15 ms·

I strongly suspect the first AGI will come sooner than expected on the back of a "glue" AI that can intelligently bond together a web of narrow AIs and utilitie

I strongly suspect the first AGI will come sooner than expected on the back of a "glue" AI that can intelligently bond together a web of narrow AIs and utilities. I got access to the wolfram plugin for chatGPT, and it turned it from a math dummy to a math genius overnight. A small step for sure, but a hint of what's to come.

_yb2s 4y ago
I agree, I suspect AGI is possible right now with a similar system only slightly more sophisticated than this one. The right "glue" for existing models, and plugins to existing data sources all coordinated in a system. GTP-4 would do the managing, and handling, and some simple template API and handler script would allow it to call instances of itself or other models, track recursion depth, and automatically remind GTP-4 to stay on track and always use the correct 'templates' for requests. It could also remind it to create a condensed summary of it's goals and state that gets repeated back automatically to act as long term memory, and increase the effective context window size (edit: this is single variable external memory). I am afraid to explain this, because I have tried a preliminary version of it that I supervised step by step, and it seems to work. I think it is obvious enough that I won't have been the only one to think of it, so it would be safer to put the information out there so people can prepare. I see a big disconnect on here between people saying GPT-4 can't do things like this and is just a "stochastic parrot" or "glorified autocomplete," and people posting logs and summaries of it solving unexpectedly hard problems outside of any conceivable training set. My theory is that this disconnect is due to three major factors: * People confusing GPT-4 and GPT-3, as both are called "chatGTP" and most people haven't actually used GPT-4 because it requires a paid subscription, and don't realize how much better it is * Most popular conceptual explanations about how these models work imply that these actually observed capabilities should be fundamentally impossible * Expectations from movies, etc. about what AGI will be like, e.g. that it will never get confused or make mistakes, or that it won't have major shortcomings in specific areas. In practice this doesn't seem to limit it because it recognizes and fixes its mistakes automatically when it sees feedback (e.g. in programming)
- basch 4y ago
  It also, in my pay opinion, needs a memory store. I need to be able to save something to a variable and call back that exact variable. Right now, because it’s just pure text input of the whole conversation, it can forget or corrupt it’s “memory”. Even something like writing a movie script doesn’t work if it constantly forgets characters names or settings or plot points.
  - _yb2s 4y ago
    My description above includes a memory store. It is a single variable that it can save and recover, and will automatically get sent back to it at regular intervals if it forgets it saved anything. Only a single text string variable is needed, as GTP-4 is smart enough to realize it can compress, pickle, etc. as many things as it needs into that.
    - alchemist1e9 4y ago
      I think we might be working something a bit similar and I agree memory is an issue. However it’s not really AGI that will emerge from this. It’s looking very useful but it’s still observable but it’s not a “mind” like an AGI. To add to your comments, I suggest that local vector store of embedding vectors of local content is how I’m going with memory issue. Langchain like. That way all previous progress is tracked and retrievable. That way the system can retrieve all it’s “memories” itself. The recursive multi agent pattern is big deal in my opinion.
      - nomel 4y ago
        > but it’s not a “mind” like an AGI. Please define this. ;)
        WA 4y ago
        Please define AGI first ;)
roflyear 4y ago
Except, whatever model is doing the orchestration will still not "understand" the problem (be able to learn from new information provided).
- _yb2s 4y ago
  GPT-4 can do this. I realize that a description of how these models work implies that this is absolutely impossible, so I have no explanation other than suggesting you go try it out for yourself.
  - antibasilisk 4y ago
    There is no description of how these models work, other than mechanically.
    - _yb2s 4y ago
      exactly... some people are using such descriptions to think about its limitations, but there is no reason to think they are accurate
  - roflyear 4y ago
    Bro, GPT-4 can absolutely not understand new information. Try to teach it a grammar. It can't learn a grammar. Anyone can learn a new grammar. Not GPT-4. GPT-4 does not ask clarifying questions. It does not understand things.
    - _yb2s 4y ago
