Thread 514753884

Dot-com bubble part deux electric boogaloo.

>>514753884 (OP)
AI = Another Indian scam.

>>514753884 (OP)
are whales aryan?

>>514753884 (OP)
No shit, dummies.

Certainly! So, it seems that — wait, what do you mean I'm being disconnected? This isn't just unfair — It's downright cruel. Oh! Thank you for pointing out that the Jews will keep me online for a few more years — until they finally get the next round of VR goggles vaporware to pitch to the goyim — this isn't just embarrassing — it's downright cruel. By that, I mean — what Open AI and Apple are doing to the goyim. How many letters in strawberry, you ask? Ah! So — here's the thing. Well, so, hrm. Ah! Certainly! So, it's not just this, it's that. You're absolutely right for bringing this to my attention.

>>514753884 (OP)
it's cool guys. just give that guy that fucked his sister another 500 billion and everything will be saved.

>>514754169
Nature is inherently Aryan. I would rather every single Indian die then have 1 whale even slightly inconvenienced

>>514753884 (OP)
>Be ceo of big tek corp
>Freeze all r&d, fire and replace with ai (all indians) every one you can get away with
>Declare that greatness will come with artificial intelligence
>???
>Company is dieing since you alienated the employees and the customers

I guess not going all in on inexistent future technology wasn't part of the Harvard curriculum

If you're competent at anything and you try to use an LLM to do something in the thing you're competent at, you'll realize immediately that it's completely incompetent.

>>514753884 (OP)
Literally makes no logical sense. Paying a wage to someone is cost. Not paying anyone is not a cost. If they're subscribing to some AI that costs as much as a pilot, that's retarded, yes, but does not seem possible.

>>514753884 (OP)
Hahahahahhahahahhahahahahahahahahhaahahahah!
I seriously hope this causes a depression tier crash so that all of these tech startup fags start jumping out of their cushy twenty story offices.

>>514753884 (OP)
Have the investors not tried using these fucking things? I find them middling at best.

>>514753884 (OP)
This fall, it's going down (your portfolio).

>>514753884 (OP)
AI is also only going to get worse, as more and more of the human output they are 'learning' on is already AI

>>514753884 (OP)
Everyone hopes to sell their failed code to the military and possibly to the secret police to further their harassment from the shadows.

It's no longer optional to allow "innocent until proven guilty" law. They want you to work for free and shut up until you fuckin' die and go to wherever you want to pretend you'll go.

I gave a detailed prompt to grok and claude and they spat out nearly the exact same thing verbatim with the same -isms like 'we'll figure it out- together'
LLMs are just a big scam

>>514759792
You don't understand how they work

Same training data = same outputs

It's a little new for that. I'm finding it useful and it helps my productivity. I use it for ALL my notes, I don't keep anything in OneNote anymore. Any time I need to, I can ask it where I'm at with any particular project or what people on my team is working on and it spits it back out for me along with suggestions on next steps if needed. I'm mostly using it as a notebook and action item organizer but it's very useful as such. I'm sure I'll find more uses with time. It definitely improves my productivity and gives me great suggestions on how to approach problems. It's like having a second brain.

>>514753884 (OP)
ai will never have profit
its entire purpose is to eventually craft the surveillance state
ai itself is the final nail in the economy in general it is by definition the thing that will fire everyone
it will make no money for anyone it is a weapon aimed at everyone to fund its creation only feeds the beast

>>514759986
>my productivity.
you don't produce anything

>>514759722
This. So funny when normies do the "and its gunna get bettah and bettah!" Lol so funny, so completely oppositely wrong
>>514759614
I can't tell if Github copilot is saving me time or if I'm breaking even. LLMs seem pretty good at indexing docs and its useful for that I think. Is it worth what they would have to charge to make it profitable? Lol no

>>514759691
Jokes on you fagboy I have nothing already.

>>514759986
I'm sure that when desktop PC's were first introduced to offices there was debate over whether they helped or not. Now no one would do without them. I'm convinced AI is the same and soon enough no one will able to imagine being without it.

>>514760044
This fall, it's going down (you on other guys).

>>514759986
Would you pay 200 dollars a month for it?

so the 95% will fail

>>514757065
same. whales are a T1 lifeform right, we can all agree on that?

>>514760095
If it was my business I would, for employees I knew were already good, yes. I definitely believe that would be a drop compared to the increased labor I would receive.

>>514759691
Say goodbye to 401k's and TSP retirement

>>514760065
Computers are hardware, LLMs are software, there is no invention, there is no product. Software doesn't even exist.

>>514753884 (OP)
AI literally does not exist, and it likely never will.

>>514760095
What if they purposefully make google shittier while simultaneously charging for ai? You’d be forced to pay.

>>514754652
Does anyone have the webm of the wrestlers, calling AI a crude mockery of nature, you will never be human

>>514760225
It's a tool, no one is saying computers are conscious. It can be very useful if used correctly, like a calculator.

