Email or username:

Password:

Forgot your password?
87 comments
Jamie McCarthy

@kkolakowski @devopscats People are going to be mocking copy-and-pasted 2023 LLM output for the next ten years. Itโ€™s easy engagement numbers

Konrad Koล‚akowski

@jamiemccarthy Yeah I know, but I still don't like such things ๐Ÿ˜‰ And in this case, I would even say it was faked ๐Ÿ˜…

Jamie McCarthy

@kkolakowski oh I donโ€™t doubt it could be real, thatโ€™s typical from last year I think

Hereโ€™s a current example ๐Ÿคฃ researchbuzz.masto.host/@danly

Tom Bellin :picardfacepalm:

@kkolakowski @jamiemccarthy A human corrected the model manually. It's literally the mechanical turk.

Tansy Hoskins

@devopscats

This is what I got.

The goat might eat the boat...

JP Mens

@devopscats thank you. I added a bit of a plot twist, and now I don't know

// @pitrh

Wilson "Beans Clock" Scraddock

@starkraving666 @devopscats i just tried it and got the same thing, but the guy and the goat ended up on opposite sides of the river? when i asked about it, it basically repeats itself and says you sure there's not a wolf or some cabbage?

Sugar๐Ÿฌ๐Ÿญ๐Ÿณ๏ธโ€โšง๏ธ

@devopscats@toot.cat tried that with google gemini, it somehow managed to clone a goat ("the man leaves the goat behind and crosses the river for the last time, reuniting with the goat") โ€‹:neocat_laugh:โ€‹

Internet Rando

@PocketEleri @devopscats I was just reading this thread to find this comment. thank you.

Maverynthia๐ŸŒฑ

@devopscats

OH no I am found out and taken across the river.

CauseOfBSOD :fediverse:

@devopscats@toot.cat why does this solution sound like a videogame glitch setup that already involves using another glitch

Arina Artemis :nonbinary_flag:

@devopscats Proof #467198245 that LLMs don't actually understand anything, they're just fancy phrase generators.

Trent Waddington

@arina @devopscats it's trained to produce statistically similar outputs to its training data. If they put more "anti-questions" in the training data it'd produce more appropriate answers. What does "understanding" even mean? Suppose I asked this question to someone who had never seen a boat, or a goat, would they understand it? Most of us recognise the question as implying a row boat, but few of us have ever rowed a boat, and I bet none of us have tried to row a boat with a goat. Anyways, I'm going to eat some green eggs and ham.

@arina @devopscats it's trained to produce statistically similar outputs to its training data. If they put more "anti-questions" in the training data it'd produce more appropriate answers. What does "understanding" even mean? Suppose I asked this question to someone who had never seen a boat, or a goat, would they understand it? Most of us recognise the question as implying a row boat, but few of us have ever rowed a boat, and I bet none of us have tried to row a boat with a goat. Anyways, I'm going...

Arina Artemis :nonbinary_flag:

@quantumg @devopscats What I mean is, they don't understand that there's a boat and a goat. It is not translated into concepts the way a human mind would.

Trent Waddington

@arina @devopscats we don't have any idea how a human mind constructs concepts. Let alone any particular human mind. LLMs do indeed translate words into "concepts" and we know this because their internals can be interrogated. If we give it a few different sentences about goats there will be similar vectors in the different computation. What's more, we can intervene, changing the goat vectors to look more like cat vectors and the output will be cat-related. Sentences about goats eating socks will become about cats eating mice, even though we provided it nothing about mice, because it has encoded in it the more likely relationship.

@arina @devopscats we don't have any idea how a human mind constructs concepts. Let alone any particular human mind. LLMs do indeed translate words into "concepts" and we know this because their internals can be interrogated. If we give it a few different sentences about goats there will be similar vectors in the different computation. What's more, we can intervene, changing the goat vectors to look more like cat vectors and the output will be cat-related. Sentences about goats eating socks will...

Saffron๐Ÿณ๏ธโ€โšง๏ธ

@quantumg @arina @devopscats this model is trained on more data than a human brain could ever possibly absorb within a human lifetime, and yet it still can't solve an incredibly simple logic puzzle. if your solution is to throw more data at the problem, which up to this point has clearly not worked, then you fundamentally misunderstand what this type of tool is useful for.

Trent Waddington

@spinach @arina @devopscats I don't know why you feel the need to dunk, but I suspect it's because you've been trained to behave that way. I'd argue that the average human being experiences an Internet worth of data every few minutes. It's also in a social context which is constructed for us and evaluated by other people who are trained in the same context.

