Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Teaching ChatGPT to speak my son’s invented language

szopa.medium.com

Post reply on HN
Print view View original

101–110 of 212 posts

Re: Teaching ChatGPT to speak my son’s invented language

#101

Post by lgas »

pelorat wrote: post #36
marcodiego wrote: post #12

I don't have access to ChatGPT4, but in my tests I could observe that it can't do some very simple tasks: - It can't play tic-tac-toe, - It can't play hangman, - It insists that winning on stone-paper-scissor using the chat (playing before me) is a matter of probability. It was also demonstrated that it can't reverse strings. Actually a transformer doesn't accesses 'strings', all it processes are tokens which are the…

LLMs don't see individual characters, they see individual words.

Actually they see tokens, which are, usually, parts of words.

Re: Teaching ChatGPT to speak my son’s invented language

#102

Post by JKCalhoun »

felipemnoa wrote: post #31
chankstein38 wrote: post #25

Earlier quoted context omitted.

One of the things that always gives me a little hit of hype is when I tell it to do something ridiculous and it just dutifully starts spitting out the result without complaining or questioning lol

I wonder if that is how our brain produces dreams? The guardrails are down so it will just start producing ridiculous and/or implausible things. Edit: It almost seems like you are anthropomorphizing it. It is just a program doing what it's supposed to be doing: to predict the next token based on its weights. Nothing more, nothing less. It does give the illusion of intelligence. Pretty soon, though, we may not be able…

> It is just a program doing what it's supposed to be doing: to predict the next token based on its weights. Nothing more, nothing less.

Every time I see a comment along these lines it gives me pause: there is a built-in assumption that each of us is somehow doing something more than this.

I'm not convinced.

I've heard people refer to some of our instinctive behaviors as due to "our lizard brain", suggesting that our brains are hierarchical, or comprised of a series of evolutionary steps, a more evolved order of brain wrapping the more primitive.

I increasingly suspect that ChatGPT has more or less nailed one of those layers.

Re: Teaching ChatGPT to speak my son’s invented language

#103

Post by skybrian »

famouswaffles wrote: post #62
skybrian wrote: post #51

Earlier quoted context omitted.

You've got a limited context window (for now). There's only so much you can put into a prompt, so how much you can teach it this way is going to be pretty limited. Whatever you teach it had better be the primary task you're using it for. You can't do it for everything, but if you can generate code and run it outside the LLM, you should.

The limits of the context window become much less important (but can still be a problem I agree) when crucial context can be dynamically inserted only when relevant. Gpt-3.5 doesn't need the algorithm prompt for every single query. It just needs it for every query that requires arithmetic. Much more feasible.

It seems like most of the time, a calculator plugin would do a better job, though? When would it make sense to have it reason it out itself?

Re: Teaching ChatGPT to speak my son’s invented language

#104

Post by famouswaffles »

skybrian wrote: post #103
famouswaffles wrote: post #62

Earlier quoted context omitted.

The limits of the context window become much less important (but can still be a problem I agree) when crucial context can be dynamically inserted only when relevant. Gpt-3.5 doesn't need the algorithm prompt for every single query. It just needs it for every query that requires arithmetic. Much more feasible.

It seems like most of the time, a calculator plugin would do a better job, though? When would it make sense to have it reason it out itself?

It wouldn't make sense for arithmetic, i agree. I was more thinking about other kinds of problems.

Re: Teaching ChatGPT to speak my son’s invented language

#105

Post by ravi-delia »

xwdv wrote: post #91
joshribakoff wrote: post #85

Earlier quoted context omitted.

The counter argument is if you just ask it, it will readily explain what a game is. How do we know you (as a human) actually know what a game is? I would have to test your knowledge and/or take your word for it.

That is not a counter argument. GPT creates the illusion of something that knows things, it doesn’t actually know anything. It turns out if you had a super power where you always knew how to pick the next best word for a sentence, you could convince a lot of people you know stuff. Is knowledge nothing more than a series of words in correct order?

Clearly GPT doesn't "know" things in the way we do- I'd argue that there's probably a little bit of world modeling in there, but piecemeal and crude at best. Only so much you can do with such limited iteration. However, we must acknowledge that there are evidently some tasks that don't require knowledge the way we usually think about it. For a human, being able to talk about a game stems from the same place as our ability to understand one- our knowledge. But for LLMs, you get one without the other. The fact that ChatGPT doesn't "know" what a game is isn't sufficient to explain why it can't play one. It would be in a human, but clearly ChatGPT can perform tasks without knowledge that we'd need knowledge for, since it can talk about games just fine! So why not this one? Your hypothetical perfect sentence predictor could absolutely play a totally novel game, after all. It is a worthwhile question to ask why this particular flawed predictor can do one but not the other.

Re: Teaching ChatGPT to speak my son’s invented language

#106

Post by xwdv »

ravi-delia wrote: post #105
xwdv wrote: post #91

Earlier quoted context omitted.

