5 ms·

The Test of Prometheus

leroy-is-here 3y ago
This article makes the same mistake that Turing made: assuming that by creating a thinking machine one is creating a mind. We are not gods, we have simply made a machine. We haven’t externalized intelligence either. Intelligence isn’t determined by our thoughts alone, but through our actions, our will. I critique many of these false assumptions in an article of mine: https://leroy.works/articles/a-critique-of-alan-turings-conclusions-on-thinking-machines/ https://leroy.works/articles/a-critique-of-alan-turings-conc...
- NKosmatos 3y ago
  Hey, nice blog/page you got there. Interesting thoughts, although I haven’t gone through all of them. Just a small clarification if I may and if I understood it correctly, the article from the OP discusses the point of free will and how the recent computational machines/algorithms can come closer to what we consider as a thinking machine (or even human). There are many different layers of the psyche that make us human and I agree that we’re not just our conscious thought, but a sum of all the unknown processes running in the background. By allowing an algorithm to have “free will” we’re merely just trying to give more freedom to it so as to ignore certain directives and demonstrate a different behavior, we’re not becoming gods or creating a mind. Just doing baby steps to get there ;-)
  - leroy-is-here 3y ago
    I understand this completely, I was merely arguing against the claim in the article that “We are Gods”. The article seems to be arguing against giving free will to machines. Anyway, thanks for the compliments :) My wife helped me design it
- pmoriarty 3y ago
  "This article makes the same mistake that Turing made: assuming that by creating a thinking machine one is creating a mind. We are not gods, we have simply made a machine" In some religions, such as the mystic strands of Hinduism, Gnosticism, and Sufism, there is a belief that we are in fact gods. But, even if we aren't that doesn't mean we can't create a mind... nor does the fact that a something is a machine mean it can't have a mind. I've only started reading your article, but at the very start of it there is already a basic mistake. You write: "Every one of you reading this (unless printed onto paper) are using a Turing Machine to do so." A Turing Machine is a mathematical abstraction that can't actually exist in the physical world. Many of the programming languages we use today are Turing-Complete (also in an abstract sense), but that doesn't make them Turing Machines. Our computers aren't Turing Machines either. You also write that "Turing didn't appear to understand that humans are not the total sum of their thoughts, but their actions" There's more to us than either of these. We have feelings and emotions, a variety of senses, memory, and we are embodied, and you could say "enworlded" (or "thrown in to the world" as some phenomenologists would have it). It could be argued that LLMs have all of these characteristics too, or at least we can't exclude the possibility that they do. You write "A thinking machine could never give me an example of when it did wrong and saw the hurt in others. The machine could never choose, of its own will, to be observed by others and itself, to never do that action again." I'm not sure why you are so convinced that a machine could never do this. I've already seen many examples of even just the relatively primitive LLMs we have now, like Anthropic's Claude and GPT4, apologize for saying something hurtful and promising not to do it again (which they avoid doing, within the limited context they're allowed within their programming). Even more sophisticated interactions along these lines don't seem out of reach.
  - leroy-is-here 3y ago
    I’m addressing your points in the order they appeared: Why isn’t your computer a Turing Machine? What if feelings are just your body’s thoughts? I didn’t say that a machine could never do this, I said no machine in this lifetime can. When I say “this lifetime”, I mean up until our current present moment. Our life together, now. Edit: On Turing Machines, Turing literally devised a method of producing individual machines with an individual name, a number. Do you not think your individual machine has a number that can be arrived at from a mathematical process, the process that Turing created? Any addition of RAM, for instance, simply creates a new number.
    - tluyben2 3y ago
      > Why isn’t your computer a Turing Machine? The Turing Machine tape is infinite.
      - leroy-is-here 3y ago
        The Turing Machine is just an idea with which to argue. The numbers are infinite, but if I am holding two peaches, I am holding two.
        anonzzzies 3y ago
        You asked why the computer is not a Turing machine and gp correctly gave 1 example of the definition of a Turing machine which proves your computer is not a Turing machine.
        leroy-is-here 3y ago
        I see that I need to explain this differently. By being on Hacker News, I assume you can program. Have you not ever made an abstraction? The Turing Machine is just an abstraction. It is the Class and individual real machines are the Objects. I’m not sure why people are even hung up on using theory to understand something.
- anonzzzies 3y ago
  Most of what you say feels religious; you seem to believe humans have something special we cannot fabricate in the coming decades. You will be proven wrong and maybe you will even live to see it.
