Yes, it’s the cool new tool. However, I’m not sure if all the people throwing it at problems understand the difference between the type of use where it can probably synthesize a reasonable answer, depending on how deep or shallow the subject matter is in the dataset, verses uses where it will produce cheerful BS, like non-existent book suggestions.
3 Likes
That would be meaningful if I was talking about religious beliefs. Everything I’m talking about is compatible with science. But reductionism, while long popular, is not the only view on the scientific reality of our universe.
3 Likes
It is, however, the only one that has proven testable so far, and ideas that are not testable may want to say something about science, but they are not in themselves science, just beliefs. It’s okay, I have metaphysical beliefs about the nature of consciousness too, you’d probably even say mystical ones. But I opened talking about scientific inquiry and I don’t find a metaphysical critique of scientific inquiry very compelling.
You seemed to be saying that because there’s something ineffable there, there’s no point in conducting a scientific inquiry. I’m responding that regardless of whatever metaphysical element there may be, there is scientific exploration we could do that might shape our beliefs in a productive way, and it’s silly to give up on this line of inquiry.
1 Like
Do some reading on emergence and complexity theory and get back to us. I’m not talking about metaphysics.
5 Likes
The idea that ChatGPT is any sort of intelligence is preposterous, and I think most of the confusion comes from a lack of understanding of what an inference engine even is.
Use the API yourself, set the “randomness” variable to 0. ChatGPT will have perfect recall. It will always give the same exact answer to the same input, forever. The only way it varies is if you intentionally distort its algorithm. That’s not learning! If you program a robot to follow a specific set of brushstrokes to paint a picture, it will do that every time. If you add some stochastic randomness to each brushstroke (so that now, every painting is unique), it didn’t “learn” anything. It just slightly fucked up everything it tried to do, which made it look “new.” Perhaps the most important part of that reality is that it also cannot learn anything from this randomness. It’s not aware of its’ output. It would not learn a thing if it painted 50,000 pictures.
ChatGPT is the same. If you run it at its default, we algorithmically add stochastic randomness to its’ output - essentially intentionally fuck up its perfect recall - to force it to give differentiated answers. But it doesn’t know we’ve done this, and it learns nothing from this at all. It isn’t capable of the sort of self-reflection required to even realize these things. The worst thing we ever did was to call an inference engine AI. They are nothing like an emergent neural network. They aren’t even designed to learn! They’re designed to respond identically from the same corpus to every stimuli, then we fuck up their recall so they will respond slightly differently at each decision gate.
13 Likes
The great thing about words, and language, is that we can have a conversation without getting into mindless pedantry about what words mean. That’s how discussion forums work.
As you may have noticed, most everyone else on the topic can manage to agree on 1) what intelligence means and 2) this isn’t it. You are welcome to try and move those goalposts and attempt to have ChatGPT into your definition if you so desire (again, discussion forum), but from the responses, I don’t think many other people care to join you in that redefinition. And the crazy thing about language is that it means what the preponderance of folks tend to believe it means, regardless of what “the establishment” would prefer it to say.
So, you’re welcome to keep trying to say that folks here just don’t understand what “intelligence” is, but we kinda sorta innately do, and many of us also kinda sorta understand what an inference engine is well enough to know that intelligent, it isn’t.
Yeah, it’s an imperfect universe. Sorry about that.
11 Likes
I’ve found that those humans who insist that LLM’s and such are truly cognizant and intelligent (or are right around the corner from being so) tend to be either not very sophisticated thinkers themselves or the kind of software coders who’ve convinced themselves that their own work product can create solutions by imposing algorithmic perfection on inherently imperfect and messy and chaotic situations (e.g. creating surefire romantic matches or spitting out a completely accurate medical diagnosis).
The latter are inclined to believe that the human brain itself is nothing more than (per the op-ed) “a lumbering statistical engine for pattern matching, gorging on hundreds of terabytes of data and extrapolating the most likely conversational response or most probable answer to a scientific question.”
8 Likes
So they think other people are like themselves?
(Could well be, given most that I’ve met.)
9 Likes
Going by my own anecdotal experience over decades in the industry, there’s a distinct subset of them. There’s often an element of a Frankenstein-like arrogance and God complex underlying their views.
The situation gets more problematic when the individual’s ideological views also inform the position on sentience (or super-sentience). Misogyny is the most common one, Libertarianism a close second in frequency, but there are others like this famous instance.
6 Likes
Tell me you were raised Canadian without telling me!
12 Likes
Of course not. Real understandingTM comes from taking the consensus of all the writing you can scrape from the internet. 
7 Likes
In English the word sentient is this single-purpose term that has no agreed-upon definition but always gets thrown around by sci-fi nerds arguing about whether robots and aliens are people or not 
4 Likes
ChatGPT is not a being who has “experiences”
It’s a text engine that responds to commands
—like /bin/bash or the Perl debugger
6 Likes
If you are going to be pedantic about the language, I’m going to be pedantic about the philosophy. I’ll change my word choice when you prove definitively to me you’re a being who has experiences, i.e. without begging the question.
(I don’t actually want to argue about semantics or philosophy, I just want to use the word by analogy, in what I think is a pretty obvious and useful way: the “experience” of ChatGPT is any training data that goes into its preparation, and to some extent the conversational history for any particular instance of it in any particular conversation. It’s not meant to romanticize what’s going on.)
8 Likes
If we’re going to go the whole Philosophical Zombie route, I reserve the right to declare that I am a Boltzmann Brain, and you are all the figments of a diseased imagination: go ahead, prove me wrong.
ChatGPT does not have experiences, because it is incapable of learning or remembering things it has done. It repackages things in a mechanistic way, and it fools us into thinking there is some sort of sentience behind it because its coders introduced a deliberate random fudge factor.
It has no memory, therefore it can’t remember, therefore it can’t learn, therefore it can’t experience anything QED.
If you’re going to attempt to be pedantic about the philosophy, it helps to recognise when the philosophy you’re regurgitating is bullshit.
5 Likes
We can’t both be Boltzmann brains, can we?
Oops, my time is up.
(Poof!)
8 Likes
Yup. This.
And from where I sit what we have now is fundamentally more or less the same, but running on hardware that is magnitudes of order more powerful than the old iron, and using vastly improved algorithms.
Which is absolutely amazing.
But it’s also
- not even close to becoming an intelligent… whatever, let alone being sentient. Or feeling.
- not the one universal solution to all our problems as a species (see also: blockchains, genetic engineering, nuclear power, petrol, physics, chemistry, steam power, alchemy, astronomy, … you get the gist)
5 Likes
“The ‘experience’ of SQL is any structured data that’s run through its algorithm.”
4 Likes
Don’t forget they now have the content of the Internet, which makes it easy to build a huge dataset, which has instant intelligence (and prejudice) already mixed in.
3 Likes
