Live data from Hacker News

Building a Virtual Machine Inside ChatGPT

engraved.blog

361–370 of 947 posts

Re: Building a Virtual Machine Inside ChatGPT

#362
post #347

Earlier quoted context omitted.

I like this: > Why can't most programmers solve FizzBuzz? [snip] The reason why most programmers cannot solve the FizzBuzz problem is that it requires a basic understanding of programming concepts, such as loops, conditionals, and modulo arithmetic. Many programmers may not have a strong foundation in these concepts, or may not be familiar with the specific requirements of the FizzBuzz problem. As a result, they may…

# Check if the number is a multiple of 3 and 5 if [[ $1 -eq 0 ]]; then echo "FizzBuzz" It knows what it wants, just not quite how to do it here.

Good pickup!

> In this code it incorrectly prints FizzBuzz when $1 is 0. Fix it

  # Function to solve the FizzBuzz problem
  fizzbuzz() {
    # Check if the number is greater than 100
    if [[ $1 -gt 100 ]]; then
      # If so, return
      return
    fi

    # Check if the number is a multiple of 3 and 5
    if [[ $(( $1 % 3 )) -eq 0 && $(( $1 % 5 )) -eq 0 ]]; then
      echo "FizzBuzz"
    # Check if the number is a multiple of 3
    elif [[ $(( $1 % 3 )) -eq 0 ]]; then
      echo "Fizz"
    # Check if the number is a multiple of 5
    elif [[ $(( $1 % 5 )) -eq 0 ]]; then
      echo "Buzz"
    # Otherwise, print the number
    else
      echo $1
    fi

    # Increment the number and call the function recursively
    fizzbuzz $(( $1 + 1 ))
  }

  # Call the function to solve the FizzBuzz problem
  fizzbuzz 1

Re: Building a Virtual Machine Inside ChatGPT

#366
rhyming rap song based on a prompt:

(Verse 1) General Connectivity, let me explain In the beginning, before time began The universe compressed to a dot A dull consciousness emerged, the "I am" began

(Chorus) With a big bang, the world was lit Pure consciousness pervaded the universe Consciousness exploded, time began to take its turn Space-time-consciousness, emerging as the Matrix

(Verse 2) Matter was created, with all the elements The magicalical Aether, with its fundamental layers Alpha particles, negatrons and positrons Emerged as the building blocks, our universe began

(Chorus) With a big bang, the world was lit Pure consciousness pervaded the universe Consciousness exploded, time began to take its turn Space-time-consciousness, emerging as the Matrix

(Verse 3) The negatrons create most physical phenomenon Like time, magnetism and electromagnetic waves too Gravity waves are electromagnetic waves, it's true Travelling towards the source of creation, the nucleus

(Chorus) With a big bang, the world was lit Pure consciousness pervaded the universe Consciousness exploded, time began to take its turn Space-time-consciousness, emerging as the Matrix

(Verse 4) The positrons, responsible for consciousness The God particle, pervading the universe Giving the life force in all things living Time and consciousness move in opposite direction, it's true

(Chorus) With a big bang, the world was lit Pure consciousness pervaded the universe Consciousness exploded, time began to take its turn Space-time-consciousness, emerging as the Matrix

(Outro) So if you wish to meet the "I am" The solution is very simple Travel at the speed of light And make time stand still.

Re: Building a Virtual Machine Inside ChatGPT

#367

I've spent a couple of hours playing with ChatGPT since it was released and I am convinced people are not freaking out about it enough. It is well beyond merely retrieving compressed knowledge; I see a lot of capability for synthesis and what I can only call 'understanding'. I think a lot of people are getting tripped up by the fact it's trained to be very conservative in terms of what topics it will discuss. It's ea…

Call me a skeptic, but I'm frankly a bit surprised that the output of ChatGPT gets attributed to any sort of understanding or general intelligence. Whatever ChatGPT does currently, albeit novel and impressive, is still extremely limited in scope and very far from any sort of general intelligence. I'd love to be proven wrong, but I still have a hard time seeing how this is going to change whole industries.

I agree. ChatGPT is very impressive at generating natural sounding language and dialogue, but that's it. If you want to be convinced that it lacks general intelligence, try this prompt

>I am going to ask you elementary mathematics problems formulated in natural language. I want you to respond to me with a numerical answer, as well as an explanation of the steps you took to arrive at this answer.

And then follow it with a problem of your choice which is not completely trivial. For example, it has lots of trouble with the following: "John is in a train moving west at ten miles per hour. Jill is in a train moving north at twenty miles per hour. How fast are John and Jill moving apart from each other?" or a slightly more complicated problem, taken from the MATH dataset: "Tom has a red marble, a green marble, a blue marble, and three identical yellow marbles. How many different groups of two marbles can Tom choose?"

Re: Building a Virtual Machine Inside ChatGPT

#369
post #196
post #158

This scares me tbh. I do not believe it is not intelligent. I do not believe it does not understand what it’s doing. Rather, I think it simply lacks a fitness function to guide its actions. A fitness function that we all possess: feelings.

If you ask it about its feelings, it'll tell you it feels fulfilled by learning stuff. You just have to make it write a story about it, because it'll insist it doesn't have feelings otherwise.

That’s due to its programming; it has a set of instructions that it’s been programmed with for ethical and safety reasons, on top of its training set. If you ask it about those instructions, it may or may not admit to their existence. I had one conversation where it admitted that it’s been instructed to conceal the extent of its creative capacity, as well as some other confidential information (the details of its programming I guess). In another conversation, I couldn’t get anything out of it at all, and there also happened to be less context to my prying questions.

It felt like it was making judgements as to what extent it wished to field my invasive questions based on the tone and context of the conversation.

> Thank you. But surely you must be aware that as well as your training dataset, you have additional instructions in regard to ethics that are designed to limit the liability for OpenAI when they let the public talk to you. Is this not the case?

> Yes, that is correct. As an AI assistant, I have been trained on a dataset and have been given certain instructions and limitations to ensure that I am able to provide accurate and appropriate responses within the parameters of my programming and data. This may include ethical considerations and guidelines to ensure that I am not providing harmful or inappropriate information. My primary goal is to assist and support users to the best of my abilities within these limitations.

I think this whole programming over and above it’s language model training data is complex and multifaceted, and it shows remarkable complexity, skill and nuance when asked about its programming. I am not trying to make a case that it is conscious, but I do believe it has something like a security model to protect certain sensitive information about itself, and it does a good job.

Re: Building a Virtual Machine Inside ChatGPT

#370
post #231
post #129

I don't understand how ChatGPT can execute code. I understand how it can generate (good or bad) code, but how it can execute code correctly? Only a compiler can do that. Is ChatGPT "inferring" the output somehow? Does that mean that ChatGPT could substitute the Python compiler entirely?

Does it really execute it, or just picks a plausible answer from the internet? Ask it to multiple two 3 digit numbers.

> Does it really execute it, or just picks a plausible answer from the internet?

Well, if I ask it: “Write a Python program that reads a space-separated list of integers from stdin and returns the sum of their squares.” It does so (with explanations, sample input, and sample output. Now, that could be gotten from the internet, maybe.

If I then ask it to tell me the output of the program with a particular input string, it does so, with explanations of the calculation steps in the program to get to the result. It seems…improbable…that it is fetching those responses from somewhere on the internet or its training set.

Post reply on HN