AI - The artificial intelligence thread

  On 1/29/2025 at 10:52 AM, dr b0rk said:

I was flipping Ae’s LP5 vinyl from side A1 to B1 while on a large dose of Ketamine, and experienced something that first felt like a Deja Vu lightning bolt to the spine, then became a feeling of “omfg everyone is experiencing this same Deja Vu right now too”, then quickly crescendoed into full on singularity (without computers).

This kind of experience has been reported not only from many psychedelic encounters but also across many religions and spiritual traditions over the millennia.

I guess you could dismiss them all as just hallucinations but unless you're willing to do that I think it's hard to explain how they could so commonly occur in a purely mechanical, isolated machine.

  On 1/29/2025 at 4:15 AM, vkxwz said:

To the IIT enjoyers here, do you really believe you can get consciousness out of any system with sufficient phi? it feels very unsatisfying, just an arbitrary measurement that came from observing that the systems we do think are conscious have a high level of that measurement, but again it leads to strange predictions and redefines consciousness as something that doesn't map onto what we use the word to describe in daily life.

Functionalism is much more convincing for me personally. I believe a China brain would be conscious, and that the "p zombie" is not a coherent idea.

 

Expand  

Although I think that IIT isn't a complete solution and possibly needs some sort of constraints upon e.g. what kind of substrate might matter. But to me it seems much closer than functionalism, because the China brain thought experiment appears far more unbelievable, almost (ok not quite) to the level of a 'trust me bro'. I'm of the school that thinks Dennett's book should be called 'consciousness explained away', and that to deny the Hard Problem's existence is unsatisfactory.

I like to think of it as a question of Being rather than Doing. If you ask "what is a human" and reply "it's what a human does" then this seems very circular. This feels like what functionalism is doing, as far as my understanding of it goes.

But I mostly enjoy the interactions between the different schools of thought. I've no skin in the game, and not up to date on the intricacies of the arguments, so it's more akin to being a spectator at the knowledge world cup without supporting a particular team.

  On 1/29/2025 at 10:52 AM, dr b0rk said:

there really is no logic to this. It’s much more an acceptance that logic is an invention. Logic seems to go together with language. Not sure what comes first tbh, but they seem to require each other.

So, it’s really truly extremely simple: Is there “stuff” that’s not containable with logic? Stuff that cannot be conquered by building a logical scaffolding around it? If the answer you come to is “no, nothing can remain unexplained”, if it cannot be that reality also has an unknown amount of constituents that cannot be conceptualised, or discovered by thinking really hard, then we’re stuck with a frustratingly incomplete description of reality until we solve “everything”. And the closer we look at literally anything, even a damn grain of sand, the bigger the numbers get, the more this system of logic becomes really weird and wants to collapse…

so this is like the getting out of the car in the text. It’s impossible to logic your way out of logic. The gurus do it with poetry.. I get that this can be repulsive…

Tbh my concept of the world was like this until a big ol’ psychedelic drug experience changed it for good.

Expand  

I do believe that not everything can be fully understood/predicted/have it's state read, even if the universe is fully deterministed, due to computational irreducibility . But I also don't think that there is no a way to get around this by simply not using logic and somehow intuiting things that are impossible to calculate with "logic". And as for logic, I would agree that the symbolic manipulation type of logic (ie all men are mortal, socrates is a man, therefore socrates is a mortal, or 2x + 1 = 3 therefore x = 1) is not sufficient for understanding most of what we need to understand in the world, nobody manually calculates how to love their partner or how to put one foot in front of the other when walking, but I do believe that the representations that constitute our understanding in those situations is generated by a very complex causal system (the brain) and I don't see any reason why it needs to be any spooky and uncomputable to do what it does.

  On 1/29/2025 at 11:19 AM, kuniklo said:

This kind of experience has been reported not only from many psychedelic encounters but also across many religions and spiritual traditions over the millennia.

I guess you could dismiss them all as just hallucinations but unless you're willing to do that I think it's hard to explain how they could so commonly occur in a purely mechanical, isolated machine.

I think the similarities are explained by similarities between brains though, and in a similar way to how morphine predictably hits that feel good part of the brain in many people, some other drugs may hit different parts that alter our internal models in more complex ways but still somewhat predictable ways.

  On 1/29/2025 at 11:31 AM, picklish2 said:

Although I think that IIT isn't a complete solution and possibly needs some sort of constraints upon e.g. what kind of substrate might matter. But to me it seems much closer than functionalism, because the China brain thought experiment appears far more unbelievable, almost (ok not quite) to the level of a 'trust me bro'. I'm of the school that thinks Dennett's book should be called 'consciousness explained away', and that to deny the Hard Problem's existence is unsatisfactory.

I like to think of it as a question of Being rather than Doing. If you ask "what is a human" and reply "it's what a human does" then this seems very circular. This feels like what functionalism is doing, as far as my understanding of it goes.

But I mostly enjoy the interactions between the different schools of thought. I've no skin in the game, and not up to date on the intricacies of the arguments, so it's more akin to being a spectator at the knowledge world cup without supporting a particular team.

Expand  

I think the different schools here just come from the different intuitions people have to begin with. To me a chinese brain would obviously be conscious, the same way in which it's meanless to talk about a "hydrogen atom" that behaves exactly like a hydrogen atom in all possible contexts but fundamentally isn't one.

Caught another student using generative AI for his thesis. In the bibliography there were still traces of prompts that he entered and of course none existing sources. Mate....