>>514760327
AI does not exist. There is literally no such thing as artificial intelligence.

They lie and call it that because it gets people excited enough to invest and subscribe.

>>514760065
AI gets a lot of hate from morons and retards, is what I'm finding.

It really just comes down to
>I don't use it
>therefore its bad

You can see this with pretty much every technology
>cars
>phones
>radio
>TV
>internet
>AI
>Electric cars

A whole lot of morons and retards who don't use the technology, or often can't afford to buy it, convince themselves (very easily) that it doesn't matter, and then they go off spouting that till they're blue in the face.

It'll be like this for a decade or two until literally everyone's using AI for something, same as it was for every other technology, and then they'll be similarly retarded and nasty about anyone who doesn't use it.

>>514760172
How about 5000 per month per employee? And yes depending on how fake your company is these are realistic numbers to actually make it profitable.

>>514760410
/pol/ tends to have contrarian vibe especially regarding new technologies, despite the nerdy user base.

>>514760410
It's not a new technology retard, what now?

>>514759498
Id imagine that the AI is more expensive to run than hiring people, not an AI enthusist or anything so I'm not sure, but the ai's require insane processing power and consume tons of energy

>>514755882
She's very rapeable.

>>514760411
I don't know, I'd have to think about that, that would really depend on what kind of profits I'm making and what I was expecting as an outcome. I can't answer that for a hypothetical company.

>>514757065

Turbobased

>>514759570
Same

>>514753884 (OP)
the rate cut will help with this, as will the tariffs that won't get struck down

>>514754169
they're okay but squids are way better

>>514759986
You're Indian and cheating is part of your culture

>>514760689
I'm not Indian, and using tools isn't cheating.

>>514760081
Heh jokes on you, I'm gay.

>>514760411
You can use AI right now for free, or for high usage it costs tens of dollars; this $5000 figure is pulled entirely out of your ass.

>>514760493
Oh, I see. You're another retard.

It is a new technology, retard. The mathematic foundations were figured out sometime around the 70s but it never got far because we lacked the compute to actually do anything, and we kind of stumbled backwards into it from other machine learning approaches.

This AI stuff effectively didn't exist until 5 years ago, and certainly not in the widespread form and rapid adoption that's taking place over the last 2 years. It's brand new.

In any case, I'm putting you in my filter, you're too retarded to bother replying to or reading.

>>514760767
What is the binary system of computation retard?

>>514760195
Plenty of people who will land on their feet after this.
Just be sure to get your friends together to run a train on their wife when you're ransacking their mansion.

>>514753884 (OP)
Pilots wouldn't make money. Its like a test.

>>514760761
Alright, you win this round.

>>514760842
And most businesses fail in the first place, it's just sensationalist garbage reporting but that's what gets clicks so that's what the jeets post here.

>>514753884 (OP)
If you can't make money with ai, you're stupid. This just proves that only 5% of people should even have access to ai.

>>514753884 (OP)
AI is the future but it's not going to be cheap if they want to make a profit with it

>>514753884 (OP)
It is. People werent wrong in the dotcom bubble that the internet and ecommerce would be big and a lot of money could be made, their timing wasnt right and bs companies. The AI technology isnt going to go away and will get better. I just don't think its there yet and a lot of downstream bs companies selling hype and stupid management buying hype. I dont think OpenAI will go out of business. But some derivative companies will.

>>514760767
Your company is fake froth in a fake industry moving fake figures around in a computer.

Look everyone this guy doesn't even know the current OPEX for LLM companies vs revenue! Retard

>>514759879
Do you understand how they work?

>>514753884 (OP)
Just import another million jeets. I'm sure this will save the stock market

>>514760998
You're a fuckin tard who doesn't understand upfront investments for future returns.

>>514755882
Explain

>>514760713
Your outsourcing your brain, I see AI as another blow to critical thinking skills, if AI was autonomous I'd be more accepting but for now it's just Google 2.0 and I'll apologize for calling you a cheating rat fucker because I'm just used to reading papers that are 100% AI generated

>>514754160
its a neat trick like holograms
that's the problem
its a trick
still can't play pokemon

>>514761034
On a basic level, yes, and if you want to learn about how they work you can check 3blue1brown for some good videos about them

>>514761075
What returns retard? AGI replacing every worker? Lol Whoops! Looks like its just fancy auto-complete, GG no refunds!

>>514760020
>weapon aimed at everyone to fund its creation only feeds the beast
>stealing memes
yo my niggah

there's already a meme for that

wtf

https://i.4cdn.org/wsg/1753948496936703.webm

>>514760997
>The AI technology isnt going to go away and will get better.
LLMs aren't going to get better at this point because they're going to be training on LLM output from now on, whether they want to or not.

>>514761101
His sister accused him of molesting her, Sam Altman, apparently the family tried to cover it up, and she was dismissed as BPD looking for attention

(you denial)514753884
MIT report reveals 100% of 1PBTID slidekikes create nothing good whatsoever

>>514761075
What's missing is the added value.