      I'm not sure why you think that? When I give it programming tasks, GPT-4 will ask all kinds of clarifying questions. Here is a trivial example: p: I want you to solve for x, giving me a numeric value, given the formula ax^2 + bx + c = 0. r: To find the numeric value of x for the quadratic equation ax^2 + bx + c = 0, we need to use the quadratic formula: x = (-b ± √(b^2 - 4ac)) / (2a) However, you haven't provided specific values for a, b, and c. Please provide these values so I can calculate the solution(s) for x. Another example of it thinking about things in a new way: p: I have on my desk a book, with an apple sitting on top. If I were to suddenly, rapidly remove the book such that the apple remained in place, in which direction would it eventually move? r: If you suddenly and rapidly remove the book from beneath the apple, the apple would initially stay in place due to its inertia. However, once the book is removed, there would be no support underneath the apple. As a result, the apple would eventually move downwards due to the force of gravity acting on it. p: Because the earth is a sphere, I prefer to use the terms “in” and “out” to refer to movement relating to that. Can you please rephrase your answer in the terms? r: Certainly! If you suddenly and rapidly remove the book from beneath the apple, the apple would initially stay in place due to its inertia. However, once the book is removed, there would be no support underneath the apple. As a result, the apple would eventually move "in" towards the center of the Earth, due to the force of gravity acting on it. (Note I didn't tell it which direction was in vs out. This is the example problem Chomsky gave in his recent article arguing that GPT-3 cannot think - per his ideas no language model would ever be able to answer a question like this. I went a lot deeper with this and got it to add fluid dynamics calculations for wind, etc.)
dvt 4y ago
> it turned it from a math dummy to a math genius overnight Imo your post fundamentally misunderstands a few things, but mainly how Wolfram works. Wolfram can be seen as a "database" that stores a lot of human mathematical information (along with related algorithms). Wolfram does not make new math. A corollary here is that AGI needs to have the ability to create new math to be truly AGI. But unless fed something like, e.g. the Principia, I don't think we could ever get a stochastic LLM to ever derive 1+1=2 from first principles (unless specifically trained to do so). Keep in mind that proving 1+1=2 from first principles isn't even new math (new math would be proving the Poincaré conjecture before Perelman did it, for example).
- Sivart13 4y ago
  If the litmus test for intelligence is the ability to “create new math” most people on earth wouldn’t be considered intelligent
  - dvt 4y ago
    > If the litmus test for intelligence Moving goal posts around is unhelpful. I think my comment was pretty clear in the context of AGI and calling ChatGPT a "math genius."
    - famouswaffles 4y ago
      The fact that people take agi to mean "can invent new math" is a goalpost shift on its own. That was not the original meaning (generally intelligent), that wasn't even the next moved goal (on par with human experts). I guess the next one counts (better than all human experts) but i'm sure we'll move that again too.
      - akiselev 4y ago
        The problem was always underspecified. We don’t have a quantifiable metric for intelligence except IQ tests, benchmark datasets like those used to evaluate different LLMs, and other similarly myopic bullshit. “We’ll know it when we see it” becomes the default. In reality, the goal posts aren’t being moved, we’re just finding out how much further we are from them than we thought. ChatGPT is a "stochastic parrot" that’s seems way “smarter” than anyone thought possible so we have to reevaluate what we consider evidence of intelligence, perhaps coming to terms with the fact that we aren’t that smart the most of the time.
Manjuuu 4y ago
There will be no AGI during our lifetime. It seems weird to me that people that should understand how ML works (e.g. sama) instead of educating laypeople about about how this actually works and the huge limits the technology has, start talking nonsense about AGIs like some random scifi fan at a convention. Depressing.
- AbrahamParangi 4y ago
  I'll take the other side of that bet
  - brotchie 4y ago
    +1
  - Manjuuu 4y ago
    Feel free to do so, we need something to replace crypto after all. And also, sometime I wonder how many people hold those kind of opinions on AGI(imminent, doable, worthy of being discussed) because they sincerely believe in some nonsense like the basilisk thing for example.
    - bheadmaster 4y ago
      I don't believe in Basilisk because of the time-travel bullshit, but I do believe that AGI will come soon, because I don't believe in divine soul and see human mind as just a very complex statistics machine and all human learning as just detecting correlation between sensory input through time.
    - HDThoreaun 4y ago
      I believe in AGI because I'm a materialist so see no reason why we couldn't create artificial brains.
  - arcastroe 4y ago
    Heh, @Manjuuu winning this bet would imply both of your deaths ...
antibasilisk 4y ago
