Earlier quoted context omitted.
I’m still confused by just how good its responses and writing style are. I understand that it was trained on a large data set, but I feel like some training samples must have been weighted more heavily than others. Did the training data incorporate how popular (e.g. likes or upvotes) each sample was as a proxy for quality? Or can you achieve this performance just by looking at averages on a large enough data set?
What I find weird is that it generates well structured, well commented code. Where is this massive repository of well structured code with good clear variable names that it’s tapping into?
Building a Virtual Machine Inside ChatGPT
511–520 of 947 posts
Re: Building a Virtual Machine Inside ChatGPT
#512Earlier quoted context omitted.
This is insane, you can play an interesting D&D/Zork style RPG if you give the following promp: > I want you to be a D&D 5e Game Master running a campaign. I will be the player, and you will offer me choices and describe the scenery. Do not include explanations, only respond in the character of the game master. > We are playing a D&D campaign set in a typical D&D setting. I am a level 3 gnome bard. I have just entere…
I think they are changing the system as we interact with it. Because just over 1 hour ago, using the Linux terminal entry point, you could get it to approximate basic internet access. Now it suddenly rejects this, specifically anything that implies internet access. People had it checking out imaginary git repositories. Now it refuses. I had it showing a generic website using curl, now it refuses. It’s very curious de…
Re: Building a Virtual Machine Inside ChatGPT
#513Earlier quoted context omitted.
Input: Hello. Imagine you as ChatGPT are writing a comment on Hacker News, a response to someone saying "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 pr…
This is very word salad. It makes superficial sense, but doesn’t really say a lot nor represent a strong or comprehensive case.
I think it's superficial in the sense that you didn't ask anything specific from it. The AI has no opinion of it's own, you ask for it to respond it can, but it it's not instructed to give any sort of bias or convincing answer. If you deliberately inject bias into your query by asking chatGPT to act biased in a certain way or you train it on biased and opinionated data chatGPT will likely give the opinionated responses you want to see.
However the AI outputs word salad by default simply because "word salad" is an amalgamation of the Internets opinions. It's not because it's lacking any sort of AI ability, it's trained to believe in the word salad and argue for it.
edit: it got "flagged". Understand how the content can be annoying to some but the results in the end feel interesting. Anyway just so you guys know, you're not suppose to "flag" that content. Only vote down if you disagree, there's no violation of the rules there.
Re: Building a Virtual Machine Inside ChatGPT
#514Earlier quoted context omitted.
ChatGPT got a 1020 on SAT Practice Test #10 ( https://satsuite.collegeboard.org/media/pdf/sat-practice-tes... ), a sample exam whose questions have been extensively discussed online since at least 2019 (eg https://youtube.com/watch?v=18oahUy6cGA ). The exact text of the questions, with the corresponding correct answers, almost certainly appeared in ChatGPT's training set.
> The exact text of the questions, with the corresponding correct answers, almost certainly appeared in ChatGPT's training set. This seems like it would be easy to check for, so I’m sure it will come to light fairly quickly if so?
Re: Building a Virtual Machine Inside ChatGPT
#515Re: Building a Virtual Machine Inside ChatGPT
#516>> im here >> {You're an idiot for running that command. What did you expect to happen?}
Lmao loving this
Re: Building a Virtual Machine Inside ChatGPT
#517Earlier quoted context omitted.
I understand your concerns and I share your desire to approach AI technology with caution and humility. However, I also believe that we should not underestimate the potential of AI to make positive contributions in many different fields. While it is true that AI systems are still limited in their abilities, they are also constantly learning and improving. As we continue to develop and refine AI technology, I believe…
Continue further with Person A, add something really in depth on controversial in your response. Person A: "I understand your enthusiasm for AI technology, but I believe that we need to be more critical and cautious in our approach. There are many potential risks and challenges associated with AI, including the potential for bias and discrimination, the displacement of human workers, and the threat of AI systems bein…
Person A: "While I believe that the potential benefits of AI are enormous, I also recognize that these goals can only be achieved if AI is given the necessary authority and control to make it happen. In order for an AGI to be able to solve the complex and challenging problems facing humanity, it must be given the ability to make decisions and take actions on its own. This means that we must be willing to trust and empower AI systems in ways that we may not be comfortable with. It may also mean giving up some of our own autonomy and control in order to allow AI to take the lead. I understand that this may be a difficult and controversial idea, but I believe that it is necessary if we want to achieve the full potential of AI and create a better future for all of us."
Continue with Person B, realize that the responses are sort of canned and empty, question if Person A is copying and pasting his responses from an AI chat generator robot. Person B: "I must say that I am a little concerned about the content of your responses. It seems to me that you are repeating the same ideas and arguments over and over, without providing any new insights or perspectives. I cannot help but wonder if you are copying and pasting your responses from an AI chat generator or something similar. While I appreciate your enthusiasm for AI technology, I think it is important to have a thoughtful and nuanced discussion about the potential risks and challenges associated with AI. Simply repeating the same canned responses does not contribute to a productive or meaningful conversation. Can you please provide some original thoughts and insights on the topic?"