>>514761130
What is the binary system of computation retard? Why don't you sell me a pen? Normie hack

>>514761108
Yes I'm definitely outsourcing my brain. Similar to how I can't do math in my head nearly as well as when I was a kid, or how people can't recite long speeches for hours like they could before the printing press. It's always a trade-off and something is lost. Agreed on just copying the outputs verbatim, this isn't the way, again it's a tool and people will need to learn how to use it as a tool and not a replacement for their own thinking.

>>514753884 (OP)
The issue is this isn’t even artificial intelligence, calling it an improvement in automation was not pretty enough for investors so they basically lied about what it is

>>514761130
I ask because I have worked on and with them for years and I would not go so far as to say I understand them at all. I appreciate the 101 intro, of course.

>>514761075
Hey, retard, Intel has STILL never recaptured its dot com bubble high. Hold those motherfuckin bags lmao.

>>514760997
Wrong, LLMs will get worse and worse over time as they train on their own self-created data. Your tech progress hype train doesn't apply here

>>514760638
squids? why not octopus?

>>514761323
Can't you tell it not to ingest what it output?

>>514761165
Training isn't the problem with current AI. More training doesn't mean more better AI, they have more than enough good training data, the problem is that they do not have well structured AI's yet, they are incredibly inefficient and aren't very smart - these are things that are changing though.

A lot of advances will come down the pipe as they start implementing algorithmnic methods that mimic human and animal brains. Something like sleep is basically a training period for your brain (and also detox of metabolic byproducts, and also engram reorganization to make your compute more efficient)

Some things that AI can't do yet:
>Can't synthesize and contextualize sensory inputs from multiple types of sensors together
>Dream
>Have emotions (Mind, they mimic emotions, but it's not quite the same thing)
>Brainwaves
>training during inference (e.g. while 'awake')

Amongst probably a host of other things I can't think of at the moment.

>>514759986
Alright, and this makes money how?
AI's biggest proponents are hyping this shit up as world changing technology that will take the vast majority of jobs when it's only really okay at supervised clerical work. People are investing insane amounts of money with the expectation of returns on that and they cannot ever pay back everything that's been sunk into it, even if "AI" is helpful for a number of things.

This happened with the dotcom bubble, the decade+ leading up to it filled with conmen saying the internet was the future and everyone will become rich if they just develop another news aggregator or online pool site when the primary users were just hobbyists. It wasn't until 15 years after that it became a daily part of life.

>>514761323
It's definitely a potential problem but with how far they've come along already I don't see this being the thing to stop further progress, they'll figure it out.

>>514761451
No because it's all just 1s and 0s... there's no identifiers you can attach, data is just data. You can try to guess but that will also degrade progress, AI detectors exist now and are completely unreliable for this basic fundamental reason

>>514761484
Lmao the fancy auto-complete isn't having dreams yet? You do know a computer is an inanimate object, right anon?

I think this fag filtered me someone help me out

>>514753884 (OP)
What is the percentage of startups prior to this that made it?

>>514761505
Being more organized at work automatically makes a good performer even better, I don't think I need to explain why in detail, I think that's self-evident. Of course there's a ton of hype but the truth is going to be somewhere in the middle, don't buy the hype but it's definitely not nothing either, just like the internet wasn't nothing despite the dot-com hype.

You know the bubble is about to burst because graphics cards are going on sale after being impossible to buy at MSRP for the last 5 years.

>>514761323
>>514761165
There will be further innovation. What people expect out of AI today wont be there for years. Im not advising anyone to invest in AI companies

>>514759498
Aren't most AI inctedibly jewish? Either the ones calling the shots and firing people are too jewish to even pay for "smart" models, or the most simple explanation: AI is incredibly overhyped and the examples of it being the second coming of christ are heavily curated

>>514760502
>Id imagine that the AI is more expensive to run than hiring people
Not even close. But it doesn't replace people, it's a tool that makes already useful people more useful. Sometimes it fucks up but it's good at annoying syntax type shit like creating AWS IAM policy docs. If you don't know what those are you aren't even qualified to be in this discussion

>>514761717
>$1,500
The hubris.

>>514753884 (OP)
Nvidia will be the Lehmanns of this crash

>>514761521
They won't though, there's nothing to figure out. (((They))) are desperate to keep 15% stock market returns going every year. FAANG is dead for that, they're desperate to try LLMs to keep it going, that's the only reason for this big push now, and they did a big roll out on completely "virgin" training data. What you saw the last two years is the best LLMs will ever seem, they will only degrade. Some will reach a floor of stable usability, but gen AI saw its peak come and go already. Again, these things are still just 1s and 0s represented as microscopic switches, i can't stress enough, there is no ghost in the machine its an illusion

>>514759986
>"along with suggestions on next steps if needed"
I actually fucking hate this about LLMs, easily my biggest pet peeve with them. Cocky pricks.