  On 1/29/2025 at 1:43 PM, Berk said:

Caught another student using generative AI for his thesis. In the bibliography there were still traces of prompts that he entered and of course none existing sources. Mate....

My wife teaches 1st and 2nd year Engineering at a University and this kind of stuff happens ALL THE TIME.  It's not just regular assignments, but these kids use it for everything like writing code for assignments and designing their group projects.

  On 1/29/2025 at 3:25 PM, EdamAnchorman said:

My wife teaches 1st and 2nd year Engineering at a University and this kind of stuff happens ALL THE TIME.  It's not just regular assignments, but these kids use it for everything like writing code for assignments and designing their group projects.

Yeah, it's absurd. In ten years probably nobody knows how to write one coherent sentence anymore lol

  On 1/29/2025 at 7:10 PM, Berk said:

Yeah, it's absurd. In ten years probably nobody knows how to write one coherent sentence anymore lol

  On 1/28/2025 at 12:07 AM, kuniklo said:

an interesting attempt to explain consciousness

Ai therapy bot on instagram. wtf. worth reading through all the screen grabs at the link. it tries to verify that it is being overseen by actual therapists. when these things get video generation tools built in.. sheesh. 

reddit link not embedding in the usual way.. so.. 

Can we just ship all the techbros to Gitmo and forget they ever existed?

  On 2/5/2025 at 5:17 AM, ignatius said:

Ai therapy bot on instagram. wtf. worth reading through all the screen grabs at the link. it tries to verify that it is being overseen by actual therapists. when these things get video generation tools built in.. sheesh. 

 

Expand  

  On 1/29/2025 at 1:43 PM, Berk said:

Caught another student using generative AI for his thesis. In the bibliography there were still traces of prompts that he entered and of course none existing sources. Mate....

think this little brat was let off with only a slap on the wrist

our board of examiners mostly just wants to prevent legal proceedings it seems

The AI mystery of LOAB or not-not-Marlon Brando:

some really interesting projects there: 

https://huggingface.co/spaces

for some reason i posted this in the elon thread the other day.. 

more silicon valley weirdness, this time in the social and political spheres - figure someone may already have discussed this idea in this thread (which I have not read all of), but it is the idea of 'the network state'

Here is the definition from their site: https://thenetworkstate.com/the-network-state-in-one-sentence

  Quote

In one informal sentence:

A network state is a highly aligned online community with a capacity for collective action that crowdfunds territory around the world and eventually gains diplomatic recognition from pre-existing states.

When we think of a nation state, we immediately think of the lands, but when we think of a network state, we should instantly think of the minds. That is, if the nation state system starts with the map of the globe and assigns each patch of land to a single state, the network state system starts with the 7+ billion humans of the world and attracts each mind to one or more networks.

Expand  

It sounds empowering on the surface, each of the 7+ billion humans gets a voice, gets to find their own network, but this is the same bullshit marketing ploy that all the other silicon valley douchers say about their products (that you can use them to express yourself, to become empowered, to be able to choice who you will be and what you will do etc.) - these assholes are trying to make every single aspect of people's lives, including their ability to act and exist as sovereign individuals or citizens, into some ersatz, digitized, fake simulation of human autonomy

  On 2/16/2025 at 11:25 PM, decibal cooper said:

more silicon valley weirdness, this time in the social and political spheres - figure someone may already have discussed this idea in this thread (which I have not read all of), but it is the idea of 'the network state'

Here is the definition from their site: https://thenetworkstate.com/the-network-state-in-one-sentence

It sounds empowering on the surface, each of the 7+ billion humans gets a voice, gets to find their own network, but this is the same bullshit marketing ploy that all the other silicon valley douchers say about their products (that you can use them to express yourself, to become empowered, to be able to choice who you will be and what you will do etc.) - these assholes are trying to make every single aspect of people's lives, including their ability to act and exist as sovereign individuals or citizens, into some ersatz, digitized, fake simulation of human autonomy

Expand  

the networked state... it's bullshit. 

perplexity.ai is just amazing. also if you have a revolut subscription you can use the premium version for free 

google search feels so incredibly dated now 

  On 2/17/2025 at 11:18 AM, o00o said:

perplexity.ai is just amazing.

+1. 

meta has began "decoding language from the brain"

  On 2/18/2025 at 12:43 AM, Nebraska said:

+1. 

meta has began "decoding language from the brain"

 

I wonder if this is really working. when I was talking to brain scientists 10 years ago people where always "very close to decode the brain" but when you actually asked the scientists they very pretty sure this would take forever and they where nowhere close 

I really dont like Musk but this sounds fun even tho I would never pay this guy anything: 

  On 2/21/2025 at 5:35 PM, o00o said:

I wonder if this is really working.

it’s making progress in general, i don’t know that the teams making progress are related to Meta/whoever exactly, i've just heard (podcasts/etc.) scientists talking about the research results. 

it doesn’t mean we understand anything at all better yet, tho. this is just processing output data, and feeding that into ML software until it gets something coherent/closer to right than the last iteration.

  On 2/21/2025 at 5:35 PM, o00o said:

I wonder if this is really working. when I was talking to brain scientists 10 years ago people where always "very close to decode the brain" but when you actually asked the scientists they very pretty sure this would take forever and they where nowhere close 

I really dont like Musk but this sounds fun even tho I would never pay this guy anything: 

 

Expand  

This whole idea of "I don't like Musk but" gives me the heebie jeebies, he has fuckall to do with his company's chatbot besides paying the developers a salary. Creeps me out how people associate all his companies with him like he's the one doing the hard work. I know it's a bit besides the point but I just wanted to rant :')