Whose to say that AGI isn't already here but relatively slow? It does seem to me that as more and more outputs of AI are pushed through various systems, that will eventually feed back into various AI. In 2017 an Anon on 4chan proposed that our society, due to content marketing and AI stock trading, is already run by a sort of emergent intelligence, and he proposed that this is why things seem to be 'getting crazy' lately. I'm inclined to agree with that perspective, and if correct more capable AI systems could really kick this into high gear.
- flangola7 4y ago
  There's arguably three kids of superintelligence: quality, speed, and collective. We definitely have #3 and some instances of #2. You can outthink a human by thinking 1000x faster, or by having 1000 minds think on the same problem. There's a reason everything explodes when communication technology gets better. Imagine how long it would take to coordinate the fundamental pieces of a laptop computer by pigeon mail or horseback.
Mizza 4y ago
I've built a service for building systems like this. I haven't officially "launched" yet, but it's working and you can play with it here, if anybody is up to giving the alpha a spin: https://aistudio.fly.dev/ https://aistudio.fly.dev/
- nomel 4y ago
  I really hope this is the future of coding.
- fnimick 4y ago
  It's oddly fitting that your example workflow appears to be scraping content from elsewhere, rewording it, and generating an image in order to produce automated blogspam.
  - sebzim4500 4y ago
    He knows his audience maybe too well.
- 93po 4y ago
  other quick feedback: add a "add node" button, it isn't immediately clear you do this with right clicking
- 93po 4y ago
  your "go pro" page is white text on white background in firefox
MichaelRazum 4y ago
"math genius" not sure about it. I gave it a problem from IMO 2022. In my opinion not a very hard one. It failed even if you give it some hints. Let R+ denote the set of positive real numbers. Find all functions f : R+ → R+ such that for each x ∈ R+, there is exactly one y ∈ R+ satisfying xf(y) + yf(x) <= 2. It just struggeld to reason. So I would be very surprised if the plugin somehow helps here.
- Donald 4y ago
  It's a language completion model, so it's not surprising at all that it struggles with logical inference. That's probably one of the reasons that instruction fine-tuning has such a dramatic effect in the performance of these models: they're finally given some of the underlying causal priors of the task domain.
- knrdev 4y ago
  Very strange. I gave this problem to GPT-3.5 and it answered `f(x)=1/x`. https://www.youtube.com/watch?v=13YrZLBmu7A https://www.youtube.com/watch?v=13YrZLBmu7A gives the same solution. GPT-4 failed but was close. Maybe GPT-3.5 is smarter ;) Query i used was: ``` Can you think step by step about this math problem and solve it? Let R+ denote the set of positive real numbers. Find all functions f : R+ → R+ such that for each x ∈ R+, there is exactly one y ∈ R+ satisfying xf(y) + yf(x) <= 2. ``` Response: https://pastebin.com/sTXM9kLt https://pastebin.com/sTXM9kLt Edit: Maybe i should say Bing not GPT-4 because i asked it there.
  - MichaelRazum 4y ago
    I asked gpt4 to think through it step by step and then to recheck the answer. At least it got 1/x right. So now kind of impressed.
- rcpt 4y ago
  Putnam asked "can a parabola of arc length 4 fit inside the unit circle?" Should be simple for a computer to just try every parabola but all the LLMs try to solve it like a human would (and get it wrong)
jmfldn 4y ago
It all depends on your definition of AGI as this term seems to be used with wildly different meanings. Before I can agree on any statement about AGI, I need to know what it is I'm being asked to agree / disagree on. Fwiw though, using this kind of mashup approach of gluing lots of things together, I suspect you'll get something that appears to match at least some people's definition of AGI.
taberiand 4y ago
I want to try providing ChatGPT with a plug-in to itself (+ whatever utility plugins, + a plug-in to create plugins), so that it can delegate sub tasks to chat sessions primed with context that it decides. I think a lot of the issues people have interacting with ChatGPT stem from overloading the context, and having a tree of specialised contexts recursively building on each other would I think focus each context while providing for broad and deep exploration of the top level request
ilaksh 4y ago
I think that's a misinterpretation of what these types of papers are. GPT3.5 and GPT4 are already general purpose with text tasks. You can give it _any_ well defined programming task that involves a fairly direct route from natural language to code, along with the API/module description (or output data format), and it can do it. That's the only reason this and 500 other papers or services like it (but doing different things) are possible. 3.5 is not multimodal though. That's why it needs the other models. But GPT4 has image understanding and can do a lot of these things without the external tools. There is no reason to think that similar variations of GPT will not be able to handle video understanding or generation at some point.