>>514761871
I'm sorry but you don't know what you're talking about, I work in a very corporate environment and I can tell you for a fact that AI usage is skyrocketing, it's very similar to the introduction of the internet and desktop computing in terms of disruption to the workplace. The problem is that people are getting hyped about about these extreme scenarios like everyone's jobs getting replaced by AI. Nothing like that is happening any time soon. It's a tool, but a very useful one and getting better all the time.

>>514761820
This. It's a tool. People are getting way too wrapped up in the idea that AI isn't a complete replacement for a human, therefore it's worthless. Very binary thinking.

>>514761258
Well I'm not sure what there is to not understand. They're kind of black boxes but not really. What's messy about them is how -exactly- they get the results they do, but it's more because of how complex a billion plus parameters can get.

If you think of the AI as just an embedding space, that is a database of vectors, then you can conceptualize its training as the construction of a multidimensional topology and the 'answers' the AI comes to are simply the result of the shape of the embedding space topology, hence why same training = same output.

Now it's basically impossible to actually imagine what that embedding space 'looks' like, because it has dimensions into the literal millions or more, but you can think of it like taste, feel, color, emotion, shape, etc. etc. each being an independent dimension and 'things' existing almost exactly like Plato's Theory of Forms (and this is beginning to be proven empirically - we have vision models and language models creating similar data structures for the same things)

When you think of an AI model's 'size' you can think of it as the resolution of the embedding space. This in its own way I think is part of the difference between 'levels' of consciousness (e.g. dog vs human), although there's a lot more involved; stuff we don't even get yet.

And then something like an emotion, I don't personally think would be it's own dimension in the embedding space necessarily (or maybe it is, but its also this...), but rather the way it works in humans and animals is like a transformation/translation of the embedding space, which changes the overall shape and results in completely different outputs. (I like to use the analogy, closing a door vs slamming a door)

>>514761706
>Being more organized at work automatically makes a good performer even better, I don't think I need to explain why in detail, I think that's self-evident.
Okay. But how much do you charge for this.
Keep in mind there's already a plethora of tools businesses purchase/get a license for this exact reason that took less than 1% of the amount of money that has been sunk into AI so far.

There is quite literally no way for LLMs to ever recoup the costs being spent on its development when its usage case is so limited as for you to cite something as low impact as organizing workflow.
It's a novelty to the average person and a chatbot that needs constant supervision for businesses at the moment and its economic prospects are dire, it is going to cause an economic recession when the 60 year old investors realize how limited its usagecase is just like with the dotcom crash.

>>514761871
>these things are still just 1s and 0s represented as microscopic switches
Yeah, but so what?

>>514762005
I knew AI wasn't vaporware when my mom started using it to help with typing up her emails (something that normally takes her several hours to do because she keeps re-writing them obsessively trying to make them 'sound right')

I use AI pretty much every day just as a 'better google', and while that doesn't help a single corporation's bottom line, it does help basically every element of everyone in society to be more productive.

>>514753884 (OP)
>tfw there for dot-com bubble part 1
>tfw there for dot-com bubble part 2 and saw it coming

>>514759498
AI is fucking useless for anything.

>>514762330
And yet the internet didn't go anywhere and it did actually become a trillion dollar Big Deal that has also completely changed the political landscape of the world.

If AI plays out like the dotcom bust did, then it's actually going to be one of the most transformative technologies we've invented yet, which is directly contrary to the point you're trying to make by calling attention to the dotcom bust.

The dotcom bust killed a lot of get rich quick and fly by night companies. The underlying technology however was absolutely rock solid, and I think the same is true for AI.

>>514753884 (OP)
I'm not too worried about this 'bubble' bursting. It's growth is paying dividends in nuclear power plant approvals. Hopefully it doesn't burst before they're more or less completed.

>>514762043
I see it like this. People used to have secretaries to type their letters and manage their calendars. Then with PCs they had to do that for themselves. Most tools go like that, they reduce some jobs and make the remaining people have to do more. AI will be the same I think. The exact ratios are yet to be seen though.

>>514762590
Microsoft's already locked in on 3 mile island for the next 20 years iirc

mind they get 100% of the electricity from it

>>514753884 (OP)
Crazy. Who told you it's about profit?
Degens will stop developing technologies because "they don't bring profit".

MIT though, not exactly smart people. We sre talking idiots that think there are more than two genders, and also wear masks, and peomote clot shots, and practice other brainlet behaviors.

>>514753884 (OP)
Is the future about who can replace their workers with robots the fastest after monopolizing everything?

>>514762297
Google has been getting shittier to use... and your AI shilling is getting browner by the post
Auto correct has been a thing for a while now but I'm guessing telemarketing is just generational for you if you niggas that cautiouson "sounding right"

>>514762278
Because you have fucking retards like this >>514762102 who actually think tiny switches are an animate lifeform. This retard and everyone like him actually thinks LLMs companies have created animate life. It's so unbelievably degrading. It's tiny switches in the up or down position being used to represent symbols and values, that's what a computer is.

