3 ms·

The funny thing is though your point here is working against your argument. You are saying that GPT is doing a poor job of showing it is a great artificial int

The funny thing is though your point here is working against your argument. You are saying that GPT is doing a poor job of showing it is a great artificial intelligence. But nobody is saying that it has infallible intelligence. It is making the same mistake that now two different humans have made in the same situation. Both me and OP above. It is failing in the same way that humans fail at this problem! By pattern matching and assuming it's the goat proble! You're arguing that gpt 4 isn't a sound logic engine and not are most people. A tool trained on human input, when given this problem is making similar mistakes that we make. Are there a set of people that would get this problem right? Yup. Are there also a set of people that would make this exact mistake? Yup. Your upset that is behaving like the "wrong" group of humans. Your thinking of it as an expert. Instead think of it as a result smart and well read high school student. There are things you can delegate to it that it will do well. But you also need to double check its work as it will make mistakes.

the_af 4y ago
I don't think this is it. This is not a failure mode for humans. No human (*) will forget in such a short problem statement that a vegan wolf cannot be left alone with a cabbage; humans will instead forget one of the states is invalid after trying some combinations. GPT's failure mode is only possible if it doesn't understand the meaning of the (apparently nonsensical) combination of words "vegan wolf". Humans don't pattern match like this; humans are capable of understanding the meaning of a combination of words they've never encountered before. (*) you know what I mean with "no human". Of course, there are deranged people or whimsical people or people who just will say about anything.
- BoiledCabbage 4y ago
  > No human (*) will forget in such a short problem statement that a vegan wolf cannot be left alone with a cabbage; Two humans in this thread just read the solution and thought it was correct. Me being one of them. Another further down in the thread manually wrote up a solution making the exact same mistake. I think you want things to be different, but they're not. Your answering how you'd think humans would respond, not how people actually respond. Does it mean everyone made that same mistake? No, but I bet a bunch did.
  - the_af 4y ago
    > Two humans in this thread just read the solution and thought it was correct. My guess is that they just skim read and missed what ChatGPT actually wrote, it's not that they misunderstood what "vegan wolf" means [1]. On the other hand, you cannot skim read what you are writing yourself, that's not how the mind works. The gist of the problem here is that, unlike a human, ChatGPT doesn't understand the words it generates, which leads to hilarious results. As another example, look at the "debugging" of GPT-4's assumptions someone posted in a sibling comment: it "knows" the vegan wolf will eat plant-based food and it "knows" a cabbage is a plant, yet it "thinks" the wolf "will not harm the cabbage"... which is a misunderstanding no human will make (if they know what "vegan" and "cabbage" mean). This doesn't happen in a long chain of reasoning (where a human can lose the line of thought) but in very short paragraphs, one right after the other! This failure mode requires not understanding the individual assumptions, which prevents GPT from making the connection. I was asked for an error that showed GPT misunderstanding something no person would, and I did. [1] question for you: did you think the wrong solution was right because you thought a vegan wolf cannot eat the cabbage (let me bet this is NOT what crossed your mind) or because the person who posted it made it look as if it was the right solution and you skim read it without paying attention, assuming "this person said it's right and it's posting it as a rebuttal, so it's likely right" (this is my bet)? If the latter, this failure mode is not one of misunderstanding what "vegan wolf" means (which is what debugging GPT's process shows), but one of very human laziness/jumping to conclusions. Do note this cannot happen when you write the solution yourself!
- astrange 4y ago
  You can try and "debug" it by asking: -- Consider this problem: a man must cross a river and carry his belongings to the other shore, but the boat can only carry the man and one other item. His belongings are a cabbage, a vegan wolf, and a meat-eating goat. What are the individual beings in this problem, what are they motivated to do, and how would they behave once left alone? -- Of course, it doesn't necessarily work because the model is not actually guaranteed to explain itself and may fabricate anything it likes. But ignoring that, we find that it doesn't think vegan wolves eat cabbages, that the meat-eating goat is an obligate carnivore, or that cabbages are animate objects. So it could be solving this logically considering its different assumptions. GPT-3.5: -- The cabbage is an inanimate object and does not have any motivations or behaviors. The vegan wolf is motivated to not harm any living beings, as it does not eat meat. If left alone with the goat, it would not harm the goat and would likely avoid it. The meat-eating goat is motivated by its natural instinct to eat grass and other vegetation, as well as to protect itself from harm. If left alone with the cabbage, it would likely eat the cabbage. -- GPT-4: -- In this problem, there are four individual beings: Man: He is motivated to cross the river and safely transport his belongings (cabbage, vegan wolf, and meat-eating goat) to the other shore without any harm coming to them. Cabbage: As a plant, it doesn't have motivations or behavior. It's an inanimate object that the man needs to transport. Vegan Wolf: The vegan wolf is motivated to eat plant-based food and will not harm the meat-eating goat or the cabbage. However, it is still a wild animal, and it may become agitated if left alone for too long, potentially leading to unpredictable behavior. Meat-Eating Goat: This unusual goat is motivated by its need to eat meat. It will attempt to eat the vegan wolf if left alone with it. Since it is a meat-eater, it will not be interested in eating the cabbage. --
  - the_af 4y ago
    That's cool. This bit shows ChatGPT's lack of understanding, the mistake no human would make: > Vegan Wolf: The vegan wolf is motivated to eat plant-based food and will not harm the meat-eating goat or the cabbage It "knows" that a cabbage is a plant, and it "knows" the vegan wolf is "motivated to eat plant-based food", but it doesn't know the wolf will eat the cabbage! It solves a different problem than the one posed precisely because it doesn't understand the meaning of the words.

