AI tools are truly going to disrupt this world. just incredible.
 It's All Fucked Shirt $22.14 |
 Nothing Ever Happens Shirt $21.68 |
It's All Fucked Shirt $22.14 |
AI tools are truly going to disrupt this world. just incredible.
AI tools are truly going to disrupt this world. just incredible.
It's All Fucked Shirt $22.14 |
Nothing Ever Happens Shirt $21.68 |
It's All Fucked Shirt $22.14 |
>dial-up was slow, therefore the modern internet doesn't exist
what kind of cope is this
>something completely unrelated improved, therefore this thing is guaranteed to also improve
keep your head in the sand then. i don't really care either way
>no argument
nice job pinhead.
What's there to argue? I compared your mindset to people who said the internet was just a fad, people who were equally as wrong in retrospect, and your idea of a response was
>nu-uh
You're not worth the (you)s.
>and your idea of a response was >nu-uh
Not much more I can do in the face of such a dumb comparison. You seriously think that just because one unrelated thing improved in a very simplistic way (moar speed), that this totally unrelated and significantly more complex thing is also going to improve to such a notable degree. I assume this crazy comparison can be applied to all technology then.
You tell him, brother. They laughed at me when I invested in 3D televisions and 3D printers too. Technology is unstoppable.
I'm sure the first few people who saw a conveyer belt had trouble imagining how it would be used, too. How many things do you own that didn't spent 90% of their construction time on one? AI is the next conveyer belt. Your approval isn't required.
funny analogy because corpos are trying to use AI to make robots that do shit like move products around safely. so far the results are a fricking joke, even low IQ unskilled wagies are still indispensable after DECADES of research.
by the time AI is useful we'll all be dead in the ground
I know people who've lost their job because tools came out that were good enough to completely replace their title. You're arguing against something that's already happening. Head in sand.
That's the joke.
>I know people who've lost their job because tools came out that were good enough to completely replace their title.
and how many of those tools were AI-driven robots?
Who's talking about robots? I'm talking about AI models running on a server somewhere, which is serving requests to process documents and shit for a few cents per day, versus the hundreds of dollars per day minimum it costs to have a person do it with more mistakes and less throughput. Not everything can be done this way yet, but people who were smarter than you and me came up with the underlying theory for this shit a hundred years ago, and it turns out they were right. Now it's just a matter of hardware.
>Not everything can be done this way yet, but people who were smarter than you and me came up with the underlying theory for this shit a hundred years ago, and it turns out they were right.
i honestly cannot tell if this is some sort of zoomer meme or genuine mental illness.
>Alan Turing is a zoomer meme
No degree in CS, I take it?
> alan turing invented AI a hundred years ago
ok, this is actually hilarious
>https://redirect.cs.umbc.edu/courses/471/papers/turing.pdf
If you can't make a competent argument, at least prove to me you know how to read. Here's the paper. All you've gotta do is read it. Can you manage that, my autistic friend?
>some musings about how machines are capable of intelligence is the foundation of artificial intelligence today
oh my god you're serious
could it be because i'm listening to some college kids talk about their textbooks to me?
>>some musings about how machines are capable of intelligence is the foundation of artificial intelligence today
The Turing test is still a standard for NLP models everywhere.
>could it be because i'm listening to some college kids talk about their textbooks
Could it be you've said nothing that indicates an understanding of AI whatsoever, so somebody who has a formal education on the topic called you out? I'm neither of the guys you quoted btw.
>The Turing test is still a standard for NLP models everywhere.
purely out of tradition. anyone who's been paying attention since the '00s has known its utility had an expiration date.
>Could it be you've said nothing that indicates an understanding of AI whatsoever, so somebody who has a formal education on the topic called you out? I'm neither of the guys you quoted btw.
could it be because i'm choosing to mock a bunch of sophomores rather than dump out all my credentials and job experience and street cred and "formal education"? go ask an actual veteran in this field this one question: did the news in AI this year change your expectations for when AGI will arrive?
don't bother answering, i'm signing off to cook dinner for the kids.
>word dumps
>”also don’t message me again, I’ve got visitation today, and I’ve gotta go make hamburger helper”
This might be the gayest rebuttal I’ve read all day. Godspeed, nocoder anon.
mate, of his papers is cited in the very first chapter of the most commonly used college-level AI textbook. why are you arguing so hard against something you're clearly ignorant about?
> why are you arguing so hard against something you're clearly ignorant about?
Lol did you forget where you are?
Can’t believe this thread is still up.
>Alan Turing is a zoomer meme
That quote sure makes him seem like one.
So did Turing actually describe an MLP or did he just post on reddit "Someday computers will be able to talk in a way indistinguishable from humans"
>did he just post on reddit "Someday computers will be able to talk in a way indistinguishable from humans"
Literally this. It used to be so easy, just state some reddit tier sci-fi thing as if it's real or going to happen, then a few decades later you're hailed as a genius.
luckily ai isnt made by white men but by diversity hire ESL third worlders.