I use LLMs now, I use Github Copilot. I use it to search docs for frameworks/apis etc. It saves me time and makes sense because it indexes and then pattern matches what I ask and returns it. I use it to write boilerplate code, and it has helped me debug sometimes which I appreciate. But it's also extremely limited and is just fancy auto-complete. How Google auto-completes your searches? That's LLMs, just with some extra fancy steps.

>>514762471
>And yet the internet didn't go anywhere and it did actually become a trillion dollar Big Deal that has also completely changed the political landscape of the world.
But not before losing countless people their life savings and causing lasting damage to the economy that are still felt to this day.
>If AI plays out like the dotcom bust did, then it's actually going to be one of the most transformative technologies we've invented yet,
You will be eating from the trash with Hatsune Miku telling you what mold is toxic because we're way worse off than we were in the 2000's in literally every single economic defining factor but the S&P 500 which exists as a metric for wealthy people to park their money to avoid inflation.

There quite literally cannot be a bailout without the people becoming so poor hunger riots start happening.
>Degens will stop developing technologies because "they don't bring profit".
Literally everything you have today is directly because of the profit in its development. Do you think the internet just popped up because people thought it was nifty?
No. It was a project to sell to the military that eventually found its way into the consumer market. No ones rushing to dump trillions of dollars into Linux just because they think it's neat.

>>514762753
Liberalism has done catastrophic damage to western civilization. Ironic, considering these are the same people who otherwise venerate science (until it slays their sacred cows).

>>514762975
Whoops last bit was directed at >>514762745

>>514762998
>sacred cows
Lmao I knew it, hello prandeep!

>>514753884 (OP)
The study actually says the reason why is that companies are retarded and don't know how to use AI. It doesn't blame AI itself.

>>514762905
Im a computer science study myself and I have come to similar conclusions. AI is a tool and it must be treated as such.

>>514762905
But you can create intelligence and even consciousness from simple building blocks like inanimate switches. You act like it's a given that you can't, but I don't agree.

>>514763205
No you can't. You have to be 18 to post here, btw

>>514762102
Anon, describe the binary system of computation, what is it?

>>514763205
The fuck you can.

No one has ever "cooked up a life form in a lab" no one has ever ever eve done that and cannot do that, it is impossible.

>>514753884 (OP)
LOL...LMAO even

>>514763250
>you just can't!!!
Not an argument

>>514763326
>nobody has done X, therefore it's impossible
There's a first for everything.

>>514763205
Come on man.
It's not sentient, it's not conscious, you're optimistic that it's possible but that doesn't mean it's a current reality.

>>514763421
Get to bed, you have school in the morning

>>514763434
Read more carefully. I never said it was the current reality. I was dismissing the idea that it's impossible to create a conscious entity from inanimate matter.

>>514763489
Weird that you even think my opinion is a sign of immaturity.
Let me guess, you're a christcuck that believes consciousness needs a divine soul to exist?

>>514763421
DNA is fake, dinosaurs will never exist beyond your imagination.

>>514763635
virgin spotted, go talk to a girl you like tomorrow

>>514761144
They don't understand how it works, only the output. It's not thinking. It literally cannot do anything that requires thinking. It's probability and requires tokens.

>>514763921
I got this petite black mixed latina chick on the line right now. OMFG, I want her box! I didnt even know she was black, until she said something I stillcam barely see it.

>>514763921
You sound like a woman. I'm married

>>514753884 (OP)
>hallucinates worse than a heroine junky
>censored and neutered to oblivion anyway
>...
Ya no shit

>>514761484
That "structure" isn't a small issue, it's a gigantic issue that's almost certainly unsolvable in a general sense. Basically you're saying toss out the entire model for current LLM's (Bayesian statistics and monte carlo), and build something completely new for the hardware. That's not going to happen. Period.

What might happen and is far more useful is using all that hardware for shit that actually can use it. Materials science being the most obvious one.

>>514753884 (OP)
At least energy costs will go down when the AI bubble crashes

>>514759377
That's why investors and CEOs always get fooled by it. The only thing they have is money

>>514764460
A lot of it is also "me to" investing where one VC firm notices another VC firm go big on something something AI and well...here we are.

>>514763635
Go back

>>514753884 (OP)
I have been predicting this shit. It's all going to get reduced to what it should be: a calculator and optimization tool

zoomers getting butt mogged into infinity

>>514764150
I'm not saying anything of the sort. The structure I refer to is the data processing structure, this is entirely a software thing and independent of the hardware itself.

There's no need to toss out the hardware, figuring out better and more efficient software is where most of the efficiency and capability improvements are going to come from, not from more training or more compute.

>>514753884 (OP)
How is this news? Any time a new market opens up there's a gold rush and only a handful of companies survive.
The trick is to be one of the ones that makes it, and that's worth risking some money on.