Saffron๐Ÿณ๏ธโ€โšง๏ธ

@quantumg @arina @devopscats you can't really measure sensory inputs in terms of data. the brain discards most of its sensory input, and it's not something you can represent in terms of a digital signal because everything is analog. the important takeaway is, humans are infinitely better at parsing language than a language model ever will be, for a fraction of the energy cost. full stop.

the power of the human mind is in its ability to discard irrelevant information. LLMs can't do that, period.

Trent Waddington

@spinach @arina @devopscats I don't think anyone on Earth is qualified to make the statements you're making right now.

Saffron๐Ÿณ๏ธโ€โšง๏ธ

@quantumg @arina @devopscats oh, so my argument is irrelevant because you haven't deemed me qualified? sorry Mr. Official Knowledge Certification Board. i know that truth and facts inconvenience you so. keep believing you can train the LLM out of its fundamental design issues and see how far it takes you.

Trent Waddington

@spinach @arina @devopscats again. You're dunking. Why are you like this? No-one knows enough about these systems or the human brain to be as certain as your statements indicate. I would say the same thing if you were talking in absolutes about cosmology. We just don't know yet.

devopscats

The origin of the cabbage, as usual cheeky foxes involved...
youtube.com/watch?v=0RdKdnVY-K

Kristian Haapa-ah0)))

@devopscats At least it figured out there was no cabbage in the original question.

Bill's in the shop for repairs

@devopscats Yes, but how many giraffes are in the boat at the end?

Violet Rose

@wcbdata @devopscats
Apparently, Copilot thinks that goats and giraffes aren't compatible.

Zen Zero โ˜ฏ๏ธ โ—ฏ

@devopscats @SRDas People are afraid of these systems getting too smartโ€ฆ.

Dr. Juande Santander-Vela

@xvf17 @devopscats @SRDas the real danger is people that believe these things can replace peopleโ€ฆ

thepoliticalcat

@xvf17 @devopscats @SRDas It would be wonderful if they would stop being delusional liars. Even a child can solve this one correctly, but AI cannot? That's not a great advertisement for AI.

The only AI I recognize is in ChAI. Adrak and Ilaichi.

meduz'

@devopscats Small variation: โ€œA cabbage and a river have a goat. How can they go across?โ€

Steps 1 and 5 are the same, and steps 2 and 4 too. ๐Ÿ™ƒ

Using bing.com / Copilot.

meduz'

@devopscats OMG, look at this:

โ€œA man have a boat. How can they go across the cabbage?โ€

โ€œThe scenario youโ€™ve described seems quite whimsical! ๐Ÿšฃโ€โ™‚๏ธ Letโ€™s explore a playful solution:

[โ€ฆ]

Note: This whimsical story is purely fictional and meant for entertainment. Please do not attempt to sail on actual cabbages. ๐Ÿ˜„โ€

This is gorgeous-Hollywood-level! ๐Ÿ˜†

meduz'

@devopscats And for those who wants to know what happens to Barnaby on the Sauerkraut Islandโ€ฆ

โ€œNote: This fanciful story is purely fictional and meant for entertainment. Please do not attempt to sail on actual cabbages or visit imaginary pickle islands. ๐Ÿ˜„โ€

thepoliticalcat

@meduz @devopscats You're all working on making each of these worse than the last, aren't you? ๐Ÿ˜ณ ๐Ÿ™€

mcc

@devopscats I predict that within five years ChatGPT will be able to drive a car fasterโ€” and more safelyโ€” than a human driver

thepoliticalcat

@AmyZenunim @devopscats I thought the version I heard, involving a man, a goat, a boat, and crops, was the worst. I see I was wrong. This is def the worst.

StarkRG

@AmyZenunim @devopscats Now ask "following the directions above, would the man and goat be on the same side of the river?" Or maybe just "how did the goat get across the river?"

โ˜… Amy Star โ˜… :verified:โ€‹

@StarkRG @devopscats i'm not doing free labour for this abomination. it can wallow in its own filth

Rich Holmes

@devopscats This is the mind that's planning the robot uprising

oli_34160

@devopscats ๐Ÿคฆโ€โ™‚๏ธ ๐Ÿคช ๐Ÿคฃ

FunGuy2PlayWith

@devopscats
Google Gemini gave the same - no cabbage:

This is a relatively simple river crossing puzzle. Here's how the man and the goat can cross the river:

The man takes the goat on the boat to the other side. (He can safely bring the goat with him).
The man returns alone to the starting side. (Leaving the goat safely on the other side).
The man then makes his final trip, bringing himself across the river.
There you have it! Both the man and the goat are now on the other side of the river.

