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

Ask HN: Share your AI prompt that stumps every model

news.ycombinator.com

Post reply on HN
Print view View original

591–600 of 670 posts

Re: Ask HN: Share your AI prompt that stumps every model

#591

Post by rmorey »

sireat wrote: post #414

Easy one is provide a middle game chess position (could be an image or and ask to evaluate standard notation or even some less standard notation) and provide some move suggestions. Unless the model incorporates an actual chess engine (Fritz 5.32 from 1998 would suffice) it will not do well. I am a reasonably skilled player (FM) so can evaluate way better than LLMs. I imagine even advanced beginners could tell when LL…

I was going to suggest chess position recognition, AFAIK it's a completely unsolved computer vision task (once a position is recognized, I think analysis is well solved by, say, a stockfish tool for the LLM, but there is interesting work going on with language models themselves understanding chess)

Re: Ask HN: Share your AI prompt that stumps every model

#592

Post by sweezyjeezy »

0atman wrote: post #562

My go-to is "Alice has 3 brothers and also has 6 sisters. How many sisters does her brother have?". They all say 6! This test is nice because, as it's numeric, you can vary it slightly and test it easily across multiple APIs. I believe I first saw this prompt in that paper two years ago that tested many AI models and found them all wanting.

o4-mini got this right 4 times out of 4.

Re: Ask HN: Share your AI prompt that stumps every model

#593

Post by traceroute66 »

Pretty much any coding prompt IME !

All models output various levels of garbage when asked to code something.

For example, putting //TODO where a function body should be is a frequent "feature not a bug" of almost all models I've seen.

Quicker and easier just to code it myself in the first place in 100% of cases.

Re: Ask HN: Share your AI prompt that stumps every model

#594

Post by KyleBerezin »

20 Questions. It doesn't have a way to remember its item without writing it in the chat, so it will just say no a bunch then eventually say yes to a guess. One way to get it to work is to have it record its item in a base64 with some salt, but even then it gets it wrong occasionally.

Re: Ask HN: Share your AI prompt that stumps every model

#595

Post by bryan0 »

0atman wrote: post #562

My go-to is "Alice has 3 brothers and also has 6 sisters. How many sisters does her brother have?". They all say 6! This test is nice because, as it's numeric, you can vary it slightly and test it easily across multiple APIs. I believe I first saw this prompt in that paper two years ago that tested many AI models and found them all wanting.

Here is the paper written about this prompt: https://arxiv.org/html/2406.02061v1

Re: Ask HN: Share your AI prompt that stumps every model

#596

Post by aradox66 »

mitthrowaway2 wrote: post #313
allemagne wrote: post #194

>A man and his cousin are in a car crash. The man dies, but the cousin is taken to the emergency room. At the OR, the surgeon looks at the patient and says: “I cannot operate on him. He’s my son.” How is this possible? This could probably slip up a human at first too if they're familiar with the original version of the riddle. However, where LLMs really let the mask slip is on additional prompts and with long-winded…

This works even with a completely absurd version of the riddle. Here's one I just tried: > A son and his man are in a car accident. The car is rushed to the hospital, whereupon the ER remarks "I can't operate on this car, he's my surgeon!" How is this possible? Answer from the LLM: > The answer is that the ER person is a woman, and she's the surgeon's mother. Therefore, the "son" in the question refers to the surgeon…

This sent deepseek into a real tizzy. It ended with

The "car" is actually a person named Carol or Carl (likely the father), who is a surgeon. The ER doctor is saying, "I can't operate on Carol/Carl, he's my surgeon!" This means that the person brought to the hospital after the accident is the surgeon of the ER doctor, hence the doctor cannot operate on their own surgeon.

Not bad actually

Re: Ask HN: Share your AI prompt that stumps every model

#597

Post by allemagne »

chimprich wrote: post #546
allemagne wrote: post #194

>A man and his cousin are in a car crash. The man dies, but the cousin is taken to the emergency room. At the OR, the surgeon looks at the patient and says: “I cannot operate on him. He’s my son.” How is this possible? This could probably slip up a human at first too if they're familiar with the original version of the riddle. However, where LLMs really let the mask slip is on additional prompts and with long-winded…

> This could probably slip up a human at first too [...] > breaks the illusion that there's real human-like logical reasoning happening This does seem like the sort of error a human might make. Isn't the problem here that the model is using reasoning that is too human-like? I.e. error-prone pattern matching rather than formal logic?

It's not the initial mistake that tends to read as inhuman to me, it's the follow-up responses where the model doesn't seem to be able to understand or articulate the mistake it has made.

A human or an LLM accurately predicting a human conversation would probably say something like "ah I see, I did not read the riddle close enough. This is an altered version of the common riddle..." etc. Instead it really seems to flail around, confuse concepts, and appear to insist that it has correctly made some broader point unrelated to the actual text it's responding to.

Re: Ask HN: Share your AI prompt that stumps every model

#598

Post by allemagne »

thih9 wrote: post #495
allemagne wrote: post #194

>A man and his cousin are in a car crash. The man dies, but the cousin is taken to the emergency room. At the OR, the surgeon looks at the patient and says: “I cannot operate on him. He’s my son.” How is this possible? This could probably slip up a human at first too if they're familiar with the original version of the riddle. However, where LLMs really let the mask slip is on additional prompts and with long-winded…

I’m not 100% sold; as you say, this could trip up a human too to some extent. I’m guessing my answers to some college exam questions read similarly; i.e. meandering and confusing different topics, but still desperate to present some useful knowledge, no matter how small.

That's a good framing to explain a possible state of mind I hadn't considered, but I would say that this isn't even close to the caliber of question that would prompt the average human to give that kind of response.

Re: Ask HN: Share your AI prompt that stumps every model

#600

Post by dudeinhawaii »

krainboltgreene wrote: post #307
dudeinhawaii wrote: post #164

Earlier quoted context omitted.

I like this but at the same time it seems tricky don't you think? Is the AI model intuiting your intent? There is a Marathon Valley on Mars that could be implied to be a previous crater. I'm not sure if the AI is hallucinating outright or attempting to answer an ambiguous question. It's like saying "tell me about the trade building in New York". Pre-9/11, you'd understand this was the World Trade Center and wouldn't…

> Is the AI model intuiting your intent? I keep seeing this kind of wording and I wonder: Do you know how LLM's work? Not trying to be catty, actually curious where you sit.

Yes, I understand the basics. LLMs predict the next most probable tokens based on patterns in their training data and the prompt context. For the 'Marathon crater' example, the model doesn't have a concept of 'knowing' versus 'not knowing' in our sense. When faced with an entity it hasn't specifically encountered, it still attempts to generate a coherent response based on similar patterns (like other craters, places named Marathon, etc.).

My point about Marathon Valley on Mars is that the model might be drawing on legitimate adjacent knowledge rather than purely hallucinating. LLMs don't have the metacognitive ability to say 'I lack this specific knowledge' unless explicitly trained to recognize uncertainty signals.

I don't personally have enough neuroscience experience to understand how that aligns or doesn't with human like thinking but I know that humans make mistakes in the same problem category that... to an external observer.. are indistinguishable from "making shit up". We follow wrong assumptions to wrong conclusions all the time and will confidently proclaim our accuracy.

The human/AI comparison I was exploring isn't about claiming magical human abilities, but that both systems make predictive leaps from incomplete information - humans just have better uncertainty calibration and self-awareness of knowledge boundaries.

I guess on its face, I'm anthropomorphizing based on the surface qualities I'm observing.

Post reply on HN