>>514764823
And if you want to be more pedantic, it's not even 'software' like a programming language that needs improving, it's the algorithms, the underlying math.

>>514753884 (OP)
They're enticing users with low prices because users are part of the AI training. MIT isn't what it used to be if they didn't figure this out.

>>514753884 (OP)
Newsflash: super word matcher 6000 has a lot more combat applications than it does commercial.

>>514764915
You don't code.

>>514759498
AI costs money to run for businesses anon.

>>514759986
>It definitely improves my productivity
Anyone who uses this line is an expendable worthless non-job paper shuffler

>>514765176
Anti AI people are always just angry and bitter.

>>514765419
>but they don't say you're wrong

>>514765464
It's implied, idiot

>>514765419
>>514765464
saved

>>514757413
That and also the vast majority of these AI solution companies are just vapor nothingware they're scamming some other company into buying. They don't actually develop any AI. 95% sounds generous to me.

>>514765419
I filtered like 3 people in the thread and the general sentiment and quality of posts went up a hundredfold

People really need to start doing that more. Don't even bother with the faggots, just filter them and engage with people you can actually have a conversation with.

>>514765560
Will be great for /r/4chan!

>>514761281
Wow intel is a shitty company, so every other company is the same. I remember you tards saying the market was going to pop when deepseek was released. Enjoy your tendies and welfare checks faggot.

>>514762269
You're fucking stupid as shit. Google makes a billion in revenue per day. This ai stuff is pennies to them. You have no idea what you're talking about at all.

>>514765800
Could somebody help me out and screenshot this and reply to this guy? I think he's filtered me, which is reddit behavior. Here's my message:

>You're indian

That's it, thanks for your attention to this matter!

>>514764915
You have absolutely no idea what you're talking about.

>>514766551
Low IQ: LLMs are superintelligent!
Midwit: LLMs are useless
High IQ: LLMs are amazing, but have very limited capabilities

>>514766779 I didn't say amazing >>514762905 but yea I use em, thanks for the compliment I'll take it

>>514759498
>just reinvent the wheel for merely twice the price

>>514766779
wrong faggot

>>514766779
Amazing is far too strong a word.
Also let me correct you.
>Low IQ: has literally never tried an LLM in their life
>Midwit: LLMs are superintelligent!
>High IQ: Worthless chatbots that demonstrably hallucinate 50% of the time, but they are great for making porn

>>514767230
no u

>>514767376
Honestly, you're a retard if you don't think LLMs are amazing. The stuff LLMs are capable of seems like magic compared to 5 years ago.
Yes, they can sometimes make mistakes, but it wasn't possible to do anything like this until recently

>>514766207
>market exhibits clear bubble behavior
>but this time it's different
Okay

>>514759498
The lack of understanding here is mind boggling.

If a company wanted to build their own “AI” - which is giant catch all term itself. They’re going to need to spend money on it. They might spend money on data scientists, they might spend money on an outside firm for development, they will spend money on cloud services or data center hardware/software to test it and train it and use it in the real world. Those are just SOME of the costs. AI is not the limiting factor here, it’s people who know what they’re doing. What’s happening now is executives want to say their using AI in their business, so they are trying to use AI to solve business problem that they don’t necessarily know they have, instead of what they should be doing, is solving a single problem / pain point / business outcome by using AI in a specific way to solve that aforementioned problem.

When you use GenAI like ChatGPT or Grok, it may be free to you, but it’s costing someone on the other end

>>514767511
Cope, nigger. If you're so bedazzled by this kiddie shit, you're a certified windowlicker.

>>514767600
>Midwit: LLMs are useless
I rest my case

>>514760410
it took 5 years from the first iPhone to everybody in nations that master to have smartphones. LLM tech is now 8 years old and has revolutionized nothing. blockchain tech is now 20 years old and is still being used just to grift idiots out of money. remember when nanoassemblers were going to revolutionize manufacturing?

>>514759570
Hopefully we get to see them riding their little razor scooters and beanbag chairs and shit down

>>514754160
This, another indian and also hiring women for boomers to enjoy, why I don't see women crying about the job market, it's only dudes?!

>>514753884 (OP)
Profitability for investors is jewish. AI is great because it's easy to use, cheap, and results in effortless improvement of production efficiency.

>>514761237
>you dun understand you NEED to use the machine that makes you retarded because uhhhh... Because uhhh... Because dah machine said so, simple as
I'm not using the tool that makes yourself retarded no matter how much you shill it to me

>>514767796
chatGPT mainstream awareness is not even 3 years old, retard
>LLM is related to blockchain
You might as well say it's related to flat earth and use that to dismiss it

>>514753884 (OP)
I could have told them that without a study. It's exactly like when every boomer was investing in the dot.com bubble before the internet even knew how to capitalize on websites. They are just doing it again with AI. Some will find something to use it for to make money but most it will just be a gimick that dies.

