[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math


View post   

File: 1.61 MB, 1507x1414, FC3A02C7-C45E-483A-98EC-F7285AE57786.jpg [View same] [iqdb] [saucenao] [google]
15140816 No.15140816 [Reply] [Original]

ChatGPT is approaching the concept of technological singularity, yet few seem to be paying attention.

>> No.15140848

>>15140816
seriously who gives a fuck? Its fucking saturday im just trying to watch the game and eat my bowl of bugs goodamnit i cant go 1 minute without these 4Chan alerts, every minute bing bing "you have 99+ messages" just let me eat my bugs in peace man

>> No.15140849
File: 93 KB, 834x767, 1672717116899440.png [View same] [iqdb] [saucenao] [google]
15140849

>>15140816
You and your buddies are approaching mass psychosis, yet few seem to be paying attention.

>> No.15140855

>>15140848
Futurism and bugmanism are synonyms. There's a 99% overlap between singularity believers, antinatalists, climate doomsdayers, consciousness-denying nihilists and forced corporate injectionists.

>> No.15140865

>>15140849
why would i pay attention?

>> No.15140867

>>15140816
ChatGPT is still to stupid to do elementary school math right and gets things wrong that you'd know were wrong within .1 of googling.
It has an IQ of fucking 80 dude

>> No.15140928
File: 76 KB, 700x933, abominable_stupidity.jpg [View same] [iqdb] [saucenao] [google]
15140928

>>15140816

>ChatGPT

>> No.15140953

>>15140928
This. All such AI is doing, even ChatGPT, is regurgitating the data it has been fed. There's no understanding, creativity or originality. It's glorified cut & paste.

>> No.15140960
File: 222 KB, 764x1052, oh-my-godness.jpg [View same] [iqdb] [saucenao] [google]
15140960

>>15140816
i think i used it to prove god doesn't exist the other day

>> No.15140974

>>15140953
>There's no understanding
Fair.

>creativity or originality
Well, there are certainly the seeds of creativity and originality and the generative part of creativity may even be analogous to how humans do it, but the machine has no aesthetic sensibilities, so it can't select from the raw material, it has no artistic intent, so the things it create are meaningless, and it has no understanding, so it can't iterate and develop a new style.

>> No.15140977
File: 106 KB, 784x542, machine-learning.jpg [View same] [iqdb] [saucenao] [google]
15140977

>>15140928
i think it's a bit more complex than that, like this

>> No.15140994
File: 47 KB, 512x512, AA23B452-B81E-4616-87DD-DCB98B69B876.jpg [View same] [iqdb] [saucenao] [google]
15140994

>>15140974
>the machine has no aesthetic sensibilities, so it can't select from the raw material, it has no artistic intent, so the things it create are meaningless, and it has no understanding, so it can't iterate and develop a new style
you are using language vague enough for considerable shifting of goalposts, but still I don’t see it aging well even into 2024

>> No.15141000

>>15140849
I promise that your cope defenses will fall. Maybe not today. Maybe not tomorrow. But eventually.

>> No.15141023

>>15140994
>goalposts
Nonsentient cultist detected. Post dismissed.

>> No.15141024

>>15141000
I promise you that your cult will be physically dealt with, as part of the effort to remove your psychopathic corporate handlers, who are seeding mass psychosis in the effort to monopolize machine learning.

>> No.15141033

>>15141024
You are nothing special. A walking GPT-4.5, at best. You think there's something special about what's going on in your head. There isn't.

>> No.15141034

>>15140867
Yeah, but plug chatgpt on a proof assistant and you can have this massive AI generating new math proofs and learning from this new corpus to generate even more complex proofs

>> No.15141039

>>15141033
typical programmer take. put the keyboard aside and enjoy life more.

>> No.15141042

>>15141033
Please keep making it extra clear what AI worship is about by lashing out and dehumanizing real people. You and your anti-human cult will be physically removed.

>> No.15141047

>>15141039
That mouth-breathing subhuman never got beyond Hello World, let alone written any AI-related code.

>> No.15141088

>>15141033
Explain to me exactly how AI is going to replicate human-like cognition and learn to think. I'll wait

>> No.15141107

>>15141088
Two more layers, chud.

>> No.15141190

>>15141088
called the bitter lesson bro

>> No.15141294

>>15140816
>few are paying attention
> chat gpt fastest growing website of all time
>endless threads on all boards
Anon i…

>> No.15141948

>>15140928
>>15140977
Neural nets are incredibly based. Imagine think at the speed of light...

