Yes, LLMs emit tokens. And human brains emit nerve impulses. Both devices are "trained" on corpora of data.
That has nothing to do with the question of whether the mechanism that generates those signals is complex enough to say that it's "creative."
4,785 karma · joined July 7, 2014
That has nothing to do with the question of whether the mechanism that generates those signals is complex enough to say that it's "creative."
Great, but we also build robots with world models and sensory pre-processing. So nothing you've described is unique to humans.
> I never mentioned creativity, nor would I in relation to LLMs
The topic of this thread is whether LLMs are capable of genuinely contributing "new" ideas. So if that's not the point you're trying to make, I'm not sure of the purpose of your comment.
> But neither POV seems interested in anything more than drive-by hot takes, debating over-simplistic strawmen or trading 'gotcha' hypotheticals.
I think you've oversimplfying the discussion here.
In my view, at least, both "LLMs are nothing more than stochastic parrots" and "LLMs can be as intelligent, creative, innovative (etc) as any human or all humans combined." are simultaneously true, for the simple reason that humans are stochastic parrots. Our brains learn patterns and respond to stimuli.
Okay but how is that relevant for measuring whether an llm is capable of creating original thoughts? Why do you think that complexity, as you say, is a more relevant Factor than simply the number of weights?
That's a rather dismissive way of putting it. Moreover it's confusing the output format with the complexity of the output. I could just as easily say that no matter what the human brain does it's just generating stimulus to motor neurons. Such a simplification is just as wrong as saying that it's just as misleading is saying that an llm is just a token generator. What makes an llm able or unable to be complex is the process that creates those tokens
But that interpretation of Free Will is a legal fiction. Because of course what people do is determined by their upbringing and their opportunities and the environment that they have. In fact, that's an argument of a lot of legal reform movements that seek to move responsibility from the individual. The goal of the law is to assign responsibility and create a set of incentives which will hopefully result in orderly society.
But in the AI debate, we're not just creating incentives for an orderly Society, we're talking about the nature of creativity and the impetus to act. And by that measure I don't think there is a significant difference between AIs and human beings.
Yesterday, my boss told me to fix a bug in our product. Then, I fixed it. Today my boss is taking credit, saying that he fixed it. I guess he's right, since he told me to do it.
That's true. Of course we can hook an llm up to Motors and sensors. That's a robot. Or a self-driving car. So would you say that those devices can confirm the ground truth about physical reality, and therefore are capable of creativity?
How are you measuring complexity here? How are you measuring inputs? Your average llm is trained on a corpus that vastly exceeds the amount of data I could read in my lifetime.
I've heard that before but it just doesn't make sense in context of what I've seen llms do. If I ask an llm to write a poem about magnetic resonance and vampire rabbits it can do that it created a new thing. I can ask it to build a website for managing rabbit breeding that's also a new thing.
Another way of looking at it is that human beings, just like llms, can produce output based on their inputs. Most literature is inspired by other literature. Most music is inspired by other music. Most software is inspired by other software.
So I think we need to work on defining " new things" before we can definitely exclude them from llm's capabilities
When that corporation is performing contracted work for the government, they act as the government's agent, making them subject to the same restrictions regarding civil rights as the government itself.
There's an interesting analogy with copyright issues surrounding AI use. Both copyright laws and surveillance were created to (respectively) creative incentives for authors to create novel works; and to protect citizens for indiscriminate surveillance. The laws made sense given the technology available at the time: it simply wasn't feasible to put a cop on every street corner, or to ingest the ideas of every work ever written.
But the laws have been eclipsed by technology. Technology makes a mockery of existing laws, by clearly violating their intent. Much has been written to the effect of "why bother to create new, free materials on the internet when it will just get scooped up by an LLM with no attribution to me." Similarly, Flock cameras have the effect or violating privacy in the exact same way the original laws sought to prevent.
Unfortunately, laws are slow to change, and monied interests will fight tooth-and-nail to put their profits above the rights of ordinary people. It's going to be a difficult decade or so coming up.
If you're genuinely curious, and can't wait to watch the film, here's a starting point: https://www.rollingstone.com/culture/culture-lists/elon-musk...
Enjoy your cheap lawyer tokens while you can.
Because you'd rather pay your $1000 to OpenAI for 5 milliseconds of work, instead of to a human lawyer for 10 seconds of work?
No intellectual capability, but non-human, inherently amoral intellectual capability that has already been proven act in hard-to-predict, illegal, and unethical ways. I think that's a very rational thing to be afraid of.
> But wouldn't we want to see the black hole? One could send a ray through every pixel of a screen, and let them go around the black hole and hit its surroundings. And if the rays are sent back in time, it's as if one received the light emitted by the surrounding objects in the past!
I wasn't aware that I was making a "subtle" point. I was joint showing that the use of articles in English is complex and hard-to-predict for non-native speakers, in contradiction to the claim (paraphrased) that "Just use the for things that are definite, duh!" Your observation that article use varies between dialects of the same language strengthens my point, insofar as encountering inconsistent use makes learning the language even harder.
> is determined arbitrarily.
I don't know what you're trying to say here. Languages vary systematically over time. Just because two different dialects, or two different languages, do things in different ways doesn't mean that those changes were arrived at arbitrarily.
No. To be "in hospital" or to go "to hospital" means that you are a patient. To be "in the hospital" describes only location.
> Your description is incomplete
Your analysis is fine, but does not contradict my comment and moreover is irrelevant to the present discussion, which pertains to the subtle use of articles in English.