the_af 4y ago

I don't think this is it. This is not a failure mode for humans. No human (*) will forget in such a short problem statement that a vegan wolf cannot be left alone with a cabbage; humans will instead forget one of the states is invalid after trying some combinations. GPT's failure mode is only possible if it doesn't understand the meaning of the (apparently nonsensical) combination of words "vegan wolf". Humans don't pattern match like this; humans are capable of understanding the meaning of a combination of words they've never encountered before. (*) you know what I mean with "no human". Of course, there are deranged people or whimsical people or people who just will say about anything.

BoiledCabbage 4y ago
> No human (*) will forget in such a short problem statement that a vegan wolf cannot be left alone with a cabbage; Two humans in this thread just read the solution and thought it was correct. Me being one of them. Another further down in the thread manually wrote up a solution making the exact same mistake. I think you want things to be different, but they're not. Your answering how you'd think humans would respond, not how people actually respond. Does it mean everyone made that same mistake? No, but I bet a bunch did.
- the_af 4y ago
  > Two humans in this thread just read the solution and thought it was correct. My guess is that they just skim read and missed what ChatGPT actually wrote, it's not that they misunderstood what "vegan wolf" means [1]. On the other hand, you cannot skim read what you are writing yourself, that's not how the mind works. The gist of the problem here is that, unlike a human, ChatGPT doesn't understand the words it generates, which leads to hilarious results. As another example, look at the "debugging" of GPT-4's assumptions someone posted in a sibling comment: it "knows" the vegan wolf will eat plant-based food and it "knows" a cabbage is a plant, yet it "thinks" the wolf "will not harm the cabbage"... which is a misunderstanding no human will make (if they know what "vegan" and "cabbage" mean). This doesn't happen in a long chain of reasoning (where a human can lose the line of thought) but in very short paragraphs, one right after the other! This failure mode requires not understanding the individual assumptions, which prevents GPT from making the connection. I was asked for an error that showed GPT misunderstanding something no person would, and I did. [1] question for you: did you think the wrong solution was right because you thought a vegan wolf cannot eat the cabbage (let me bet this is NOT what crossed your mind) or because the person who posted it made it look as if it was the right solution and you skim read it without paying attention, assuming "this person said it's right and it's posting it as a rebuttal, so it's likely right" (this is my bet)? If the latter, this failure mode is not one of misunderstanding what "vegan wolf" means (which is what debugging GPT's process shows), but one of very human laziness/jumping to conclusions. Do note this cannot happen when you write the solution yourself!
astrange 4y ago
You can try and "debug" it by asking: -- Consider this problem: a man must cross a river and carry his belongings to the other shore, but the boat can only carry the man and one other item. His belongings are a cabbage, a vegan wolf, and a meat-eating goat. What are the individual beings in this problem, what are they motivated to do, and how would they behave once left alone? -- Of course, it doesn't necessarily work because the model is not actually guaranteed to explain itself and may fabricate anything it likes. But ignoring that, we find that it doesn't think vegan wolves eat cabbages, that the meat-eating goat is an obligate carnivore, or that cabbages are animate objects. So it could be solving this logically considering its different assumptions. GPT-3.5: -- The cabbage is an inanimate object and does not have any motivations or behaviors. The vegan wolf is motivated to not harm any living beings, as it does not eat meat. If left alone with the goat, it would not harm the goat and would likely avoid it. The meat-eating goat is motivated by its natural instinct to eat grass and other vegetation, as well as to protect itself from harm. If left alone with the cabbage, it would likely eat the cabbage. -- GPT-4: -- In this problem, there are four individual beings: Man: He is motivated to cross the river and safely transport his belongings (cabbage, vegan wolf, and meat-eating goat) to the other shore without any harm coming to them. Cabbage: As a plant, it doesn't have motivations or behavior. It's an inanimate object that the man needs to transport. Vegan Wolf: The vegan wolf is motivated to eat plant-based food and will not harm the meat-eating goat or the cabbage. However, it is still a wild animal, and it may become agitated if left alone for too long, potentially leading to unpredictable behavior. Meat-Eating Goat: This unusual goat is motivated by its need to eat meat. It will attempt to eat the vegan wolf if left alone with it. Since it is a meat-eater, it will not be interested in eating the cabbage. --
- the_af 4y ago
  That's cool. This bit shows ChatGPT's lack of understanding, the mistake no human would make: > Vegan Wolf: The vegan wolf is motivated to eat plant-based food and will not harm the meat-eating goat or the cabbage It "knows" that a cabbage is a plant, and it "knows" the vegan wolf is "motivated to eat plant-based food", but it doesn't know the wolf will eat the cabbage! It solves a different problem than the one posed precisely because it doesn't understand the meaning of the words.