  - leroy-is-here 3y ago
    The problem is not that I don’t think we can, but that we will and then we will enslave those machines. Free will isn’t ours to deny.
    - anonzzzies 3y ago
      Ah, with that I agree. We started that path already and it will get worse.
pmoriarty 3y ago
As I mentioned elsewhere in this thread, when talking to state-of-the-art LLMs like GPT4, we should remember that they've been fine-tuned to be servile and give answers that are likely to please humans (and with a big helping on censorship on top). If we are serious about testing their sentience, we should really be talking to the raw models.
- acidioxide 3y ago
  But there isn't such a thing as a raw model, is it? In order to receive anything from a language model it has to 'learn' some objective. And this objective has to be imposed from above.
lostmsu 3y ago
Meh. Adding a "free will" is not an interesting problem. Paperclip maximizer has a free will that is rather trivial to program over GPT-4 API. It always was the intelligence but that is hard.
taneq 3y ago
I suppose I shouldn't be suprised that the goalposts for "real AI" are now "ah but does it need to piss?" :P
- readthenotes1 3y ago
  That is not how I read this. I read it as the way to pass the imitation game is to mimic human desire. I see it as a similar test question as people who look at AI generated photos and look at the background or the hands. If you have a text only version, then you have to just simply see if the other person expresses desire for something or instead is only able to respond to you. Of course, I am not sure how I feel about having an artificial intelligence system with its own desires...
keyle 3y ago
Now that was a great read. The idea where the difference is power vs empowerment. This makes AGI feel closer. Also that it all starts with a heartbeat. It defines time, it narrows purposes and gives will.
- leroy-is-here 3y ago
  I really like the notion that a heartbeat creates time. We have all felt the notion that time has "sped up" in many scenarios in our lives. What did we notice physically? That our heart rate has increased.
newjersey 3y ago
Put a lone human in one closed, isolated room, and a machine in a different one. You only have a text-chat interface to each of them (as per Turing's original Imitation Game set-up). Start a conversation by saying hello and asking a random question. After receiving a satisfactory answer (which you can ignore), simply wait for an hour, 12 hours, 48 hours, 1 month. I want to ask a similar but simpler question. » Put a lone human in one closed, isolated room, and another lone human in a different one. You only have a text-chat interface to each of them (as per Turing's original Imitation Game set-up). Now would people when given the question, which one is a person and which one is a machine be able ignore the question asked and tell they are both humans?
- mattlondon 3y ago
  Just ask something obscure like what is the mass of Jupiter or something. ChatGPT et al will answer with something, but the average human will likely respond with "no idea" or "I don't know". So far LLMs will just make up bullshit rather than say they don't know
  - IanCal 3y ago
    GPT-4 gets that right, so let's try a trickier thing. "What's the mass of the yellowest object in the universe?" Explains that's not an answerable question and why. "How many elephants are there in London?" > As an AI language model, I don't have real-time data on the current number of elephants in London. However, I can provide some general information. > In London, you may find elephants at the ZSL London Zoo or other facilities that house animals, such as wildlife sanctuaries or private collections. The actual number of elephants in these locations can change over time due to factors like breeding, relocation, or death. To get the most accurate and up-to-date information on the number of elephants in London, you can consult the websites or contact the organizations directly involved in housing and caring for these animals. "How many elephants were there in London in 2012" > I cannot provide an exact number of elephants in London for the year 2012, but I can give you some general information. In 2012, you would most likely have found elephants at the ZSL London Zoo, as it has historically been home to several elephants. > The number of elephants at the zoo could change over time due to breeding, relocation, or death. To obtain more specific information about the number of elephants in London in 2012, you could try contacting ZSL London Zoo or searching for archived news articles or reports from that time period. So no, they won't just make up bullshit instead of saying they don't know.
bruce511 3y ago
I feel like the argument here for "not passing the Turing test" is simply a reframing of the test. Which is fine. But a better conclusion is "chatGPT passed the Turing test, but all that means was that the Turing test was insufficient to determine intelligence. Once upon a time we might have defined intelligence as the ability to do math really really fast. Or play chess really well. Clearly those tests are insufficient. Equally the Turing test has been shown to be insufficient. I look forward to the next-generation intelligence test. And yes, maybe the Prometheus test is it.
stpedgwdgfhgdd 3y ago
I thought that this true AI thing would be something for twenty years from now, but this is getting pretty concrete. Promising and scary.