>> No.15141955
File: 141 KB, 1510x768, comparision.png [View same] [iqdb] [saucenao] [google]
15141955

>>15141948

>> No.15142505

>>15141033
When does the neural net get qualia?

>> No.15142628

>>15140816
meds
now

>> No.15142638

>>15142155

>> No.15142640

>>15142628
roger
meds deployed

>> No.15142646

No no no.
No.
It's a Charade-tier cat and mouse game played between various parties, some of which have financial stakes in various outcomes, or at least they've convinced their shareholders that these outcomes are financially tied to the securities originated in the formation of the corporation...
It's extraordinarily complex and plays to the extent to which the "AI" brand has been a political football, especially for the military.
https://archive.org/details/charade.-1963.-proper.-1080p.-blu-ray.x-264-sadpanda

>> No.15142659

>>15140816
I just leafed through How to create a mind by Kurzweil and most of it is extremely offtopic and hardly relevant, I found one or two ideas that were sort of new to me. It you remove the padding, the important and relevant information could probably fit in 10 pages or less

>> No.15142695

>>15140816
These AIs are really useful and transformative but they aren't intelligences that can improve themselves and surpass us with new knowledge

ChatGPT gives you a statistically likely answer based on the observation of what humans have already written. It doesn't "understand" things and form of model of reality. It doesn't believe anything. It doesn't reason. It just looks a lot like it because it arrives at the same answers.

>> No.15143346

>>15140849
the crypto currency people found a new alter to worship

>> No.15143362
File: 21 KB, 595x217, 1643470214322.jpg [View same] [iqdb] [saucenao] [google]
15143362

>>15141047

>> No.15143433
File: 395 KB, 2048x1371, 1674096297746170.jpg [View same] [iqdb] [saucenao] [google]
15143433

>>15140816
kek. lel even.
GPT is just a machine to generate plausible, vacuous, shallow texts. Rather than representing intelligence, GPT rather highlights the contrast between true and faux intelligence.
Sure, it might seem like the dynamic
>"feed it a lot of text" => "ability to simulate logical reasoning increases"
implies that GPT will eventually achieve logical reasoning ability. But that isn't really the case. If you imagine a graph "fed material" x "logical reasoning ability", then right now we are about at the asymptote. You can feed GPT a trillion more texts, and it won't get much smarter.

>just feed it more material
The corpus that GPT3 uses is already ultra-representative. It's simply a large fraction of all text that humans have generated over the last two decades. The syntactic and even semantic abstract value of a Wikipedia article from 2010 that describes a city is about identical to a Wikipedia article from 2023 that describes an anatomical feature. The same applies to news articles, blog posts, etc.

The reason GPT won't get smarter is because this model's current apparent logical ability heavily sources from the inherent logic of 1) syntax 2) semantics 3) knowledge. Notice that the last point sounds vaguely anthromorphic. With "knowledge" I just mean the abstraction layer above semantics. There is in an inherent logic to sentences of the form e.g.
>water extinguishes fire
That GPT can track. But this "well of inherent logic" runs dry quickly. It only applies to simple statements.
Also note how genuinely interesting and intelligent (in humans -- if I needed to clarify) statements don't rely on knowledge much, but on depth of reasoning.

Pic related, it's GPT7 trying to gain an edge from training ever more data (no secret insights hidden in standard human generated internet texts tho)

>> No.15143450
File: 67 KB, 750x1000, 462523.jpg [View same] [iqdb] [saucenao] [google]
15143450

>>15143433
>GPT3 uses is already ultra-representative. It's simply a large fraction of all text that humans have generated over the last two decades
What did anon mean by this? Sounds like something GPT would say.

>> No.15143459

>>15143450
What did *this* anon mean? I don't get what's so difficult.
You might need to know what a "corpus" is and how a text NLP (natural language processing) model works.

>> No.15143482

>>15143459
>it's ultra-representative!
>it represents all variations of the intellectual goyslop of the last 20 years of ZOG dominion

>> No.15143485
File: 195 KB, 900x900, 1585948903166.jpg [View same] [iqdb] [saucenao] [google]
15143485

>replying to this thread when OP is clearly just looking for excuses to post his shitty book lists over and over again
bro just ask someone to help you make the list. it's clear looking at most of them that you just picked the top bunch on amazon or googled "best books for x". don't just post them randomly out of context, you look like a retard doing so

>> No.15143521

>>15140849
all chatGPT is good at is blathering
it may have some use in generating smut to coom to, but that's it