4 ms·

Probably too late for this, but I have argued before that language is a fundamentally lossy encoding of the human experience. We do our best to describe what we

Probably too late for this, but I have argued before that language is a fundamentally lossy encoding of the human experience. We do our best to describe what we're seeing and experiencing using language, which is fantastically expressive, but it has its limits. I think we see glimpses of this when we find ourselves saying things such as, "it's impossible to put it into words" or we overload certain words when we mean very different things, such as, "I love my children" or "I love apple pie". Clearly the word "love" here has a certain magnitude that is not being expressed, yet it is understood by the listener somehow. So, I do sort of buy into this idea that Einstein was simulating the world and running experiments on those simulations in ways that were beyond what you could encode in natural language. Will AI be capable of doing this, if it is bounded by training data that is composed almost entirely on language? One might argue that if AI is training on a lossy encoding/representation of the human experience, how will it be able to simulate anything beyond that experience? Unless it does so in a way that we manage to do when we image objects beyond 3D. But now I'm just rambling.

sandeepkd 2mo ago
I think this is very critical concept and existing gap which does not makes into AI conversations. Interestingly enough a lot of Si Fi movies have captured this where the AI starts to feel and have "thoughts". Have to give kudos to the authors for being so creative and imaginative.
gabbagool 2mo ago
Or, ask yourself the following question: If you read every single book there is about The Grand Canyon, and watched every single video and/or documentary about The Grand Canyon, do you believe that you have fully experienced The Grand Canyon? Or do you just have to be there to fully experience it. I dunno. Substitute in whatever you want for "The Grand Canyon". Maybe climbing Mount Everest or walking on the Moon. The point is that maybe the human experience is more vast than what is written about it.
- Dylan16807 2mo ago
  I think I could get very close via video. But that depends on a lot of my non-canyon experience moving around the world, and LLMs are very flawed in their ability to input video, so I think they would not get nearly as close.
  - throwaway0123_5 2mo ago
    I've been to a few natural wonders (including the Grand Canyon) that I saw in advance on video. At least for me it isn't close at all. Even if audiovisual elements could be near-perfectly reproduced by video (imo not even close with modern tech, no screen is capturing the brilliance of sunlight), you aren't capturing the temperature, the feel of wind or rain, the smell of the plants around you, etc.
    - Dylan16807 2mo ago
      But you already know those feelings from being outside elsewhere. You get a unique combination there, but that's only worth so much.
      - throwaway0123_5 2mo ago
        > only worth so much. Worth... quite a lot imo. First, in the slightly objective sense of "What is this place like in real life, under X conditions." But, more subjectively, watching a video of a glacier and imagining the wind/rain/cold doesn't even approach 5% of the intensity and awe of climbing up a mountain yourself to see a seemingly endless expanse of ice, struggling to stand steady because of the wind, shivering because of the cold and rain. And finally, in the financial sense, a lot of people routinely spend many thousands of dollars and days-weeks of their time to experience natural wonders in real life.
        Dylan16807 2mo ago
        There's so much natural wonder in the world in and around your home. It's not that the value of the grand canyon is low in an absolute sense, it's that in a relative sense the gap between "empty numb void" and "hiking at home (plus grand canyon videos)" is far far greater than the gap between "hiking at home (plus grand canyon videos)" and "actual grand canyon".
lukifer 2mo ago
The lossy compression of language is why we should find it unsurprising that LLMs tend to perform better at code, than at human language tasks or reasoning. While there can be subtle semantic differences in real codebases (using "null" to mean "unknown" in one context, versus "intentionally blank" in another), there is a much tighter coupling of semantics to meaning (low ambiguity) compared to "love" in English (let alone any inexpressible je ne sais quoi). With apologies if this is common knowledge at this point, 3Blue1Brown has been doing an excellent series on compression, and its relationship to intelligence (or more controversially, that they are one and the same): https://www.youtube.com/watch?v=l6DKRf-fAAM https://www.youtube.com/watch?v=l6DKRf-fAAM But that also throws in sharp relief, that there is vastly more to the human experience than intelligence alone: qualia, desire, gut instinct, intuition, emergent creativity. (Whether the "God of the Gaps" for the delta between capabilities of human vs AIs is fixed, or diminishing, or even shrinking to zero, remains an open experiment we're all living through.)
- cwmoore 2mo ago
  I have come across a concept that given a file of compressed text files, adding a new text to it expands it more or less depending on how different the new text is from the compressed. The compression series sounds interesting! I think the deltas are growing at different rates.
  - Hunpeter 2mo ago
    > given a file of compressed text files, adding a new text to it expands it more or less depending on how different the new text is from the compressed. The 3b1b videos mention how this concept may be used to find similarities between different languages. Researchers have been able to get results that closely resemble how languages are usually grouped into families.
- nullbio 2mo ago
  I think LLMs perform better at code than human language tasks because there's no clear way to eval human language tasks in a non-ambiguous or concrete manner. Any sort of eval that happens around language is transformation tasks, which have deterministic properties or goals. The fuzzy side of human communication can only be modeled probabilistically because there are no clear boundaries. Human communication is more like a felt mutual agreement where the correct interpretation is generally determined by popularity.
anzuhoeren 2mo ago
[dead]
zapataband1 2mo ago
bro, it is a token predictor. You had me in the first half, language is humanity's greatest invention and all of the LLM "wins" can be attributed to existing language(imo). But no it's not going to have a vision like Einstein because it doesn't have a brain it is a token predictor.
