Kogasa

Well, we knew he was a shitbag beforehand, so that’s not really what’s in question

We aren’t trying to establish that neurons are conscious. The thought experiment presupposes that there is a consciousness, something capable of understanding, in the room. But there is no understanding because of the circumstances of the room. This demonstrates that the appearance of understanding cannot confirm the presence of understanding. The thought experiment can’t be formulated without a prior concept of what it means for a human consciousness to understand something, so I’m not sure it makes sense to say a human mind “is a Chinese room.” Anyway, the fact that a human mind can understand anything is established by completely different lines of thought.

This fails to engage with the thought experiment. The question isn’t if “the room is fluent in Chinese.” It is whether the machine learning model is actually comparable to the person in the room, executing program instructions to turn input into output without ever understanding anything about the input or output.

You can get blight free versions of JavaScript too, if you use TypeScript.

In a hypothetical and highly unlikely world where everyone had to pay Oracle to use Java, everyone would switch to something else. It would be guaranteed suicide. Anyway, in that world, they would need to both make this ridiculous decision and win an unwinnable legal battle afterwards. It’s not a realistic concern.

3 billion devices

What’s the lockin? Is it really harder than just swapping the jdk path to switch between Coretto and OpenJDK? I understand Coretto being preferable for performance and security patches but I don’t imagine it’s that big of a deal if one eventually had to switch

Yup, you’ll notice the only thing distinguishing C from R^(2) is that multiplication. That one definition has extremely broad implications.

For fun, another definition is in terms of 2x2 matrices with real entries. The identity matrix

1 0
0 1

is identified with the real number 1, and the matrix

0 1
-1 0

is identified with i. Given this setup, the normal definitions of matrix addition and multiplication define the complex numbers.

One definition of the complex numbers is the set of tuples (x, y) in R^(2) with the operations of addition: (a,b) + (c,d) = (a+c, b+d) and multiplication: (a,b) * (c,d) = (ac - bd, ad + bc). Then defining i := (0,1) and identifying (x, 0) with the real number x, we can write (a,b) = a + bi.

I’m not suggesting that. It just looks like he did in fact choose to have that hair color.

It looks dyed

It does, which is why it’s autocompletion and not auto-gibberish.

Yes, but that’s not how LLMs work. My statement depends heavily on the fact that a LLM like GPT is coaxed into coherence by unsupervised or semi-supervised training. That the training process works is the evidence of an internal model (of language/related concepts), not just the fact that something outputs coherent statements.

deleted by creator

It must have some internal models of some things, or else it wouldn’t be possible to consistently make coherent and mostly reasonable statements. But the fact that it has a reasonable model of things like grammar and conversation doesn’t imply that it has a good model of literally anything else, which is unlike a human for whom a basic set of cognitive skills is presumably transferable. Still, the success of LLMs in their actual language-modeling objective is a promising indication that it’s feasible for a ML model to learn complex abstractions.

That’s not necessarily wrong, but not the big explaining factor here I think. The technological challenges behind aligning ML models with factual reality aren’t solved, so it’s not an engineering decision. It’s more that AI is remarkably easy to market as being more capable than it is

deleted by creator

deleted by creator

Mesa is usually pretty quick to update, it’s just that stable distros won’t update mesa all that quickly. I assume most of them have some way to install a newer mesa from a community repo or something.

Docker is lighter and easier to manage than a VM. I run a collection of services as docker compose services inside a NixOS host VM. It’s easy to start, stop, monitor, update etc. even from a different computer (via ssh or docker contexts). It’s great.