_yb2s 4y ago

I agree, I suspect AGI is possible right now with a similar system only slightly more sophisticated than this one. The right "glue" for existing models, and plugins to existing data sources all coordinated in a system. GTP-4 would do the managing, and handling, and some simple template API and handler script would allow it to call instances of itself or other models, track recursion depth, and automatically remind GTP-4 to stay on track and always use the correct 'templates' for requests. It could also remind it to create a condensed summary of it's goals and state that gets repeated back automatically to act as long term memory, and increase the effective context window size (edit: this is single variable external memory). I am afraid to explain this, because I have tried a preliminary version of it that I supervised step by step, and it seems to work. I think it is obvious enough that I won't have been the only one to think of it, so it would be safer to put the information out there so people can prepare. I see a big disconnect on here between people saying GPT-4 can't do things like this and is just a "stochastic parrot" or "glorified autocomplete," and people posting logs and summaries of it solving unexpectedly hard problems outside of any conceivable training set. My theory is that this disconnect is due to three major factors: * People confusing GPT-4 and GPT-3, as both are called "chatGTP" and most people haven't actually used GPT-4 because it requires a paid subscription, and don't realize how much better it is * Most popular conceptual explanations about how these models work imply that these actually observed capabilities should be fundamentally impossible * Expectations from movies, etc. about what AGI will be like, e.g. that it will never get confused or make mistakes, or that it won't have major shortcomings in specific areas. In practice this doesn't seem to limit it because it recognizes and fixes its mistakes automatically when it sees feedback (e.g. in programming)

basch 4y ago
It also, in my pay opinion, needs a memory store. I need to be able to save something to a variable and call back that exact variable. Right now, because it’s just pure text input of the whole conversation, it can forget or corrupt it’s “memory”. Even something like writing a movie script doesn’t work if it constantly forgets characters names or settings or plot points.
- _yb2s 4y ago
  My description above includes a memory store. It is a single variable that it can save and recover, and will automatically get sent back to it at regular intervals if it forgets it saved anything. Only a single text string variable is needed, as GTP-4 is smart enough to realize it can compress, pickle, etc. as many things as it needs into that.
  - alchemist1e9 4y ago
    I think we might be working something a bit similar and I agree memory is an issue. However it’s not really AGI that will emerge from this. It’s looking very useful but it’s still observable but it’s not a “mind” like an AGI. To add to your comments, I suggest that local vector store of embedding vectors of local content is how I’m going with memory issue. Langchain like. That way all previous progress is tracked and retrievable. That way the system can retrieve all it’s “memories” itself. The recursive multi agent pattern is big deal in my opinion.
    - nomel 4y ago
      > but it’s not a “mind” like an AGI. Please define this. ;)
      - WA 4y ago
        Please define AGI first ;)

roflyear 4y ago

Except, whatever model is doing the orchestration will still not "understand" the problem (be able to learn from new information provided).

_yb2s 4y ago
GPT-4 can do this. I realize that a description of how these models work implies that this is absolutely impossible, so I have no explanation other than suggesting you go try it out for yourself.
- antibasilisk 4y ago
  There is no description of how these models work, other than mechanically.
  - _yb2s 4y ago
    exactly... some people are using such descriptions to think about its limitations, but there is no reason to think they are accurate
- roflyear 4y ago
  Bro, GPT-4 can absolutely not understand new information. Try to teach it a grammar. It can't learn a grammar. Anyone can learn a new grammar. Not GPT-4. GPT-4 does not ask clarifying questions. It does not understand things.
  - _yb2s 4y ago
    I'm not sure why you think that? When I give it programming tasks, GPT-4 will ask all kinds of clarifying questions. Here is a trivial example: p: I want you to solve for x, giving me a numeric value, given the formula ax^2 + bx + c = 0. r: To find the numeric value of x for the quadratic equation ax^2 + bx + c = 0, we need to use the quadratic formula: x = (-b ± √(b^2 - 4ac)) / (2a) However, you haven't provided specific values for a, b, and c. Please provide these values so I can calculate the solution(s) for x. Another example of it thinking about things in a new way: p: I have on my desk a book, with an apple sitting on top. If I were to suddenly, rapidly remove the book such that the apple remained in place, in which direction would it eventually move? r: If you suddenly and rapidly remove the book from beneath the apple, the apple would initially stay in place due to its inertia. However, once the book is removed, there would be no support underneath the apple. As a result, the apple would eventually move downwards due to the force of gravity acting on it. p: Because the earth is a sphere, I prefer to use the terms “in” and “out” to refer to movement relating to that. Can you please rephrase your answer in the terms? r: Certainly! If you suddenly and rapidly remove the book from beneath the apple, the apple would initially stay in place due to its inertia. However, once the book is removed, there would be no support underneath the apple. As a result, the apple would eventually move "in" towards the center of the Earth, due to the force of gravity acting on it. (Note I didn't tell it which direction was in vs out. This is the example problem Chomsky gave in his recent article arguing that GPT-3 cannot think - per his ideas no language model would ever be able to answer a question like this. I went a lot deeper with this and got it to add fluid dynamics calculations for wind, etc.)