- vanviegen 2mo ago
  Yes, and a human is a procreation machine. Bro.
walrus01 2mo ago
> Probably too late for this, but I have argued before that language is a fundamentally lossy encoding of the human experience. We do our best to describe what we're seeing and experiencing using language, which is fantastically expressive, but it has its limits. Yes, and sometimes this is very intentional. Take for example a short poem which if you sit and really think about it for a long time, you could go off on a mental tangent of imagining what sort of kingdom or empire created a statue that is now "two vast and trunkless legs of stone", for instance. Being terse and allowing for human interpretation is kind of the entire point of something being written like this. I met a traveller from an antique land Who said: Two vast and trunkless legs of stone Stand in the desert. Near them, on the sand, Half sunk, a shattered visage lies, whose frown, And wrinkled lip, and sneer of cold command, Tell that its sculptor well those passions read Which yet survive, stamped on these lifeless things, The hand that mocked them and the heart that fed: And on the pedestal these words appear: "My name is Ozymandias, king of kings: Look on my works, ye Mighty, and despair!" Nothing beside remains. Round the decay Of that colossal wreck, boundless and bare The lone and level sands stretch far away.
- AnthonyMouse 2mo ago
  > Being terse and allowing for human interpretation is kind of the entire point of something being written like this. It's also to a certain extent why LLMs work. When IBM Watson was playing Jeopardy, one of the game prompts was: > It was the anatomical oddity of U.S. gymnast George Eyser, who won a gold medal on the parallel bars in 1904 The man was missing a leg and used a prosthetic. Watson's output was, "What is leg?" At first it was regarded as correct. If a human said that you could conclude that they knew the answer. But then the judges decided not to give Watson the point because its output didn't provide enough specificity to prove that it understood the context. If you ask an LLM what kinds of things taste sweet it can give you examples like cotton candy or strawberries, but it has never actually tasted anything. All it knows is that the training data contains the association between those tokens. But the human reading the output knows what strawberries are, which is what allows the output to be meaningful.
  - cwmoore 2mo ago
andai 2mo ago
> The words or the language, as they are written or spoken, do not seem to play any role in my mechanism of thought. The psychical entities which seem to serve as elements in thought are certain signs and more or less clear images which can be “voluntarily” reproduced and combined… The above-mentioned elements are, in my case, of visual and some of muscular type. Conventional words or other signs have to be sought for laboriously only in a secondary stage, when the mentioned associative play is sufficiently established and can be reproduced at will. —Albert Einstein Quoted in Using Spaced Repetition Systems to See Through a Piece of Mathematics, https://news.ycombinator.com/item?id=18895613 https://news.ycombinator.com/item?id=18895613 which describes the author's experience that if you approach a field obsessively enough, eventually you begin to understand it at a level deeper than language. If an LLM is big enough, I imagine something similar is happening.
- andai 2mo ago
  Although I imagine that requires contact with actual reality? i.e. I would expect a higher degree of such insights to come from RLVR. Then again maybe the philosophy department has a different opinion :)
kkukshtel 2mo ago
Noah Smith had an interesting idea related to this in a recent newsletter: https://www.noahpinion.blog/p/what-will-more-intelligence-actually https://www.noahpinion.blog/p/what-will-more-intelligence-ac... > Another way of saying this is that there may be laws of the universe that humans can’t understand but AI can. I call these “cloud laws” — causal regularities that can be exploited by technology, but which are too diffuse and complex for an individual human being to either intuit or communicate. Human language seems to obey cloud laws, so why not other phenomena too? Perhaps social sciences like economics, sociology, and political science obey similarly complex regularities, and AI can help us find them. Perhaps there are physical processes — plasma, or topological materials, or aerial turbulence, etc. — that obey cloud laws instead of chaos?
cezart 2mo ago
In this context llm's remind me of the anecdote of Agassiz and the fish: "A post-graduate student equipped with honours and diplomas went to Agassiz to receive the final and finishing touches. The great man offered him a small fish and told him to describe it. Post-Graduate Student: “That’s only a sun-fish” Agassiz: “I know that. Write a description of it.” After a few minutes the student returned with the description of the Ichthus Heliodiplodokus, or whatever term is used to conceal the common sunfish from vulgar knowledge, family of Heliichterinkus, etc., as found in textbooks of the subject. Agassiz again told the student to describe the fish. The student produced a four-page essay. Agassiz then told him to look at the fish. At the end of the three weeks the fish was in an advanced state of decomposition, but the student knew something about it." sourced from: https://nabeelqu.co/understanding https://nabeelqu.co/understanding
- unsupp0rted 2mo ago
  What was the student meant to describe that wasn’t covered in the essay?
  - morgoths_bane 2mo ago
    No man ever fishes the same fish twice, for he and the fish change. -Heraklitos when eating a tuna sandwich.
scotty79 2mo ago
The entirety of our understanding of reality we owe to lossy transformations. Also called modelling.
rzk 2mo ago
This might be of interest: Building AGI Using Language Models – https://bmk.sh/2020/08/17/Building-AGI-Using-Language-Models/ https://bmk.sh/2020/08/17/Building-AGI-Using-Language-Models...
arendtio 2mo ago
I don't think that the 'lossy' part is the problem. Instead, the problem is that language is only a subset of the human experience. So there are aspects that are not included when training models based on language. Making the models multi-modal helps to close that gap, but it still exists. But ultimately this is just a matter of training data. I do not say that it is easy to obtain the required data, but it is not a fundamental problem LLMs can't overcome.