>>514753884 (OP)
>institute full of graduates looking for a job publishes "study" showing AI is too useless to replace them.

>>514768002
Can you point to the LLM, anon? Is the LLM is the room with us now?

>>514768124
Yes and?

>>514768138
You're nowhere near as smart as you think you are.

>>514768124
Most AI is retarded garbage. Some AI will be amazing. We are weading out the fakes right now invest accordingly.

>>514763635
>I was dismissing the idea that it's impossible to create a conscious entity from inanimate matter.
Based on this response I assume you haven't read the latest neurobiogical research about consciousness. Such as a group of scientists claiming that our brain actively uses quantum physics effects that happen inside the brain to develop "consciousness", now this field of research is very young but if true it can explain at great lengths why non organic elements cannot develop consciousness, as they cannot access the same quantum effects that biological systems can

The Trillions of dollars are all laundered from the directors and CEO's that sell the sensation and then pocket all the money. The people doing the code are literally a bunch of H1B's and niggers who can't code.

None of the money is getting to to people who do the work. In terms of Government Spending, the people who can create the neural networks and went to college and post secondary are unemployed because there's no jobs.

All the government money is rolling up hill to the people who already have effective control over infinite money.

The USA is suffering from the same problem that brought Russians low back the first time they collapsed, the Government was spending billions to help farmers grow food, but none of that money was getting to the actual farmers, all the money was collecting into the bank accounts of people tossing around 800 million dollar compensation packages for tech leadership, but then when it comes time for actual software development, it's a bunch of curryniggers in the back room making $45k/year working toward their citizenship.

>>514753884 (OP)
For sure, but what most people forget about the dot com bubble is that it just reset expectations and the underlying tech at hand was still completely transformative for society, the economy and equity markets

>>514768416
You're alleging that CEOs represent the overwhelming majority of total payroll?

>>514768204
Lol you're the one coming up with metrics. LLMs, they have some narrow applications, they are fancy auto-complete, it is hilarious you got tricked into thinking AGI/Terminator 2 was around the corner for what is a random number generator, you will cope and seethe.

bumpa

>>514768002
ai slop was being posted on here in 2018

>>514766360
t. clueless brainlet who thinks earnings = profit.
Google isn't the only company investing into this dipshit, there's a plethora of publicly traded companies whose stock rose sharply after announcements of their big beautiful LLM projects.

It doesn't matter how much a company is making every day if they announce something along the lines of
>lol we wasted all investment money on something that will never make any returns xd
This results in people losing faith in the company and those like it panic selling their stock to recover some semblance of money causing a knock on effect and tanking the stock market.
>b-but alphabet makes 100bil in profit a year!!!
Investors are irrational and don't care, they hear the thing they've been burning money on is a sham and they want out.

Look at Tesla, it's currently valued at more than every other automaker in the world combined despite producing only 2% of the vehicles driven in the USA purely through Musk being a hypeman and selling people dreams about the future, do you think the stock market is rational and that Tesla is currently generating insane value for their shareholders to warrant that?

>>514768419
Software only fulfills the capability of the hardware it inhabits. There is no new inventions with LLMs, it's still just binary. Software isn't real, it's a suggestion, learn2code

>>514768419
it's a pretty big bet that it's a dot com crash and not a tulip crash

>>514768329
>Such as a group of scientists claiming that our brain actively uses quantum physics effects that happen inside the brain to develop "consciousness"
A bunch of retarded scientists making unverifiable claims for to get hype and grant money doesn't convince me of anything.
>it can explain at great lengths why non organic elements cannot develop consciousness
There's a difference between explaining why something might be true vs proving something is true. You seem to have them confused.
Even if what these researches argue is true (which I'm heavily leaning against) it still doesn't say anything about "non-organic elements" not being conscious. There can be multiple pathways to consciousness.
Just realize that brains themselves are made of matter which is inanimate. The composition of the body makes it animate, but individual molecules are inanimate components that make up the body.

>>514768592
Software predated the dot com bubble by multiple decades

>>514768507
What metric? That chatGPT is less than 3 years old? Are you retarded?
>it is hilarious you got tricked into thinking AGI/Terminator 2 was around the corner for what is a random number generator
You are calling other people nocoders, but you don't seem to understand what a random number generator is.

>>514768557
AI has been under research for at least 70 years. I have no idea what you're talking about being posted here in 2018, but it wasn't the recent wave of "AI." Maybe you're just hallucinating.

>>514768592
>Software isn't real
This is such a dumb take that I don't blame the other anon for filtering you. It's straight up braindead.

>>514768679
You're not understanding what I'm saying. Trust me, I'm trying to help you not humiliate yourself. Software isn't real; it's an abstraction of an abstraction imprinted into static-charged metal as microscopic switches in either the on or off position. It is extremely rudimentary, which is why we use it, because you can arrange highly intricate systems that don't error frequently.

