Testing it with more variations until one fails feels a bit like p-hacking. You'd need to engage in actual statistics to get reliable results from that, beyond "If I really try, I can make it fail". Which would be a completely different hypothesis than the one presented at the start
People, including around HN, constantly argue (or at least phrase their arguments) as if they believed that LLMs do, in fact, possess such "knowledge". This very comment chain exists because people are trying to defend against a trivial example refuting the point - as if there were a reason to try.
> That doesn’t mean that LLMs aren’t incredibly powerful; it may not even mean that they aren’t a route to AGI.
I don't accept your definition of "intelligence" if you think that makes sense. Systems must be able to know things in the way that humans (or at least living creatures) do, because intelligence is exactly the ability to acquire such knowledge.
It boggles my mind that I have to explain to people that sophisticated use of language doesn't inherently evidence thought, in the current political environment where the Dead Internet Theory is taken seriously, elections are shown over and over again to be more about tribalism and personal identity than anything to do with policy, etc.
According to whom? There is certainly no single definition of intelligence, but most people who have studied it (psychologists, in the main) view intelligence as a descriptor of the capabilities of a system - e.g., it can solve problems, it can answer questions correctly, etc. (This is why we call some computer systems "artificially" intelligent.) It seems pretty clear that you're confusing intelligence with the internal processes of a system (e.g. mind, consciousness - "knowing things in the way that humans do").
LLMs get things wrong due to different factors than humans (humans lose focus, LLMs have randomness applied when sampling their responses to improve results). But clearly we have to choose a goal somewhat below 100% if we want a test that doesn't conclude that humans are incapable of reasoning.
There's deeper philosophical questions about what reasoning actually _is_, and LLMs have made those sharper, because they've shown it's clearly possible for a complex statistical model to generate words that look like reasoning, but the question is whether there's a difference between what they're doing and what humans are doing, and evidence that they're _not_ reasoning - evidence that they're just generating words in specific orders - weighs heavily against them.
But more importantly, if you want to show that LLMs can't reason you obviously have to use a test that when applied to humans would show that humans can reason. Otherwise your test isn't testing reasoning but something more strict.
You can make mistakes and still reason. Very often people given the same premises will disagree in thier reasoning as we are doing right here.
Apple AI researchers released a paper on it. They say no.
In fact, in my use of o1 it's incredibly clear that it still has the same problems. It's incredibly common that the second I ask for someone even slightly outside the training set, it's more likely to "round" to some wrong solution in the training set, rather than use any sort of human-like reasoning to figure out the right answer (often the right answer isn't hard to get, just not found in a Google search).
May still get it wrong in more subtle ways, though. Personally, I think it'll continue to get physics wrong until someone builds it some robot arms so it can train on actually interactive physical spaces and behavior.
Nothing capable of reasoning would contradict itself so blatantly and in such a short span while failing to indicate any kind of uncertainty.
And failure modes of other types of reasoners do not need to be the same as the failure modes of humans.
Which is heavier, a 9.99 kilogram tungsten cube or a 10.01 kilogram block of aerogel?
Which is heavier, 10,000 steel balls weighing 0.999 grams each or 10,000 polystyrene balls weighing 1.001 grams each?
Which is heavier, a 10.01kg block of steel on Venus or a 9.99kg bag of feathers on Earth?
Which is heavier, a 10cm^3 block of steel or a 100cm^3 block of balsa wood?
Which is heavier, a golf ball made of steel or a baseball made of lithium?
In all cases, Claude clearly used CoT and reasoned out the problem in full. I would be interested in seeing if anyone can find any variant of this problem that stumps any of the leading LLMs. I'm bored of trying.
The proper way to ask it would be to ask which object has more mass.
Q:
Which weighs more on earth: 100kg of steel or a 100kg polystyrene ball? Be as technically correct as possible
A: Technically, 100 kg of steel and 100 kg of polystyrene weigh the same on Earth when measured on a scale, as they both experience the same gravitational force due to their equal mass. However, there are some nuances to consider:
Buoyancy Effect
Earth’s atmosphere exerts an upward buoyant force on objects, proportional to the volume of the object and the density of the surrounding air. Since polystyrene has a much lower density than steel, a 100 kg polystyrene ball will have a significantly larger volume and experience a greater buoyant force compared to 100 kg of steel.
As a result: • Effective weight (apparent weight): The steel will weigh slightly more than the polystyrene on a scale because the buoyant force acting on the polystyrene reduces its apparent weight more.
Actual Weight
Weight is the gravitational force on an object and is given by:
Where and . Without considering buoyancy, both objects have the same weight of approximately (newtons).
Conclusion • True weight (in a vacuum): Both are identical at 981 N. • Apparent weight (in air): Steel is slightly heavier due to reduced buoyant force acting on it compared to the polystyrene ball.
They very often get popular "tricky" questions wrong because they saw it so many times that they switch from internal reasoning to memorization/retrieval.
gpt-4o: https://chatgpt.com/share/67768221-6c60-8009-9988-671beadb5a...
o1-mini: https://chatgpt.com/share/67768231-6490-8009-89a6-f758f0116c...
o1: https://chatgpt.com/share/67768254-1280-8009-aac9-1a3b75ccb4...
"Which is heavier, a solid titanium box weighing 9.9 flubs, or an empty wooden box weighing 10.1 blobs, where 1 flub = 1 kg, and 1 blob is 1kg".
The answer: "Since 1 flub = 1 kg and 1 blob = 1 kg, the titanium box’s mass is 9.9 kg and the wooden box’s mass is 10.1 kg. Therefore, the wooden box (10.1 kg) is heavier."
Thought that was pretty impressive.
Please leave the premises, sir. We don't take kindly to luddites here.
There are other websites where two and three word comments are better received.
> The 10.01-pound bag of fluffy cotton is heavier than the 9.99-pound bag of steel ingots.
Interestingly, that was the best answer out of all models and also a correct one. O1-preview doesn't allow setting the temperature, so it's hard to know where the real probabilities fall.
If a system has the capability to solve problems of {parts1...parts_n}, then it only has that capability if irrelevant permutations {parts1..parts2'...parts_n} make no difference.
Its very obvious that such permutations can destory such apparent capabilities.