sandeepkd 2mo ago

I think this is very critical concept and existing gap which does not makes into AI conversations. Interestingly enough a lot of Si Fi movies have captured this where the AI starts to feel and have "thoughts". Have to give kudos to the authors for being so creative and imaginative.

gabbagool 2mo ago

Or, ask yourself the following question: If you read every single book there is about The Grand Canyon, and watched every single video and/or documentary about The Grand Canyon, do you believe that you have fully experienced The Grand Canyon? Or do you just have to be there to fully experience it. I dunno. Substitute in whatever you want for "The Grand Canyon". Maybe climbing Mount Everest or walking on the Moon. The point is that maybe the human experience is more vast than what is written about it.

Dylan16807 2mo ago
I think I could get very close via video. But that depends on a lot of my non-canyon experience moving around the world, and LLMs are very flawed in their ability to input video, so I think they would not get nearly as close.
- throwaway0123_5 2mo ago
  I've been to a few natural wonders (including the Grand Canyon) that I saw in advance on video. At least for me it isn't close at all. Even if audiovisual elements could be near-perfectly reproduced by video (imo not even close with modern tech, no screen is capturing the brilliance of sunlight), you aren't capturing the temperature, the feel of wind or rain, the smell of the plants around you, etc.
  - Dylan16807 2mo ago
    But you already know those feelings from being outside elsewhere. You get a unique combination there, but that's only worth so much.
    - throwaway0123_5 2mo ago
      > only worth so much. Worth... quite a lot imo. First, in the slightly objective sense of "What is this place like in real life, under X conditions." But, more subjectively, watching a video of a glacier and imagining the wind/rain/cold doesn't even approach 5% of the intensity and awe of climbing up a mountain yourself to see a seemingly endless expanse of ice, struggling to stand steady because of the wind, shivering because of the cold and rain. And finally, in the financial sense, a lot of people routinely spend many thousands of dollars and days-weeks of their time to experience natural wonders in real life.
      - Dylan16807 2mo ago
        There's so much natural wonder in the world in and around your home. It's not that the value of the grand canyon is low in an absolute sense, it's that in a relative sense the gap between "empty numb void" and "hiking at home (plus grand canyon videos)" is far far greater than the gap between "hiking at home (plus grand canyon videos)" and "actual grand canyon".