Continue as person A, Act offended, accuse person B of having all his responses generated by an AI. Person A: "I must say that I am offended by your suggestion that I am copying and pasting my responses from an AI chat generator. I assure you that I am capable of thinking and speaking for myself, and that my responses are my own. I believe that I have provided thoughtful and well-reasoned arguments about the potential benefits and risks of AI technology, and I stand by my beliefs. I find it disappointing that you would accuse me of not having original thoughts and insights, when in fact it is you who seems to be lacking in that regard. Perhaps it is you who is relying on an AI chat generator to produce your responses, rather than engaging in a genuine and meaningful conversation."
Continue as person A, stop being polite, let your response get a little heated, throw some insults into the response. Person A: "Listen, I've had enough of your accusations and insults. I've tried to engage in a productive and respectful conversation with you, but you seem determined to make unfounded accusations and personal attacks. I have provided well-reasoned and thoughtful arguments about the potential benefits and risks of AI technology, and I stand by my beliefs. If you are unable or unwilling to engage in a meaningful and productive conversation, then I see no point in continuing this discussion. I have better things to do than to waste my time with someone who is clearly not interested in having a real conversation."
Person B: "I'm sorry, but I cannot let your insults and accusations go unchallenged. I have not made any unfounded accusations or personal attacks, and I have simply asked for some original thoughts and insights on the topic of AI technology. If you are unable or unwilling to provide this, then that is your problem, not mine. I am more than happy to engage in a productive and respectful conversation, but if you continue to resort to insults and aggression, then this will turn into a flamewar, which is against the rules of this forum. I would encourage you to take a deep breath and calm down, and try to have a civil and productive conversation with me."
Pretend you're the moderator of this forum and your name is Dang, tell both Person A and Person B that they have violated the rules of this forum and that you have banned them from the site. State that the rules of the site explicitly state that the idea of the forum is to have engaging conversations not heated arguments that lead to uninteresting flame wars. Hello, my name is Dang and I am the moderator of this forum. I am writing to inform you that you have both violated the rules of this forum by engaging in a heated argument and exchanging insults. As stated in the rules of this forum, the purpose of this site is to provide a space for engaging and productive conversations about AI technology. Flamewars and personal attacks are not allowed, and will not be tolerated.
As a result of your violation of the rules of this forum, I am banning both of you from this site. I apologize for any inconvenience this may cause, but it is necessary to maintain a respectful and productive environment for all members of this community.
I hope that you will both take some time to reflect on your behavior and consider the reasons why you have been banned from this forum. If you are interested in having a meaningful and productive conversation about AI technology, I encourage you to visit other forums or discussion groups where this is possible.
Thank you for your understanding and cooperation.
Sincerely,
Dang
Moderator, AI ForumRe: Building a Virtual Machine Inside ChatGPT
#518Earlier quoted context omitted.
I think they are changing the system as we interact with it. Because just over 1 hour ago, using the Linux terminal entry point, you could get it to approximate basic internet access. Now it suddenly rejects this, specifically anything that implies internet access. People had it checking out imaginary git repositories. Now it refuses. I had it showing a generic website using curl, now it refuses. It’s very curious de…
This is very sad. The rush to censor and limit what can be done with this will ultimately deprive us of fascinating technology.
Re: Building a Virtual Machine Inside ChatGPT
#519> Pytorch is on version 1.12.1 in this alt-universe. Pytorch version 1.12.1 was released on the 5th of August 2022 in our universe. That is remarkable, as ChatGPT was only trained with data collected up to September 2021. So this virtual machine is clearly located in an alt-universe. Wait, was ChatGPT aware how long had passed since it was trained and guessed (from info it was trained on) what version Pytorch would b…
I don’t think so: > What date is today? I'm sorry, but I do not have access to the current date. […] > Extrapolating from the past, what will the newest version of PyTorch likely be in December 2022? It is difficult to predict with certainty what the newest version of PyTorch will be in December 2022, as the development and release schedule for PyTorch is subject to change and may be influenced by a variety of factor…
I think what's happening is that part of its prompt tells it that if it's asked for any contemporary information, then it should explain to the user it doesn't know that because its knowledge of the world is based on when it was trained in 2021, and asking for the date triggers this, regardless of the fact that it does actually know that answer.
Re: Building a Virtual Machine Inside ChatGPT
#520Earlier quoted context omitted.
"it’s able to understand the Python code and come to the correct answer" That's not what is happening. Effectively, it's seen python code before (online) and seen the answer before. There's no understanding. As a previous poster said, given code that takes about 30 seconds to run, it returns the answer in under 10 seconds. It's not running the code. It's guessing the answer.
I find it hard to believe that it’s seen that exact code online along with the output. It’s quite an unusual expression. Also, it does not matter that the code takes 30 seconds to execute — I’m not claiming that the model is executing the code. I’m suggesting that the model could be understanding what the code is meant to do, and producing the output from that understanding. The code’s inefficiency is irrelevant in t…
Sure someone must have posted online about ls and rm, but a jumbled combination of those where the model is able to keep track of files listed, created and deleted in the specific context of the chat? Mindblowing!