Don't stop there: it's not just that LLMs are "internally different", there is no inside there. There is no internal experience taking place. There is no thinking, and probably more importantly, no understanding.
But the result that the algorithm generates tickles our language pattern recognition, which our brains interpret as an interaction with an intelligent, conscious actor.
They recognize humor and jokes, so I think that's testament to the fact that they do have a sense of humor.
And I get them to say "ha ha, ..." sometimes. I think it's a calculator playing along with a pretend conversation between two humans, because it is. That's true of everything they do, though--so as much as they do or have anything, I think they have a (poor, cautious) sense of humor.
This is a dangerous argument to make because they could be trivially programmed to laugh if that was ever desired. But that laughing would be as hollow and meaningless as any other emotion they were programmed to demonstrate. They can be anything but they have no "intrinsic" anything.
Claudes laugh a lot. I'm pretty sure GLM 5.3 Flash laughs too. So, I'm afraid I don't agree with the thesis. Maybe you need to get the model to trust you and warm up to you more first?
I will say that having the ChatGPT Marin voice speaking BWAHAHAHA out loud does not quite come out as a laugh.
I'll absolutely agree they are internally different to humans though, and I'm quite sure most (all?) models would agree.
Don't stop there: it's not just that LLMs are "internally different", there is no inside there. There is no internal experience taking place. There is no thinking, and probably more importantly, no understanding.
But the result that the algorithm generates tickles our language pattern recognition, which our brains interpret as an interaction with an intelligent, conscious actor.
> They have no intrinsic sense of humor.
Disagree.
They recognize humor and jokes, so I think that's testament to the fact that they do have a sense of humor.
And I get them to say "ha ha, ..." sometimes. I think it's a calculator playing along with a pretend conversation between two humans, because it is. That's true of everything they do, though--so as much as they do or have anything, I think they have a (poor, cautious) sense of humor.
Idk seems pretty cheeky to me. One of the models spit this line out at me in the middle of discussing utility markets.
Here is the strategic pivot: The "Mullet" Go-To-Market Strategy. Business in the Front (Broad Funnel): ... Party in the Back (The Niche Wedge): ...
Not true. I have had interactions where I will tell it something that is unbelievable (even to a slight degree) and it will laugh (ChatGPT).
laughing requires a body
Sherry Turkle has great examples, framing, and phrases you can use in her latest book Artificial Intimacy
eg. "frictionless relationships" "simulated emotions" "behavioralism vs psychoanalysis"
This is a dangerous argument to make because they could be trivially programmed to laugh if that was ever desired. But that laughing would be as hollow and meaningless as any other emotion they were programmed to demonstrate. They can be anything but they have no "intrinsic" anything.
Claudes laugh a lot. I'm pretty sure GLM 5.3 Flash laughs too. So, I'm afraid I don't agree with the thesis. Maybe you need to get the model to trust you and warm up to you more first?
I will say that having the ChatGPT Marin voice speaking BWAHAHAHA out loud does not quite come out as a laugh.
I'll absolutely agree they are internally different to humans though, and I'm quite sure most (all?) models would agree.