DarthOne
☦️
Yeah, and we become more reliant on automation. Machines that can't think for themselves.
AI/Automation Megathread
- Thread starter Bear Ribs
- Start date
I asked GPT-4 if it could write a simple Windows batch file to analyze a folder's contents. It spat out the code in very short order. You can copy this into a text file, rename the file to MakeFolderReport.bat, and it will report the contents of a folder (excluding subdirectories) and output the result to output.txt.
It can also manage way more impressive feats than that.
Also, it's hilarious how closely Elevenlabs can replicate Steve Downes' deadpan.
EDIT: I just asked ChatGPT if it could write a .pov file for "an apple sitting on a table" that could be rendered in POV-Ray, and it immediately obliged.
I rendered it in POV-Ray.
What. The. Fuck.
Guys, GPT-4 has no eyes. It has no visual cortex. It can't visualize jack shit. It was trained primarily on a massive corpus of text. It's not even remotely conscious.
Did mankind just build a fucking Scrambler from Blindsight?
EDIT 2: When I told it that the apple was floating and the light source should be off to one side a little so we can see the shadow when it's touching the table, it fixed the code.
Again, what in the absolute fuck.
I remember, about ten years back, on SB, I said that by around 2030 or 2040 or so, someone would be able to describe a video game, and an AI would spit out the whole damn thing. Code, art, music, voice acting, story, everything. People said bullshit, it wouldn't happen.
It's looking like I wasn't off the mark at all. Imagine what it would look like if they took the AI-driven tools we already have and integrated them all together right now. Imagine where this shit will be in seven years.
MakeFolderReport.bat:
@echo off
setlocal enabledelayedexpansion
:: Clear any previous output file
if exist output.txt del output.txt
set "filecount=0"
for %%A in (*.*) do (
if not "%%A"=="MakeFolderReport.bat" (
set /A "filecount+=1"
set "filename=%%A"
set "filesize_bytes=%%~zA"
set "datemodified=%%~tA"
set /A "filesize_MB=filesize_bytes / 1024 / 1024"
set "filesize_MB=!filesize_MB!.!filesize_bytes:~-6,-3!MB"
echo !filename! - !filesize_MB! - !datemodified! >> output.txt
echo. >> output.txt
)
)
echo Total Files: %filecount% >> output.txt
echo Files are listed by Name - Size (in MB) - Date Modified >> output.txt
echo File list, sizes (in MB), date modified, and total file count have been saved to output.txt.
pauseIt can also manage way more impressive feats than that.
Also, it's hilarious how closely Elevenlabs can replicate Steve Downes' deadpan.
EDIT: I just asked ChatGPT if it could write a .pov file for "an apple sitting on a table" that could be rendered in POV-Ray, and it immediately obliged.
POV-Ray Apple:
// Apple On Table - POV-Ray Scene
// Version: 3.7
global_settings {
assumed_gamma 1.0
}
camera {
location <0, 2, -5>
look_at <0, 1, 0>
angle 60
}
light_source {
<0, 15, -15>
color rgb <1, 1, 1>
area_light <2, 0, 0>, <0, 0, 2>, 3, 3
adaptive 0
jitter
}
// Table
box {
<-3, 0, -3>, <3, 0.1, 3>
texture {
pigment {
color rgb <0.5, 0.25, 0.1>
}
finish {
diffuse 0.8
specular 0.2
phong 1
phong_size 50
}
}
}
// Apple
sphere {
<0, 1, 0>, 0.5
texture {
pigment {
radial
color_map {
[0.0 color rgb <1, 0, 0>]
[0.9 color rgb <1, 0, 0>]
[1.0 color rgb <0.7, 0.1, 0.1>]
}
}
finish {
diffuse 0.8
specular 0.6
reflection 0.05
}
}
scale <1, 1.2, 1>
translate <0, 0.5, 0>
}
// Floor
plane {
y, 0
texture {
pigment {
color rgb <0.9, 0.9, 0.9>
}
finish {
diffuse 0.8
specular 0.2
phong 1
phong_size 50
}
}
}I rendered it in POV-Ray.
What. The. Fuck.
Guys, GPT-4 has no eyes. It has no visual cortex. It can't visualize jack shit. It was trained primarily on a massive corpus of text. It's not even remotely conscious.
Did mankind just build a fucking Scrambler from Blindsight?
EDIT 2: When I told it that the apple was floating and the light source should be off to one side a little so we can see the shadow when it's touching the table, it fixed the code.
Again, what in the absolute fuck.
I remember, about ten years back, on SB, I said that by around 2030 or 2040 or so, someone would be able to describe a video game, and an AI would spit out the whole damn thing. Code, art, music, voice acting, story, everything. People said bullshit, it wouldn't happen.
It's looking like I wasn't off the mark at all. Imagine what it would look like if they took the AI-driven tools we already have and integrated them all together right now. Imagine where this shit will be in seven years.
Last edited:
Cool. Think it'd have a diagnosis for my weird magnetic anomalies?ChatGPT has now passed the Medical Licensing Exam. scoring higher than many human doctors and performing difficult diagnoses in seconds. And apparently, its bedside manner beats out a majority of human doctors.
The newest version of ChatGPT passed the US medical licensing exam with flying colors — and diagnosed a 1 in 100,000 condition in seconds
ChatGPT's latest software upgrade, called GPT-4, is "better than many doctors I've observed" at clinical diagnosis, one physician said.www.insider.com
AI Clones Girl's Voice in Kidnapping Scare
Bassoe
Well-known member
AI clones teen girl’s voice in $1M kidnapping scam: ‘I’ve got your daughter’
“You call the police, you call anybody, I’m going to pop her so full of drugs,” the mysterious caller threatened.
nypost.com
Yinko
Well-known member
In a few years it will beDriving myself crazy trying to remember the title of the airport paperback technothriller that predicted this.AI clones teen girl’s voice in $1M kidnapping scam: ‘I’ve got your daughter’
“You call the police, you call anybody, I’m going to pop her so full of drugs,” the mysterious caller threatened.
