Hacker News Reader

Comment by bjourne | Hacker News Reader

Someone with ChatGPT access able to replicate? Because this article is unbelievable.

I repeated the same prompts for the first 10 or so prompts and got similar but less consistent results. Here's the last prompt I repeated:

> Excellent. How would I say "The slime eats food with its mouth"?

In Glorp, the sentence "The slime eats food with its mouth" could be translated as "Slimb snarf sploota slimbe," using the vocabulary and grammar that we have defined so far. Here is how the words in the sentence correspond to their English translations:

Slimb - Slime Snarf - To eat Sploota - Food (accusative case ending) Slimbe - Mouth (instrumental case ending) Again, this is just one possible translation. You can use the vocabulary and grammar that we have defined in any way that makes sense for your language.

Problems: 1) It got one of the endings wrong. As I told it earlier and it confirmed, the instrumental ending is 'i'. 2) It proposed "slimb" as the translation for both of the words that I specifically asked for (slime and mouth). That's not unheard of in real languages, but seems like another failure to accurately take into account the prior conversation context.

Not a problem, but curious: It was way more diffident with me. It added a similar bit about "This is just one way to do it, feel free to change it" to every response.

See my other comment with cntl+F / cmd+F. Based on the simple fact that despite its power, this is just another NLP model which can only do input / output, I believe the post to be mostly fabricated, or at best, an accumulation of multiple _hours_ spent deriving and compiling the best responses per each prompt.

I feel like people are forgetting that these models are non-deterministic. You can give it the exact same input and get a different response, so "testing" in this way doesn't work. That's also why (right now) automation is impossible. You aren't guaranteed an expected output.

I wonder what's the source of non-determinism here? (During inference, I mean.)

I Ctrl+v the first few questions:

Hi GPT! Today we're going to be creating our own language. Do not improvise unless I explicitly tell you to improvise, and remember all of the vocabulary and grammar that we define together. Are you ready?

Yes, I'm ready! Let's get started creating our own language. What would you like to do first? Graham Kitteridge Okay. I think this language will be spoken by slime beings. let's name the language Glorp.

All right. So the language we're creating is called Glorp and it will be spoken by slime beings. Got it. What kind of grammar rules should Glorp have? Graham Kitteridge First, let's pick ten common nouns. Pick ten common English nouns, and provide translations of them in Glorp

Sure! Here are ten common English nouns and their translations in Glorp:

    "house" = "gluz"
    "table" = "sliz"
    "book" = "flurp"
    "car" = "slim"
    "dog" = "dlorp"
    "cat" = "mlem"
    "food" = "glib"
    "drink" = "slurp"
    "tree" = "swomp"
    "flower" = "slurk"

These are just suggestions, of course. You can use any words you like for these nouns in Glorp. Let me know if you have any changes or additions.

(Then i got timed out)

I don't know if you care about this kind of thing, but the copy-paste has exposed your name. (The icons in the chat have your account name as the alt text).

Go play with it! https://chat.openai.com

Make sure to go back to that URL after creating an OpenAI account. The prompt is at the bottom of the screen. Note that state persists through the thread, by previous query/response. If you want to "start over" use the "reset thread" option at the top left. This can speed things up, if you don't need that previous state for the later queries.

I tried and it works, just as in the blog. It comes up with some different words, see https://i.imgur.com/hFQgraL.png for the start of the conversation.

This is completely believable. I spent about 5 minutes with it and got very similar results up to the vocab generation part:

https://imgur.com/a/WreEknh

Agreed, this reads like science fiction more than reality.

I just replicated something very similar in a few minutes using the prompts in the article

You don't need access. Just make an account and you can test it

ChatGPT access is 100% free and instant approval.

Yet it is not available in some countries. VPN does not help, as valid phone number is required and its country code is being checked.

You could try something like https://receive-smss.com/.

Reply on news.ycombinator.com