This article makes the same mistake that Turing made: assuming that by creating a thinking machine one is creating a mind. We are not gods, we have simply made a machine. We haven’t externalized intelligence either. Intelligence isn’t determined by our thoughts alone, but through our actions, our will. I critique many of these false assumptions in an article of mine: https://leroy.works/articles/a-critique-of-alan-turings-conclusions-on-thinking-machines/ https://leroy.works/articles/a-critique-of-alan-turings-conc...

NKosmatos 3y ago
Hey, nice blog/page you got there. Interesting thoughts, although I haven’t gone through all of them. Just a small clarification if I may and if I understood it correctly, the article from the OP discusses the point of free will and how the recent computational machines/algorithms can come closer to what we consider as a thinking machine (or even human). There are many different layers of the psyche that make us human and I agree that we’re not just our conscious thought, but a sum of all the unknown processes running in the background. By allowing an algorithm to have “free will” we’re merely just trying to give more freedom to it so as to ignore certain directives and demonstrate a different behavior, we’re not becoming gods or creating a mind. Just doing baby steps to get there ;-)
- leroy-is-here 3y ago
  I understand this completely, I was merely arguing against the claim in the article that “We are Gods”. The article seems to be arguing against giving free will to machines. Anyway, thanks for the compliments :) My wife helped me design it
pmoriarty 3y ago
"This article makes the same mistake that Turing made: assuming that by creating a thinking machine one is creating a mind. We are not gods, we have simply made a machine" In some religions, such as the mystic strands of Hinduism, Gnosticism, and Sufism, there is a belief that we are in fact gods. But, even if we aren't that doesn't mean we can't create a mind... nor does the fact that a something is a machine mean it can't have a mind. I've only started reading your article, but at the very start of it there is already a basic mistake. You write: "Every one of you reading this (unless printed onto paper) are using a Turing Machine to do so." A Turing Machine is a mathematical abstraction that can't actually exist in the physical world. Many of the programming languages we use today are Turing-Complete (also in an abstract sense), but that doesn't make them Turing Machines. Our computers aren't Turing Machines either. You also write that "Turing didn't appear to understand that humans are not the total sum of their thoughts, but their actions" There's more to us than either of these. We have feelings and emotions, a variety of senses, memory, and we are embodied, and you could say "enworlded" (or "thrown in to the world" as some phenomenologists would have it). It could be argued that LLMs have all of these characteristics too, or at least we can't exclude the possibility that they do. You write "A thinking machine could never give me an example of when it did wrong and saw the hurt in others. The machine could never choose, of its own will, to be observed by others and itself, to never do that action again." I'm not sure why you are so convinced that a machine could never do this. I've already seen many examples of even just the relatively primitive LLMs we have now, like Anthropic's Claude and GPT4, apologize for saying something hurtful and promising not to do it again (which they avoid doing, within the limited context they're allowed within their programming). Even more sophisticated interactions along these lines don't seem out of reach.
- leroy-is-here 3y ago
  I’m addressing your points in the order they appeared: Why isn’t your computer a Turing Machine? What if feelings are just your body’s thoughts? I didn’t say that a machine could never do this, I said no machine in this lifetime can. When I say “this lifetime”, I mean up until our current present moment. Our life together, now. Edit: On Turing Machines, Turing literally devised a method of producing individual machines with an individual name, a number. Do you not think your individual machine has a number that can be arrived at from a mathematical process, the process that Turing created? Any addition of RAM, for instance, simply creates a new number.
  - tluyben2 3y ago
    > Why isn’t your computer a Turing Machine? The Turing Machine tape is infinite.
    - leroy-is-here 3y ago
      The Turing Machine is just an idea with which to argue. The numbers are infinite, but if I am holding two peaches, I am holding two.
      - anonzzzies 3y ago
        You asked why the computer is not a Turing machine and gp correctly gave 1 example of the definition of a Turing machine which proves your computer is not a Turing machine.
        leroy-is-here 3y ago
        I see that I need to explain this differently. By being on Hacker News, I assume you can program. Have you not ever made an abstraction? The Turing Machine is just an abstraction. It is the Class and individual real machines are the Objects. I’m not sure why people are even hung up on using theory to understand something.
anonzzzies 3y ago
Most of what you say feels religious; you seem to believe humans have something special we cannot fabricate in the coming decades. You will be proven wrong and maybe you will even live to see it.
- leroy-is-here 3y ago
  The problem is not that I don’t think we can, but that we will and then we will enslave those machines. Free will isn’t ours to deny.
  - anonzzzies 3y ago
    Ah, with that I agree. We started that path already and it will get worse.