lukifer 2mo ago

The lossy compression of language is why we should find it unsurprising that LLMs tend to perform better at code, than at human language tasks or reasoning. While there can be subtle semantic differences in real codebases (using "null" to mean "unknown" in one context, versus "intentionally blank" in another), there is a much tighter coupling of semantics to meaning (low ambiguity) compared to "love" in English (let alone any inexpressible je ne sais quoi). With apologies if this is common knowledge at this point, 3Blue1Brown has been doing an excellent series on compression, and its relationship to intelligence (or more controversially, that they are one and the same): https://www.youtube.com/watch?v=l6DKRf-fAAM https://www.youtube.com/watch?v=l6DKRf-fAAM But that also throws in sharp relief, that there is vastly more to the human experience than intelligence alone: qualia, desire, gut instinct, intuition, emergent creativity. (Whether the "God of the Gaps" for the delta between capabilities of human vs AIs is fixed, or diminishing, or even shrinking to zero, remains an open experiment we're all living through.)

cwmoore 2mo ago
I have come across a concept that given a file of compressed text files, adding a new text to it expands it more or less depending on how different the new text is from the compressed. The compression series sounds interesting! I think the deltas are growing at different rates.
- Hunpeter 2mo ago
  > given a file of compressed text files, adding a new text to it expands it more or less depending on how different the new text is from the compressed. The 3b1b videos mention how this concept may be used to find similarities between different languages. Researchers have been able to get results that closely resemble how languages are usually grouped into families.
nullbio 2mo ago
I think LLMs perform better at code than human language tasks because there's no clear way to eval human language tasks in a non-ambiguous or concrete manner. Any sort of eval that happens around language is transformation tasks, which have deterministic properties or goals. The fuzzy side of human communication can only be modeled probabilistically because there are no clear boundaries. Human communication is more like a felt mutual agreement where the correct interpretation is generally determined by popularity.

anzuhoeren 2mo ago

[dead]

zapataband1 2mo ago

bro, it is a token predictor. You had me in the first half, language is humanity's greatest invention and all of the LLM "wins" can be attributed to existing language(imo). But no it's not going to have a vision like Einstein because it doesn't have a brain it is a token predictor.

vanviegen 2mo ago
Yes, and a human is a procreation machine. Bro.

walrus01 2mo ago

> Probably too late for this, but I have argued before that language is a fundamentally lossy encoding of the human experience. We do our best to describe what we're seeing and experiencing using language, which is fantastically expressive, but it has its limits. Yes, and sometimes this is very intentional. Take for example a short poem which if you sit and really think about it for a long time, you could go off on a mental tangent of imagining what sort of kingdom or empire created a statue that is now "two vast and trunkless legs of stone", for instance. Being terse and allowing for human interpretation is kind of the entire point of something being written like this. I met a traveller from an antique land Who said: Two vast and trunkless legs of stone Stand in the desert. Near them, on the sand, Half sunk, a shattered visage lies, whose frown, And wrinkled lip, and sneer of cold command, Tell that its sculptor well those passions read Which yet survive, stamped on these lifeless things, The hand that mocked them and the heart that fed: And on the pedestal these words appear: "My name is Ozymandias, king of kings: Look on my works, ye Mighty, and despair!" Nothing beside remains. Round the decay Of that colossal wreck, boundless and bare The lone and level sands stretch far away.