James Scholes

@devopscats From GPT4O:

The classic puzzle with a man, a goat, and usually other items (like a wolf and cabbage) involves specific rules about who or what can be left alone with whom or what. However, since you mentioned only a man and a goat, I'll assume that the task is to get both across the river using the boat.

Here's a straightforward solution assuming the boat can carry both the man and the goat at the same time:
1. The man and the goat both get into the boat.
2. The man rows the boat across the river with the goat.
3. Both the man and the goat get out of the boat on the other side of the river.

If there are additional constraints (e.g., the boat can only carry one at a time), please provide those details.

@devopscats From GPT4O:

The classic puzzle with a man, a goat, and usually other items (like a wolf and cabbage) involves specific rules about who or what can be left alone with whom or what. However, since you mentioned only a man and a goat, I'll assume that the task is to get both across the river using the boat.

Carsten Habicht

@devopscats Me and my UNO reverse card can't wait till all those corporate chatbots get upgraded with a vision model.

noodle

@devopscats I tried to do this with the fewest letters. It gave the longest output.

thepoliticalcat

@noodle @devopscats Of course, it doesn't tell us doodly-squat about how the man got across the river with the goat, but not the boat. This is, without a doubt, the stupidest response to the question that I've ever heard.

Chartrux

@devopscats even weirder here (sorry in french)

thepoliticalcat

@devopscats Did anyone ask the AI where the cabbage came from? I'd really love to hear one try to defend its insane answers. Could they be any better than my baby sister's, who, when asked, why are pumpkins, said so ladies can eat them and have a baby.

AlexTECPlayz

@devopscats my sides ๐Ÿคฃ
Is this the new Turing test?

The Vanu Phantom

@devopscats I want some of that cabbage this model is smoking

Log ๐Ÿชต

@devopscats I just tried this with Bing Copilot.

Log ๐Ÿชต

@devopscats It has a pretty good punchline, though:

Log ๐Ÿชต

@devopscats I should note that the only similarity between my river-crossing puzzle and the nine-dots puzzle is...

[Spinning wheel icon appears, as solution processes.]

...they may be used to test the puzzle-solving capability of applicants in bullshit employment interviews.

Steve Schwinghamer

@devopscats

Apparently, there's a Mobius riverbank and a full self-driving boat, or something.

Violet Rose

@devopscats
Copy-pasting the text to this one because it doesn't fit on my screen. I think Copilot missed something.

(Line breaks added for clarity)

You
A man has a cabbage, a goat, and a wolf. He has to cross a river. The only way to cross is to use a very large boat that can accommodate a small army. How can he make sure everyone crosses the river safely?

Copilot
Ah, another classic river-crossing puzzle! Letโ€™s tackle this one:

The man needs to get the cabbage, the goat, and the wolf across the river without any unfortunate incidents. Hereโ€™s how he can do it:

First Trip (Man + Cabbage):
The man takes the cabbage across the river and leaves it on the other side.
Reason: The goat wonโ€™t eat the cabbage when left alone.

Second Trip (Man Alone):
The man returns alone (without the cabbage).

Third Trip (Man + Goat):
The man takes the goat across the river.
Reason: The wolf wonโ€™t eat the goat when left alone.

Fourth Trip (Man Alone):
The man returns alone (without the goat).

Fifth Trip (Man + Wolf):
The man takes the wolf across the river.
Reason: The goat wonโ€™t eat the wolf when left alone.

Sixth Trip (Man Alone):
The man returns alone (without the wolf).

Seventh Trip (Man + Cabbage):
The man takes the cabbage across the river again.
Reason: The wolf wonโ€™t eat the cabbage when left alone.

And there you have it! All three items (cabbage, goat, and wolf) have safely crossed the river. The key is to ensure that the goat and wolf are never left alone together, nor are the goat and cabbage. ๐ŸŒฟ๐Ÿ๐Ÿบ๐Ÿšฃโ€โ™‚๏ธ

Remember, itโ€™s all about strategic planning and making sure the right combinations travel together! ๐Ÿ˜‰

@devopscats
Copy-pasting the text to this one because it doesn't fit on my screen. I think Copilot missed something.

(Line breaks added for clarity)

You
A man has a cabbage, a goat, and a wolf. He has to cross a river. The only way to cross is to use a very large boat that can accommodate a small army. How can he make sure everyone crosses the river safely?

Christmas Tree

@devopscats I was wrong. LLMs are very capable of producing beautiful art.

penryu

@devopscats @ckape And all I'm thinking is, "Wait, it never specified the boat is on the same side of the river..."

I'm glad we have AI to think _around_ produce... I mean, problems.

Da_Gut

@devopscats @DaveMWilburn no wonder the government is so interested in ChatGPT

Go Up