty. Mark Twain quote on originality: “ There is no such thing as a new idea. It is impossible. We simply take a lot of old ideas and put them into a sort of mental kaleidoscope. We give them a turn and they make new and curious combinations. We keep on turning and making new combinations indefinitely; but they are the same old pieces of colored glass that have been in use through all the ages.” I am not sure how humans “come up with new ideas” themselves. It does seem to be that creativity is simply combining information in new ways.

dunefox 4y ago
I don't think I would take Mark Twains word on such topics.

scrollaway 4y ago

If you’ve pair programmed with GPT you can absolutely tell it has the ability to reason. I’m sure you could get it to reason about new physics. You’re underestimating how much work went into discovering these new concepts; it’s not just a dude having a eureka moment and writing down an equation.

killerstorm 4y ago

> I would posit that reasoning is the ability to construct new, previously-unexpressed information from prior information. We should test it on a small scale, with synthetic examples. Not "invent Quantum Mechanics please". And yes, people already tested it on reasonable-sized examples, and it does work, indeed. E.g. ability to do programming indicates that. Unless you believe that all programming is just rehash of what was before, it is sufficient. Examples in the "Sparks of AGI" paper demonstrate ability to construct new, previously-unexpressed information from prior information. "It's not intelligent unless it is as smart as our top minds" is not useful. When it reaches that level you with your questions will be completely irrelevant. So you gotta come up with "as intelligent as a typical human", not "as intelligent as Einstein" criterion.

dudeinhawaii 4y ago

I was having a discussion with a colleague about how all knowledge that is "new" is necessarily derived from previous knowledge and a chance interaction with either unexpected consequences or unexpected ideas. I don't think our brains aren't magical devices that can "new up" concepts into existence that hadn't existed in some manner in which we could iterate on. Of course, there's no way to prove this at the moment. Would Einstein have invented relativity if instead he had become an art student and worked at a Bakery?