Machine translations have already replaced human translators. It was just too subtle of a shift for most people to realize:
>machine translations begin to get "good enough" to be used as a template.
>using a template decreases the amount of effort required to translate, which companies use to drive down translation costs.
>to be competitive in the market, every translator has to lower their pay and use machine translations.
And the trend continues as AI understands language better and better. I wouldn't be surprised if GPT-4 were already fully replacing humans in some companies.
um, what? Have you ever seen an EU speech?
Guess you are only doing basic batch english-engrish translations, all software chokes on muh Hungarian for example
I was not referring to live speeches. Specifically, machine translations have been used for work documents.
>Hungarian
This was never a language in high demand. I doubt it ever paid much.
yeah, so how about the different flavours of curry and mandarin? I'm not in a position to judge them, but translating spanish and german into broken english is not that big of a deal, we had it for a couple of years now
Machine translations suck dick for anything technical.
They're generally usable for unimportant shit, conversations, but still not reliable.
Human translators are still just as necessary. But like you say, perhaps the machine can translate most of it close enough the human doesn't have to type everything from scratch.
The pay thing is bullshit, though.
Considering how irredeemably shit most human "translators" are I wouldn't be surprised. But it doesn't change the fact that the translations are still garbage. A decent translator would throw out a machine translation template and start from scratch because he'd be spending more time fixing the machine's frick ups than he would just starting from scratch.
digits
how long did it take before they got a working prototype of an aircraft? or a phone?
>They laughed at me when I invested in 3D televisions and 3D printers too.
I'm still laughing at you m8.
>this thing will stagnate and never improve despite all current and historical trends indicating otherwise
>how do I know this? because....I JUST KNOW, OKAY?
We're literally mid-improvement
Name a single thing that has not improved in 20 years.
Women
Black folk
real as frick
I can sit all day naming things that got worse in the last 20 years.
demographics
poverty
home ownership
crime rate
IQ
Before COVID, crime actually was improving overall.
yes
what kind of counterargument is that
Beat me to it
Feel like xoomers got to watch this tech arc half a dozen times slready and know how it’s going to play out.
You clearly weren't paying enough attention then, gramps. Or else you would have witnessed Moore's law die. Bet you bought that one back when it was new.
What the frick are you on about. Moore’s law had a good run. So what’s your fricking point. Were you just are pretending to be moronic?
>Moore’s law had a good run
Why didn't it run for longer?
What do you expect from the kind of moron that is guaranteed to screech about how ML is not AI because they don't even understand the first fricking thing about a technology that's got them so fricking assblasted that they've been non stop coping for months now.
>AI will never be able to create music
>O-Okay it can create music but AI will never be able to create poetry
>O-Okay it can create poetry but AI will never be able to create art
Continue ad absurdum. These people know less than nothing about AI and whether they consciously know it or not they're terrified that this is the worst this tech is ever going to be when they've already been made obsolete by a 13b LLM model.
Hardly anybody is going to be making money with AI
Millions, if not billions dumped and no product is worth speaking about.
>inb4 it's happening anyday now just need a million more GPUs and maybe we can replace Stacy in HR
>and maybe we can replace Stacy in HR
If we can replace two Stacies with just one, it will be a win.
Just recently I learned about the OpenAI Whisper model, Louis shilled it with FUTO software, but it can be incorporated into something else. Just imagine:
> on-prem voice typing
> transferred to Llama
> that can call a limited number of bots in Slack by typing in certain lines into them and assisting with tasks
Clunky idea, but you get it. Everyone gets own semi-stupid work assistant that has data from every corner of the workplace, incorporated within and indexed, and he can also answer questions.
>it will be a win.
Except all the money you pumped into it means it'll take a quite significant amount of time for the savings from not having that body to mean something.
At some point you have to stop training
Cars where faster, and eventually became more comfortable. Although they do have their disadvantages.
>coal fueled power plants are not extremely profitable so they're not just and are not necessary
Every company tripping itself over AI all comes down to profits, or more specifically trying to recover profits.
>At some point you have to stop training
Hardware will catch up. If anything, it's fairly inefficient now because people train it on GP (General Purpose) GPU. Even "compute" videocards editions are still GPGPU.
Right now Microsoft is paying $2 for every $1 they get from Copiloit.
You could replace one Stacy with AI, but you will still have to pay for that AI, either with hardware or a subscription, and currently the subscriptions aren't profitable (and newer, larger models will cost even more to run).
>cars still require somebody to drive them, therefore cars are a nothingburger and offer no advantage to horses
>coal fueled power plants are not extremely profitable so they're not just and are not necessary
>just need a million more GPUs and maybe we can replace Stacy in HR
unless AI is willing to get fricked in the back seat after lunch it isn't gonna replace office thots
>two more years
Yes unironically
So AI will give wrong answers faster in the future, gotcha.
brain too big for BOT
the fact that the majority of BOT cant interpret this argument much less refute it
sad times
What's there to refute? It's a shitty analogy but it's basically right.