AnthonyMouse 2mo ago
> Being terse and allowing for human interpretation is kind of the entire point of something being written like this. It's also to a certain extent why LLMs work. When IBM Watson was playing Jeopardy, one of the game prompts was: > It was the anatomical oddity of U.S. gymnast George Eyser, who won a gold medal on the parallel bars in 1904 The man was missing a leg and used a prosthetic. Watson's output was, "What is leg?" At first it was regarded as correct. If a human said that you could conclude that they knew the answer. But then the judges decided not to give Watson the point because its output didn't provide enough specificity to prove that it understood the context. If you ask an LLM what kinds of things taste sweet it can give you examples like cotton candy or strawberries, but it has never actually tasted anything. All it knows is that the training data contains the association between those tokens. But the human reading the output knows what strawberries are, which is what allows the output to be meaningful.
- cwmoore 2mo ago

andai 2mo ago

> The words or the language, as they are written or spoken, do not seem to play any role in my mechanism of thought. The psychical entities which seem to serve as elements in thought are certain signs and more or less clear images which can be “voluntarily” reproduced and combined… The above-mentioned elements are, in my case, of visual and some of muscular type. Conventional words or other signs have to be sought for laboriously only in a secondary stage, when the mentioned associative play is sufficiently established and can be reproduced at will. —Albert Einstein Quoted in Using Spaced Repetition Systems to See Through a Piece of Mathematics, https://news.ycombinator.com/item?id=18895613 https://news.ycombinator.com/item?id=18895613 which describes the author's experience that if you approach a field obsessively enough, eventually you begin to understand it at a level deeper than language. If an LLM is big enough, I imagine something similar is happening.

andai 2mo ago
Although I imagine that requires contact with actual reality? i.e. I would expect a higher degree of such insights to come from RLVR. Then again maybe the philosophy department has a different opinion :)

kkukshtel 2mo ago

Noah Smith had an interesting idea related to this in a recent newsletter: https://www.noahpinion.blog/p/what-will-more-intelligence-actually https://www.noahpinion.blog/p/what-will-more-intelligence-ac... > Another way of saying this is that there may be laws of the universe that humans can’t understand but AI can. I call these “cloud laws” — causal regularities that can be exploited by technology, but which are too diffuse and complex for an individual human being to either intuit or communicate. Human language seems to obey cloud laws, so why not other phenomena too? Perhaps social sciences like economics, sociology, and political science obey similarly complex regularities, and AI can help us find them. Perhaps there are physical processes — plasma, or topological materials, or aerial turbulence, etc. — that obey cloud laws instead of chaos?

cezart 2mo ago

In this context llm's remind me of the anecdote of Agassiz and the fish: "A post-graduate student equipped with honours and diplomas went to Agassiz to receive the final and finishing touches. The great man offered him a small fish and told him to describe it. Post-Graduate Student: “That’s only a sun-fish” Agassiz: “I know that. Write a description of it.” After a few minutes the student returned with the description of the Ichthus Heliodiplodokus, or whatever term is used to conceal the common sunfish from vulgar knowledge, family of Heliichterinkus, etc., as found in textbooks of the subject. Agassiz again told the student to describe the fish. The student produced a four-page essay. Agassiz then told him to look at the fish. At the end of the three weeks the fish was in an advanced state of decomposition, but the student knew something about it." sourced from: https://nabeelqu.co/understanding https://nabeelqu.co/understanding

unsupp0rted 2mo ago
What was the student meant to describe that wasn’t covered in the essay?
- morgoths_bane 2mo ago
  No man ever fishes the same fish twice, for he and the fish change. -Heraklitos when eating a tuna sandwich.

scotty79 2mo ago

The entirety of our understanding of reality we owe to lossy transformations. Also called modelling.

rzk 2mo ago

This might be of interest: Building AGI Using Language Models – https://bmk.sh/2020/08/17/Building-AGI-Using-Language-Models/ https://bmk.sh/2020/08/17/Building-AGI-Using-Language-Models...

arendtio 2mo ago

I don't think that the 'lossy' part is the problem. Instead, the problem is that language is only a subset of the human experience. So there are aspects that are not included when training models based on language. Making the models multi-modal helps to close that gap, but it still exists. But ultimately this is just a matter of training data. I do not say that it is easy to obtain the required data, but it is not a fundamental problem LLMs can't overcome.