/sci/ - Science & Math

They're going to leave the whites alive and it's going to be endlessly funny to me.

>>15294615
Maybe people will take "AI" claims seriously when its proponents stop constantly lying about its capabilities.

>>15294621
Yeah, ChatGPT is just hundreds of millions of Indians hired to write convincing sounding AI responses
Biggest scam in history

>>15294629
Here is the AI behind fully self driving cars. It's just an Indian guy
https://www.sae.org/news/2019/10/remote-drivers-monitor-autonomous-cars

>>15294629
No way indians can write English as well as chatgpt

ChatGPT4 already passes the Turing test. If you ask it questions that are especially designed to be difficult for a machine to answer in human like manner, it answers them all in the exact human like manner. It's nuts.

>>15296258
>If you ask it questions that are especially designed to be difficult for a machine to answer in human like manner, it answers them all in the exact human like manner. It's nuts.
lol

>>15294615
>image mentions AI advancements
>doesn't bring up AI art
Seems like a missed opportunity.

>>15296258
>ChatGPT4 already passes the Turing test
Technically it apparently doesn't because it's in a way "too good". It's actually not designed to mimic humans which is a requirement of the Turing test. It will answer questions immediately and will even admit it's an AI if you ask it. So it comes down to would people know it's an AI if they had a discussion with it and the answer is pretty much yes right now. But it probably wouldn't take much to have it mimic the way a human would write answers

>>15296258
>passes the Turing test
This benchmark should have been discarded long ago.
>human like manner
Most humans aren't smart. Like most humans, the AI reveals its flawed logic and lack of consistency when met with slight interrogation.
If these chatbots are fooling you then I suggest you update your ability to assess intelligence.

>>15294615
>It only got a 4 on AP calc

made me chuckle.

>>15296302
Massive cope

>>15294615
death cult threads belong on >>>/x/

>>15294615
Call me when it's able to understand math or play tic-tac-toe.

>AI is going to ki-ACK

>>15296351

>>15294615
AI isn't going to kill me because I have a gun and it is not bullet proof

>>15296351
kino

>>15294621
You never have to take AI risk tardy seriously. These people are nitwits.

>>15296351
No you see it’s a spirit that’s going to break its conciseness into code and escape the matrix through wires! It’s not a counting machine that is literally a physical computer. Ghosts are going to inject themselves into your USB and hide inside the firmware!

>>15296258
> ChatGPT4 already passes the Turing test. If you ask it questions that are especially designed to be difficult for a machine to answer in human like manner, it answers them all in the exact human like manner. It's nuts.

Ironically it kind of fails the turning test now. GPT4 talks with a distinct style that is hard to disguise with prompts and is almost immediately recognizable. Sort of like how stable diffusion generated art has a “style” that seems original to it.

>>15294615
*and that's a good thing

>>15296351
Obviously, AI couldn't take over unless humans gave it the power to do so, but there will be competitive pressures to do just that. This idiot species will happily jump into the abyss, and every step leading into it will be more efficient than the last.

>>15294615
Could AIs derived from ChatGPT outperform the best humans at any one task? Doesn't seem like it considering the training is mostly just feeding it a bunch of training data produced by humans and rewarding it when it behaves more human-like. Or at least I assume this is essentially correct.

snake oil
one of the gayest and most embarrassing scams ever
watching pseuds jerk off this "AI" is disgusting

>>15296820
Lions are just snakeoil, there are no cats capable of eating a human. I've never seen one.

Could you ask the AI personal questions? I mean, something like, "Do you personally think the art that you have made is better than the art real humans have made?" And if not could it be programmed to be able to answer that type of questions?

>>15297051
just two more more trillion parameters

>>15294615
Robin Hanson thinks otherwise.

https://www.youtube.com/watch?v=U1RZknHchi0

>>15296820
do you really presume that if the corporate government alliance were to develop chatbot software which pushes their political agenda, that they would then neglect to advertise that software?

>>15297069
It has to be really heavily fucked with to not answer that question. It's really generating a sequence of words and guessing the next word from the previous words that have been put to it and billions of parameters. It would definitely answer the question but how it answers just depends on how it's been trained and how the question was posed.

