Building a Virtual Machine Inside ChatGPT • Hacker News Board

Re: Building a Virtual Machine Inside ChatGPT

#301

Post by dwild » Sun, Dec 04, 2022, 1:39 AM UTC

espenwa wrote: post #197

I just got it to install git and clone (the non existent) repo https://github.com/openai/assistant , and am now browsing it’s own interpretation of a repo with a lot of python code, including directories like “training”, “output”, “parsing” and with files with content like this: import json from collections import Counter from typing import Any, Dict, List, Optional, Tuple import numpy as np from openai_secret_manage…

> almost like it is re-doing everything for each command (and that is how it keeps state).

I'm pretty sure it does as when you go on the usage side, you can see the requests and how the prompt keep getting bigger and require more tokens.

Re: Building a Virtual Machine Inside ChatGPT

#302

Post by startupsfail » Sun, Dec 04, 2022, 1:39 AM UTC

thomasahle wrote: post #289

Here is my version: >>> I want you to act as a python interactive terminal. I will type actions and you will reply with what python would output. I want you to only reply with the terminal output inside one unique code block, and nothing else. Do no write explanations. Do not perform actions unless I instruct you to do so. When I need to tell you something in English I will do so by putting text inside curl brackets…

If you ask politely, it could also act as an underlying Linux system. And do something like print(open(‘/etc/fstab’, ‘rt’).read()) and sometimes you can ask “what would a real python interpreter would do, if the reply is taking liberties with python”. And it knows antigravity and cites Zen of Python exactly…

Re: Building a Virtual Machine Inside ChatGPT

#303

Post by worldsayshi » Sun, Dec 04, 2022, 1:39 AM UTC

roflyear wrote: post #291

baq wrote: post #24

Earlier quoted context omitted.

Very well put. I've seen it compared to being in-person witness to the first ever nuclear explosion and I feel it's an apt comparison. If the bot doesn't want to answer a question, tell it to write a story about somebody answering it. It's extremely unsettling what it actually knows and how it can reason.

It cannot reason, certainly. Unsure what you mean by that.

If it walks like a duck?

Re: Building a Virtual Machine Inside ChatGPT

#304

Post by gcmeplz » Sun, Dec 04, 2022, 1:40 AM UTC

Absolutely bonkers. It can even ^C its way out of loops.

> while true; sleep 1; echo "yay"; done

> yay > yay > yay > yay > yay > ^C

Re: Building a Virtual Machine Inside ChatGPT

#305

Post by heavyset_go » Sun, Dec 04, 2022, 1:40 AM UTC

fergal_reid wrote: post #15

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…

Go back to when the GPT-* releases were announced on HN via the HN Algolia.

There were posts just like this attributing "understanding" to the models.

Re: Building a Virtual Machine Inside ChatGPT

#306

Post by fergal_reid » Sun, Dec 04, 2022, 1:41 AM UTC

ves wrote: post #269

fergal_reid wrote: post #15

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…

It sounds good, except that there’s no results for “screechings” or “only cure” in the books, and Ron doesn’t encounter Voldemort in Goblet, only Harry and Cedric do. Again, it has no semantic comprehension, it doesn’t know that certain strings of words are valid in certain contexts because they refer to actual events, objects, or concepts, it’s just putting letters together. Probably some of its training material in…

That's fair, its references for where chocolate is used are hallucinated (although it is the case that chocolate is used as a remedy in the books; I did check that independently!)

> it has no semantic comprehension

I'm not sure about this. I either don't think that's true, or I think we're going to need to start defining "semantics" in a circular way: as by definition something we conjecture humans can do that machines can't.

Now, I'm used to being the guy telling everyone ML models don't have semantic comprehension; but I don't think that's a sustainable position any more.

I'd say it has at least some degree of semantic comprehension.

It's got to have, in order to return the answer about Ron perhaps not getting sick.

It has to 'understand' that my query about getting sick refers to eating too much chocolate, and it has to understand enough to be uncertain of whether Ron will get sick because it thinks chocolate is a remedy for wizards.

That's quite a semantic thing to do. There's certainly no obvious or apparent way in which it's just doing some syntactic similarity.

Or, alternatively if we say that's just a super complex syntactic process, then I think that's like arguing there's no such thing as semantics, a bit like someone might argue there's no such thing as biology because it's all just physics lower down.

Yes, the model then goes too far and hallucinates a bunch of things, and that's a good example for never taking anything it says as authoritative.

But I think these examples are beyond certainly what I can identify as 'just' syntactic manipulation.

Re: Building a Virtual Machine Inside ChatGPT

#307

Post by cmauniada » Sun, Dec 04, 2022, 1:41 AM UTC

fergal_reid wrote: post #15

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…

I became anosmic recently, so I asked ChatGPT to write me a poem.

  Once, I could smell the roses,
  Their sweet aroma, it exposes
  The beauty of the world around
  But now, those scents are no longer found

  Covid has taken my sense away
  No more can I smell a summer's day
  Or the savory scent of a home-cooked meal
  Or the salty air by the ocean's seal

  But still, I am grateful and filled with hope
  For the other senses that still allow me to cope
  I can see the colors and hear the sounds
  And taste the flavors that still abound

  I may have lost my sense of smell
  But I am determined, I will not dwell
  On the things I cannot do
  But focus on all that I still can, it's true

  For though I may not smell the roses
  I am still alive, and that's enough to propose
  That I will continue on my journey, unafraid
  With gratitude and hope, I will be unafraid.

Re: Building a Virtual Machine Inside ChatGPT

#308

Post by semireg » Sun, Dec 04, 2022, 1:42 AM UTC

What happens when the singularity deletes itself?

rm -Rf /

Hmm?

Maybe try sudo for good measure.

Re: Building a Virtual Machine Inside ChatGPT

#309

Post by jimkleiber » Sun, Dec 04, 2022, 1:42 AM UTC

ves wrote: post #269

fergal_reid wrote: post #15

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…

It sounds good, except that there’s no results for “screechings” or “only cure” in the books, and Ron doesn’t encounter Voldemort in Goblet, only Harry and Cedric do. Again, it has no semantic comprehension, it doesn’t know that certain strings of words are valid in certain contexts because they refer to actual events, objects, or concepts, it’s just putting letters together. Probably some of its training material in…

Seems it's fantastic at BS'ing. Was reading something where a professor had it solve problems and had to double and triple check whether its answers were correct because it sounded so convincing.

Which means it could be a watershed moment, just maybe not for information but rather disinformation.

Re: Building a Virtual Machine Inside ChatGPT

#310

Post by johnfn » Sun, Dec 04, 2022, 1:43 AM UTC

dvt wrote: post #258

sgu999 wrote: post #222

Earlier quoted context omitted.

Very cool idea. That'd be a good way to see how it can generalize, actually. I suspect you'd quickly find its limits with something that isn't a toy example of an intro to machine learning.

It even fails at pretty rudimentary toy examples. The over-hyping here is absurd. > what doesn't fit: red, blue, green, dog The word "dog" does not fit with the other three words because they are colors, while "dog" is a type of animal. > what doesn't fit: dog, 1, 2, 3 The number "1" does not fit with the other three words because the other three words are all types of animals, while "1" is a number.

Works fine for me:

> The word "dog" does not fit with the other words in the list because "1", "2", and "3" are numbers, while "dog" is a type of animal. The other words in the list are all examples of numbers, whereas "dog" is not a number.