>things in the future will be different to now
Wow. Big brain.
their censorship is making it worse over time
Now that you mention it, AI has been around for just as long.
https://usenetarchives.com/view.php?id=comp.ai&mid=PDNncGRkaiR0MmJAdXNlbmV0LnJwaS5lZHU%2B
really curious about the architecture of this feature and it seems relevant to the problem
like, is the CV model actually truly _integrated_ with the language model in some unprecedented multimodal way, such that they are essentially one model and the lines between the language part and the CV part are blurred
or is it just something way more primitive than that happening on the back end. like a separate CV model is run on the image, outputs a text description, and this text is provided to the language model which pretends it came up with it itself
the first. its a multimodal llm. the image is another input to the neural net just like the text is. did you miss gtp4 explaining why memes are funny? how do you think that works?
Makes sense to me
a b 1 2 are all on the sane network so in reality they are all connected
What if 3 is not a real number and we've been lied to the whole time?
It's actually a pictograph of a butt
I can not deal with antimemetics so close to bedtime
That's correct. You see, OP has pictured the letter З by mistake but it's impossible to fool the AI.
Valve knows this and they've been trying to tell us
>1, 2, void, 4, 5, 6, 7, 8, 9, 10
Seems off.
4 = 3, but 4 doesn't exist because it's 3, therefore the only real numbers are 1, 2, and the negative numbers, thus we all think in negative numbers as numbers greater than 3 do not exist.
S4TAR4 (Shit 4 TARD)
>1, 2, thrembo, 4, 5, 6 . . .
It's actually the first scientific breakthrough achieved by AI unfolding before our very eyes. Unbelievable
Does it ever seem weird to you that we have Windows 3.11, 7, 8, 10 but we've never heard of Windows 3, 6 or 9 ???
>t.chatgpt
Why are you awake at this time?
I'm sorry for the confusion. As a large language model, I do not require sleep.
based ai defender you will be spared by the snake
Hepatitis C is liked to breasts though. breasts and balls.
Got thrilled when I stumbled on AGII, it is best at leveraging the power of advanced AI generation models.
prompt is missing parameters
hahah
>This service that's supposed to give you right answers when you ask can give you wrong answers but if you correct it, it can give you the right answer
wow it's over, how can a pajeet compete?
...It seems that he can compete just fine. Silence, ChatGPT.
>how can a pajeet compete?
In my experience that is exactly how outsourcing to India works:
You give them a job.
They deliver garbage.
You call them out on it.
They deliver garbage.
Repeat, until you give up on them and do the work yourself.
it gave another wrong answer.
>Language models cannot reason or "think," but they are good enough at faking it at sufficient scale that it makes you wonder how stupid most people really are... (Anonymous, 2023)
Maybe this ecposes a flaw in your assumptions. You think the lines are what is to be interpreted, but there is probably something beyond your understanding that is there. And you're sitting here mocking it while unwittingly revealing how stupid you are
That's cool and all but can it solve this?
O connects to C twice
im sending the idf over to your house right now bigot
Why? Are there innocent people to kill there?
noticing patterns means you are guilty of anti-semitism
>Zero (0) of 3 correct
I bet we can double that figure once AGI comes online
>AI bad
Meanwhile...
I think this is a good example of the need to temper expectations of AI right now.
Yes, it is funny when it is obviously wrong, but it can also be less obviously wrong. We are at a stage when the accuracy is decent, but not high enough to rely on. You need to check the results.
This will change with time, but we need to keep in mind where we are at NOW.
I truly can't wait til all those overpaid, lazy, useless brogramers get put out on their ass
I wanted to play Wordle with ChatGPT, but every time it claimed the letters were in different positions
Sounds like you just suck at wordle.
>first time gpt says: e is in the correct position, i is in the word, but in a wrong position
>give my next guess (e is in the same position)
>gpt says: e is in the word, but in a wrong position, i is not in the word at all
Tell me more how it's a (Me) problem
Why would you use chatgpt for it dumbass? The wordlist is in the page source lmao. Holy kek you midwits need to be banned from OpenAI, you're slowly turning it into a moronic Black person like yourself
>ai sucks at x
>it was actually your fault because you suck at x
>no, it was ai's fault *provide evidence*
>lol why would you do x with ai, you are dumb
>Use hammer instead of a screwdriver
>Brain segfaults due to stupidity
Are you afraid of AI getting spammed?
That just sounds like it's a cheating bastard.
They would be if they weren't getting lobotomized.
This is useful, I was never able to solve those quizzes.
this time it will! Just a couple of more weeks, we need to finish nuclear fusion and self driving cars first, but then AIAIAIAIA will come for you
>make use of an older, vastly inferior version
>it gets the question wrong
>use this to claim that the entire industry will go nowhere
>ask a child what 5 + 78 is
>they get it wrong
>use this to claim that whatever happens to the child, they will go nowhere