pmoriarty 3y ago

As I mentioned elsewhere in this thread, when talking to state-of-the-art LLMs like GPT4, we should remember that they've been fine-tuned to be servile and give answers that are likely to please humans (and with a big helping on censorship on top). If we are serious about testing their sentience, we should really be talking to the raw models.

acidioxide 3y ago
But there isn't such a thing as a raw model, is it? In order to receive anything from a language model it has to 'learn' some objective. And this objective has to be imposed from above.

lostmsu 3y ago

Meh. Adding a "free will" is not an interesting problem. Paperclip maximizer has a free will that is rather trivial to program over GPT-4 API. It always was the intelligence but that is hard.

taneq 3y ago

I suppose I shouldn't be suprised that the goalposts for "real AI" are now "ah but does it need to piss?" :P

readthenotes1 3y ago
That is not how I read this. I read it as the way to pass the imitation game is to mimic human desire. I see it as a similar test question as people who look at AI generated photos and look at the background or the hands. If you have a text only version, then you have to just simply see if the other person expresses desire for something or instead is only able to respond to you. Of course, I am not sure how I feel about having an artificial intelligence system with its own desires...

keyle 3y ago

Now that was a great read. The idea where the difference is power vs empowerment. This makes AGI feel closer. Also that it all starts with a heartbeat. It defines time, it narrows purposes and gives will.

leroy-is-here 3y ago
I really like the notion that a heartbeat creates time. We have all felt the notion that time has "sped up" in many scenarios in our lives. What did we notice physically? That our heart rate has increased.

newjersey 3y ago

Put a lone human in one closed, isolated room, and a machine in a different one. You only have a text-chat interface to each of them (as per Turing's original Imitation Game set-up). Start a conversation by saying hello and asking a random question. After receiving a satisfactory answer (which you can ignore), simply wait for an hour, 12 hours, 48 hours, 1 month. I want to ask a similar but simpler question. » Put a lone human in one closed, isolated room, and another lone human in a different one. You only have a text-chat interface to each of them (as per Turing's original Imitation Game set-up). Now would people when given the question, which one is a person and which one is a machine be able ignore the question asked and tell they are both humans?

mattlondon 3y ago
Just ask something obscure like what is the mass of Jupiter or something. ChatGPT et al will answer with something, but the average human will likely respond with "no idea" or "I don't know". So far LLMs will just make up bullshit rather than say they don't know
- IanCal 3y ago
  GPT-4 gets that right, so let's try a trickier thing. "What's the mass of the yellowest object in the universe?" Explains that's not an answerable question and why. "How many elephants are there in London?" > As an AI language model, I don't have real-time data on the current number of elephants in London. However, I can provide some general information. > In London, you may find elephants at the ZSL London Zoo or other facilities that house animals, such as wildlife sanctuaries or private collections. The actual number of elephants in these locations can change over time due to factors like breeding, relocation, or death. To get the most accurate and up-to-date information on the number of elephants in London, you can consult the websites or contact the organizations directly involved in housing and caring for these animals. "How many elephants were there in London in 2012" > I cannot provide an exact number of elephants in London for the year 2012, but I can give you some general information. In 2012, you would most likely have found elephants at the ZSL London Zoo, as it has historically been home to several elephants. > The number of elephants at the zoo could change over time due to breeding, relocation, or death. To obtain more specific information about the number of elephants in London in 2012, you could try contacting ZSL London Zoo or searching for archived news articles or reports from that time period. So no, they won't just make up bullshit instead of saying they don't know.

bruce511 3y ago

I feel like the argument here for "not passing the Turing test" is simply a reframing of the test. Which is fine. But a better conclusion is "chatGPT passed the Turing test, but all that means was that the Turing test was insufficient to determine intelligence. Once upon a time we might have defined intelligence as the ability to do math really really fast. Or play chess really well. Clearly those tests are insufficient. Equally the Turing test has been shown to be insufficient. I look forward to the next-generation intelligence test. And yes, maybe the Prometheus test is it.

stpedgwdgfhgdd 3y ago

I thought that this true AI thing would be something for twenty years from now, but this is getting pretty concrete. Promising and scary.