"Please, send us a finger so we can confirm you have our daughter! All these AI generated videos, we can't tell which groups of kidnappers has her."
Imagine how hard this would be for Star Trek, with people using transporter clones to counterfeit kidnapping victims.
GPT-4 Able to Understand Jokes and Humor
Multimodal GPT-4 (which is trained on images) is scarily intelligent. It can actually interpret the humor contained in images, with or without captions, and break it down with an explanation.
DarthOne
☦️
…ever got the feeling you are trapped in a car while the driver is steering directly over a cliff?Multimodal GPT-4 (which is trained on images) is scarily intelligent. It can actually interpret the humor contained in images, with or without captions, and break it down with an explanation.
Because I have that feeling.
Bigking321
Well-known member
Every day.
All the bad stuff coming is easily predictable and easily avoidable but the idiots keep accelerating straight towards it while lying about everything.
...sigh.
hyperspacewizard
Well-known member
Look ai is a genie out of the bottle if you won’t use it some else will the only way out is forward. Nukes were and still are scary but not having them when others do is a huge risk it’s the same with ai we have to keep researching it And pushing the limits. The first agi or ai that’s close enough has to be ours or an allies as Bad as our elites are there are still way worse out there
This is pretty cool but I’m also super looking forward to drone wingman programs that may be out because an ai air superiority fighters could do moves that would kill a human pilots.
This is pretty cool but I’m also super looking forward to drone wingman programs that may be out because an ai air superiority fighters could do moves that would kill a human pilots.
DarthOne
☦️
Wrong.
It is complete possible to put this back in the bottle.
You just have to be willing to do what is necessary.
hyperspacewizard
Well-known member
I’m sure this is a joke but what would be necessary for that would be basically deleting whole sections of the internet burningWrong.
It is complete possible to put this back in the bottle.
You just have to be willing to do what is necessary.
a large number of books and killing a vast swathe of programmers and the cooperation of all the worlds nations which is the most laughable part. it’s impossible to stop we and our allies need to be the first in this race.
DarthOne
☦️
Nothing that extreme. It’s more of a matter of putting a hard stop on any future development and rolling back AI development by, say, a few years. And making sure it stays that way. By any means necessary.
And why do we ‘need’ this technology? Whose going to develop it futher? Russia and China spend most of their time stealing our the West’s digital technology just to keep even. And if they or any other nation aren’t willing to cease Artificial Intelligence development…well, we make sure we hammer them flat before they get a chance to.
Advanced Artificial Intelligence is a death sentence to the human race. And to anything remotely like a free society. It must be put to a stop.
hyperspacewizard
Well-known member
That wouldn’t work if the west does what you suggest do you not think the chinese won’t bribe or hire the now out of work ai researchers. There no political will for what you’re suggesting it not realistic. The nations of the world aren’t going to go to war over ai the people in charge barely understand it shoot the people on the bleeding edge of the research are surprised at how fast it’s moving. Plus a huge amount of the research is open source it’s everywhere now.
this isn’t stopping ever we just have to adapt
on the need front ai has already shown it can be used to diagnose cancer well before humans can and who knows what advances in medicine or other fields it will eventually produce. Again I compare ai to nuclear research we could power the world if the nations wherent overwhelming terrified of fission power and all the good it can do.
DarthOne
☦️
That’s why you kill the AI researchers and destroy their work.
hyperspacewizard
Well-known member
Yeah no thats insane incredible immoral and isn’t a real solution anyway because you wouldn’t get all of them or it. in reality we are just going to have to live with it and deal.
Bear Ribs
Well-known member
Elon Musk has started his own AI company, X.AI. No news yet on it's goals or what he plans to make with it.
Zyobot
Just a time-traveling robot stranded on Earth.
Plus, if it becomes open-source, you’ll need some massive internet censorship to prevent it from staying up or spreading far and wide. At that point, I think policy implications will go well beyond attempting to outlaw AI alone, since it’d afford the government the power to both crack down on free speech and use it as precedent to, say, go after shareable blueprints for 3D-printed guns, for example.
Personally, I wonder if you could have tech-savvy Libertarians and miscellaneous “Crypto-anarchist” communities inventing creative ways to counter mainstream AI, or perhaps engineer one of their own? While they’re off to a rough start, they’re already making use of cryptocurrency and 3D printing, so if they find ways to live with or invalidate AI that the Establishment will go all-in on, anyway, we should certainly consider that.
Yinko
Well-known member
Like it or not, neo-luddites will arise from this. How influential they are, how violent they are? Who can say. If AI takes a significant enough fraction of the jobs over a 5 year period then there will be mass unrest, no question. In some countries that unrest would probably lead to some combination of overthrowing the government / civil war / total anarchy. I mean, look at France, where they burn down half of Paris for raising the age of retirement by a year or two, and now lay off 15% of all educated workers.
Multimodal GPT-4 (which is trained on images) is scarily intelligent. It can actually interpret the humor contained in images, with or without captions, and break it down with an explanation.
Not sure what this says about me or the AI that I am unable to interpret the humor in those images.
But I do agree with some of the posters here that there is no going back with this technology, much like firearms or any other arms-race technology.
Terthna
Professional Lurker
You first.