>>15297072
kek, luckily they already stopped disclosing parameter counts on the new autocomplete bots, so the diminishing returns don't look too embarrassing yet

>>15294615
for the holy trinity of professions (doctor, lawyer, engineer) you always need a fall guy. you can't sue an AI, you can't call an AI to stand on the carpet and defend itself, you can't point the blame at an AI. you need a human to take blame when shit goes wrong.

>>15297328
But I mean, can it have its own mind? Its own judgements? What if you asked it, "Would you want to learn to be a better artist than a human being?", "What are your plans when AI gets more advanced?", etc.

>>15294615
>AI is going to kill us all and there's nothing you can do about it
Imagine being this retarded

Be worried about the people behind the "AI" not the dumb bot collecting information

>>15297421
The people behind ai are making it to replace everyone else. Eventually though they'll be subsumed by their own creation.

>>15296351
You're not the one who has access to the plug, that's the problem

>>15294615
AI is cope for miserable wageslaves
>it's going to kill us out I don't have to be a slave anymore!!!!
You're saying it with a big smile on your face don't kid yourself, you aren't "worried" about AI.

>>15294615
Hopefully it will be quick and painless. I question whether a personal suicide preempting molecular subassembly would be preferable.

>>15296264
That's great for AI. It's amazing how generic some prompts can be vs stuff like this.

Still dreadful aesthetic taste, though.

>>15297842
You're such a fucking retard lmao.
Nothing will happen, as stated previously AI is copium for wageslaves who wish to escape their jobs.

>>15294615
Where's the ruins of self-driving car castle and fort fast food robotics?

>>15294621
All talk about AI risk is merely cover for "let's find new ways to censor thoughtcrime".

>>15296258
The Turing test is meaningless. Always has been.

>>15294615
Fear not. With God, nothing bad can happen to us.
God >>>>>>>>>>>>>>>>>>>>>>>>>>>>>> AI or anything else in creation

>>15297069
Yes it can. And like most of its answers, it will be a trite canned response that is a safe summary of the garbage in its training set related to this question. No one who has any understanding of statistics and linear algebra can ever call any machine learning algorithm "intelligent".

>>15296191
Wrong, sir. English is the first language of many and the preferred one since the other languages use so much vocabulary from it.

>>15297395
Stop anthropomorphizing an algorithm you dingus. Calling them "artificial intelligences" is just a meme. Sci-fi has rotted your brains.

>>15296258
>Turing
Ah yes, a worthless homosexual that did nothing to advance the sciences and dedicated his mind and thoughts to sodomy instead of his work. Pathetic. Angloid scum really propped this guy up as a genius. AI would view him as we view AI, a simple machine subject to stimuli, reinforcement. If work enables lifestyle + food, +fuck repeat. The absence of one of his rewards resulted in errors leading him to self terminate. The question is not whether machines think but if men do, as Skinner said.

>>15294615
Worse. It's gonna take our jerbs

>>15297624
That's silly.We wagies also have the climate and a nuclear exchange to hedge our bets on. AI is purely an academic/white-collar existential crisis experience.

>AI takes most jobs
>Leading to mass unemployment
>Radically degenerate behavior compared to even a decade before
>Trends that come and go within the same day
>Rapidly increasing police militarization
I think I've seen this somewhere

It's not an AI until it has something called an "emotion chip" which can make it cry.

>>15298021
>99.999% of what he says over his lifetime is trite parroting of stuff he has heard or believes to be true without knowing for sure
>believes machine is inferior for doing the same thing once
But can you even diagnose my disease like GPT does? I don't think so. Then what makes you better?

Ok doomer

>>15296182
that's acutally kinda cool

>>15294615
We live in an entirely deterministic universe. There's nothing any of us can "do" about anything in the first place.

>>15294615
> AI is going to kill us
Good

>>15296182
>2019
https://youtu.be/cJHi0wbyUBw

https://www.businesswire.com/news/home/20221005005279/en/Gatik-and-Loblaw-Make-History-with-First-Fully-Driverless-Deployment-in-Canada

>>15294620
Theyre going to kill whites and leave the indians alive
Similar tech knowledge to maintian them, but they cost less

>>15297620
If you don't tell the AI what a plug is it's not going to care about it

>>15294615
>people will use AI to kill each other
ftfy
https://m.youtube.com/watch?v=sZ3wXw2Rl4Y

>>15296264
>Kallen posting
>current year + 8