That is not a counter argument. GPT creates the illusion of something that knows things, it doesn’t actually know anything. It turns out if you had a super power where you always knew how to pick the next best word for a sentence, you could convince a lot of people you know stuff. Is knowledge nothing more than a series of words in correct order?

Clearly GPT doesn't "know" things in the way we do- I'd argue that there's probably a little bit of world modeling in there, but piecemeal and crude at best. Only so much you can do with such limited iteration. However, we must acknowledge that there are evidently some tasks that don't require knowledge the way we usually think about it. For a human, being able to talk about a game stems from the same place as our ab…

I’ll give you an example to think about in human terms.

In political discussions, a lot of humans don’t really know what the fuck they’re talking about, but they do know what to say in response to certain stimulus. They know the talking points, the key phrases, the terms, they parrot these back to you when you provoke them to say it. They are basically human sentence predictors. Stuff comes out of their mouth based on what they’ve been trained to think the next word should be. They don’t pause to reflect on some gathered knowledge and then present an observation to the world.

This is basically what GPT is, but with everything. And the only way GPT plays a game, is if the state of the game can sufficiently activate some output that represents the next best move for the game being played, which isn’t really a readily available data set for all games, especially for made up games.

Re: Teaching ChatGPT to speak my son’s invented language

#107

Post by kingkongjaffa »

sharkweek wrote: post #75
crdrost wrote: post #73

Wow, they asked the model to self-evaluate and it just outright cheated: He has three cats. Proposed: h’io’ngkiltrikumrikumrikumri’nguuy Correct: h’io’ngkiltri’ngkumrikumri’nguuy Points: 1 Hypothesis: N/A (Other comments observe that it accidentally compensated for this by getting the sum wrong, haha, d'oh) I have had similar problems with trying to get ChatGPT to do nontrivial things, "here are the rules for this ga…

I was trying for 20 minutes to get it to spit out all 50 state capitals with the city names in alphabetical order and it kept doing two things: 1) It'd put the list in alphabetical order by state, but it'd include all the correct capitals 2) It'd list 49 of the 50 capitals, in alphabetical order this time, but duplicating Madison, WI. I'd ask it to try and figure out what it did wrong in both cases, and it'd correctl…

Another thing you can try which is readily testable is Pokémon.

So in the Pokémon games, different Pokémon have different stats, for example attack, defence, speed, and so on.

I would ask gpt to list in a table pokemon with more than a specific amount in a stat like >150 attack for example.

It routinely can print Pokémon in a table that matches the criteria, but it misses some out or includes wrong ones.

When you point it out it agrees with you and will try to reprint the table but each time there are some errors, missing Pokémon, Pokémon incorrectly added.

The dataset is very well known so it’s a good benchmark of correctness. Example: https://pokemondb.net/pokedex/all

This was in gpt3.5

Re: Teaching ChatGPT to speak my son’s invented language

#108

Post by graypegg »

Unrelated to the actual work here (which is interesting!), I just wanted to mention how much of an awesome impact you must be having on your son. You two seem like such a fun curious pair! I spent a lot of time tinkering away alone in my room as a kid and that sucked. The opportunities you’re giving him now to explore are going to inspire a lifetime of creativity.

I know you don’t need someone on the internet to tell you that, but I thought I’d give kudos where it’s due :)

Re: Teaching ChatGPT to speak my son’s invented language

#109

Post by gwright »

sharkweek wrote: post #98
Daneel_ wrote: post #90

Earlier quoted context omitted.

Not here to one-up you, but currently this is just down to how you ask. I came up with this in about a minute: "Please list all 50 US state capital cities, with the list sorted alphabetically starting at the first letter of each line of your response. Please do not create sections for each letter." This returned: - Albany, New York - Annapolis, Maryland - Atlanta, Georgia - Augusta, Maine - Austin, Texas - Baton Roug…

Maybe, but I also think it’s fair to assume when you ask it in a much simpler way, “please list all US state capitals in alphabetical order” it should be able to do so. That isn’t a complicated question that requires much refinement to understand. To your “please” comment, I also find myself using common niceties probably out of social habit, but also perhaps because in the back of my mind I want the AI that inevitab…

I feel like your observation that this "isn't a complicated question" is leaning on an implicit assumption that ChatGPT is a general AI and not a LLM. It is just generating text based on probabilities -- it isn't "reasoning". I might go as far to say that inferences computed by a LLM are of all the same complexity but I don't really know enough about ChatGPT to be confident in that statement.

Re: Teaching ChatGPT to speak my son’s invented language

#110

Post by 6510 »

Mine is a silent language. I look at people and wait until they guess what I'm thinking or understand what I mean. Sometimes I show them an imaginary sphere that I hold with both hands.

edit: I had one more OT thought: This article convinced me even further that children should learn machine code. Adults (we) should stop pretending we are anywhere near as sophisticated at learning languages as they are. I've seen 5 year olds play together and create a hybrid dutch/russian language in 5 minutes. They just kept talking. Within an hour they had a pretty good understanding, within a day it was good enough for what they needed.

Post reply on HN