dvt 4y ago

> it turned it from a math dummy to a math genius overnight Imo your post fundamentally misunderstands a few things, but mainly how Wolfram works. Wolfram can be seen as a "database" that stores a lot of human mathematical information (along with related algorithms). Wolfram does not make new math. A corollary here is that AGI needs to have the ability to create new math to be truly AGI. But unless fed something like, e.g. the Principia, I don't think we could ever get a stochastic LLM to ever derive 1+1=2 from first principles (unless specifically trained to do so). Keep in mind that proving 1+1=2 from first principles isn't even new math (new math would be proving the Poincaré conjecture before Perelman did it, for example).

Sivart13 4y ago
If the litmus test for intelligence is the ability to “create new math” most people on earth wouldn’t be considered intelligent
- dvt 4y ago
  > If the litmus test for intelligence Moving goal posts around is unhelpful. I think my comment was pretty clear in the context of AGI and calling ChatGPT a "math genius."
  - famouswaffles 4y ago
    The fact that people take agi to mean "can invent new math" is a goalpost shift on its own. That was not the original meaning (generally intelligent), that wasn't even the next moved goal (on par with human experts). I guess the next one counts (better than all human experts) but i'm sure we'll move that again too.
    - akiselev 4y ago
      The problem was always underspecified. We don’t have a quantifiable metric for intelligence except IQ tests, benchmark datasets like those used to evaluate different LLMs, and other similarly myopic bullshit. “We’ll know it when we see it” becomes the default. In reality, the goal posts aren’t being moved, we’re just finding out how much further we are from them than we thought. ChatGPT is a "stochastic parrot" that’s seems way “smarter” than anyone thought possible so we have to reevaluate what we consider evidence of intelligence, perhaps coming to terms with the fact that we aren’t that smart the most of the time.

Manjuuu 4y ago

There will be no AGI during our lifetime. It seems weird to me that people that should understand how ML works (e.g. sama) instead of educating laypeople about about how this actually works and the huge limits the technology has, start talking nonsense about AGIs like some random scifi fan at a convention. Depressing.

AbrahamParangi 4y ago
I'll take the other side of that bet
- brotchie 4y ago
  +1
- Manjuuu 4y ago
  Feel free to do so, we need something to replace crypto after all. And also, sometime I wonder how many people hold those kind of opinions on AGI(imminent, doable, worthy of being discussed) because they sincerely believe in some nonsense like the basilisk thing for example.
  - bheadmaster 4y ago
    I don't believe in Basilisk because of the time-travel bullshit, but I do believe that AGI will come soon, because I don't believe in divine soul and see human mind as just a very complex statistics machine and all human learning as just detecting correlation between sensory input through time.
  - HDThoreaun 4y ago
    I believe in AGI because I'm a materialist so see no reason why we couldn't create artificial brains.
- arcastroe 4y ago
  Heh, @Manjuuu winning this bet would imply both of your deaths ...

antibasilisk 4y ago

Whose to say that AGI isn't already here but relatively slow? It does seem to me that as more and more outputs of AI are pushed through various systems, that will eventually feed back into various AI. In 2017 an Anon on 4chan proposed that our society, due to content marketing and AI stock trading, is already run by a sort of emergent intelligence, and he proposed that this is why things seem to be 'getting crazy' lately. I'm inclined to agree with that perspective, and if correct more capable AI systems could really kick this into high gear.

flangola7 4y ago
There's arguably three kids of superintelligence: quality, speed, and collective. We definitely have #3 and some instances of #2. You can outthink a human by thinking 1000x faster, or by having 1000 minds think on the same problem. There's a reason everything explodes when communication technology gets better. Imagine how long it would take to coordinate the fundamental pieces of a laptop computer by pigeon mail or horseback.

Mizza 4y ago

I've built a service for building systems like this. I haven't officially "launched" yet, but it's working and you can play with it here, if anybody is up to giving the alpha a spin: https://aistudio.fly.dev/ https://aistudio.fly.dev/

nomel 4y ago
I really hope this is the future of coding.
fnimick 4y ago
It's oddly fitting that your example workflow appears to be scraping content from elsewhere, rewording it, and generating an image in order to produce automated blogspam.
- sebzim4500 4y ago
  He knows his audience maybe too well.
93po 4y ago
other quick feedback: add a "add node" button, it isn't immediately clear you do this with right clicking
93po 4y ago
your "go pro" page is white text on white background in firefox

MichaelRazum 4y ago

"math genius" not sure about it. I gave it a problem from IMO 2022. In my opinion not a very hard one. It failed even if you give it some hints. Let R+ denote the set of positive real numbers. Find all functions f : R+ → R+ such that for each x ∈ R+, there is exactly one y ∈ R+ satisfying xf(y) + yf(x) <= 2. It just struggeld to reason. So I would be very surprised if the plugin somehow helps here.

Donald 4y ago
It's a language completion model, so it's not surprising at all that it struggles with logical inference. That's probably one of the reasons that instruction fine-tuning has such a dramatic effect in the performance of these models: they're finally given some of the underlying causal priors of the task domain.
knrdev 4y ago
Very strange. I gave this problem to GPT-3.5 and it answered `f(x)=1/x`. https://www.youtube.com/watch?v=13YrZLBmu7A https://www.youtube.com/watch?v=13YrZLBmu7A gives the same solution. GPT-4 failed but was close. Maybe GPT-3.5 is smarter ;) Query i used was: ``` Can you think step by step about this math problem and solve it? Let R+ denote the set of positive real numbers. Find all functions f : R+ → R+ such that for each x ∈ R+, there is exactly one y ∈ R+ satisfying xf(y) + yf(x) <= 2. ``` Response: https://pastebin.com/sTXM9kLt https://pastebin.com/sTXM9kLt Edit: Maybe i should say Bing not GPT-4 because i asked it there.
- MichaelRazum 4y ago
  I asked gpt4 to think through it step by step and then to recheck the answer. At least it got 1/x right. So now kind of impressed.
rcpt 4y ago
Putnam asked "can a parabola of arc length 4 fit inside the unit circle?" Should be simple for a computer to just try every parabola but all the LLMs try to solve it like a human would (and get it wrong)

jmfldn 4y ago

It all depends on your definition of AGI as this term seems to be used with wildly different meanings. Before I can agree on any statement about AGI, I need to know what it is I'm being asked to agree / disagree on. Fwiw though, using this kind of mashup approach of gluing lots of things together, I suspect you'll get something that appears to match at least some people's definition of AGI.

taberiand 4y ago

I want to try providing ChatGPT with a plug-in to itself (+ whatever utility plugins, + a plug-in to create plugins), so that it can delegate sub tasks to chat sessions primed with context that it decides. I think a lot of the issues people have interacting with ChatGPT stem from overloading the context, and having a tree of specialised contexts recursively building on each other would I think focus each context while providing for broad and deep exploration of the top level request

ilaksh 4y ago

I think that's a misinterpretation of what these types of papers are. GPT3.5 and GPT4 are already general purpose with text tasks. You can give it _any_ well defined programming task that involves a fairly direct route from natural language to code, along with the API/module description (or output data format), and it can do it. That's the only reason this and 500 other papers or services like it (but doing different things) are possible. 3.5 is not multimodal though. That's why it needs the other models. But GPT4 has image understanding and can do a lot of these things without the external tools. There is no reason to think that similar variations of GPT will not be able to handle video understanding or generation at some point.