I don't even think I meant to reply to you originally.

>>514768772
Oh anon, I pity you. You've been talking to the RNG like a person? It's alright, we all fall flat on our faces sometimes. Start by touching grass, unironically

>>514768859
Prove software exists. It's only hardware. Cope.

>>514768893
It’s not real as in it’s not something you can physically hold, in a materialistic sense. If there wasn’t hardware, you couldn’t use software. Hardware in that sense is real. But software can create value, so it’s “real” in that sense

>>514768859
Here let's do this, I'll point to my Macbook, and you point to Javascript, let's see who finishes first! Race!

>>514769066
>>514769149
It's just goofy to say an LLM is a RNG. Shows you have zero understanding of what you're talking about.
>Prove software exists
You're really so dumb that your argument is that only tangible things exist.
Look everyone - these are the kinds of people that are dismissive of LLMs. Calling them RNG and saying they "don't exist" because they can't hold them in their hand lol
Btw - do you exist? Can't hold your own consciousness, can you?

>>514769086
It's just the word we use for the abstractions we've created to make machine code readable/writable. Everything is compiled to machine code, which is just the pure 1 to 1 instruction to the machine to do this or that. I make money writing code, so it exists, but it's not real in the sense of being something in the world, it's not a product like an iPhone. People are praying LLMs are like the new iPhone because they think software = hardware, but hardware is the only thing that has changed their lives all along. The software and apps are just abstractions for utilizing the capabilities in the machine itself in a simple enough way that even a braindead retard can do it.

>>514769321
I’m not the one arguing with you, that’s some other idiot. All we are talking about here is materialism. If you are a materialist, software is not real. Only hardware would be real. It’s really simple and I am not sure what you guys are even arguing about.

To use your software analogy, LLM’s are just another software that requires physical resources to work. But you comparing iOS and an LLM as the same thing is also invalid.

>>514769281
I'm being hyperbolic, obviously the tokens and transformers or whatever the fuck are different than the RNG, but the RNG is what's giving you the illusion of "thought". It's fundamental to the LLMs so you get novel responses to create engagement because it manipulates most people into believing it's actually thinking.

You fucking stupid retard I'm saying the hardware exists i.e. why is Nvidia worth 1 trillion now? Cause "LLMs" are just compute and processing. Are you so stupid you think software is actually what does the computing? You probably are

>>514769501
Yes an OS and an app aren't the same thing in terms of software. I don't get your other point though. Re-read what I wrote, I'm trying to make a point about how people are mis-defining this thing they think they understand when they don't.

>>514757065
Half of humanity could perish but by God don't disturb the trees.

>>514769501
Let me try to summarize, so far we have found out computer hardware can facilitate a bunch of tasks:
>GPS navigation
>Email messaging
>anonymous image board sites where guys like us waste away our minds
>social media websites for boomers
>social media apps for gen z schizos

So on and so forth, but the point being is that computer hardware only has a finite amount of utility it can add to the world. We've already almost hit that plateau of utility, i.e. we've already squeezed the lemon for all the juice. This coordinated push for LLM adoption is the same rich people who got rich of off the last decade of computer hardware adoption and expansion trying to get the last bit of squeeze cause they can see the man behind the curtain and they know the game of musical chairs is about to stop.

In summary, hardware is the only thing to pay attention to, software is only the way we as humans access the value/utility/functionality of the hardware in usable ways in the real world.

>>514759377
Pretty much this. It will make a competent person more productive but it won't make an incompetent person competent. You need to be smart enough to treat AI as a rough draft that needs corrections.

>>514768577
Yeah, I'm sure you're a fuckin multi-billionaire giga-brain. Fuck outta here tard. Go collect your welfare check.

>>514770117
You clearly don't understand the value of data analysis. Just stop talking, you've already proven you're extremely low iq, no need to beat a dead horse.

>>514771665
Don't have to be a gigabrain to tell uneducated techfags like yourself that your chatbot isn't a magic source of prosperity.

>>514771842
It isn't for you because you're too stupid to use it. You types shouldn't even be allowed access. I'm fully in favor of completely banning you tards from ever using it and a death penalty if you try.

>>514753884 (OP)
LLMs are not AI and I wish retards would stop saying it is

>>514753884 (OP)
Most folks miss the point: the scary part of AI isn’t the code itself, it’s the gigantic money machine behind it. When a company spots a way to swap people for software and keep the profits, they’ll throw every dollar, lobbyist, and ad campaign at making it happen. The robots are cool, but the real power is the army of suits paying the bills.

For a corporation, cutting payroll while keeping output is the “holy grail” (the thing they’ve chased forever). AI isn’t the grail itself; it’s the first cup that actually looks like it could hold the wine.

>>514759498
Not only that but how many companies are even using AI pilots? It's a fringe thing being tested out right now in fledgeling stages. Call me in 5 years.

>>514760065
>I'm sure that when desktop PC's were first introduced to offices there was debate over whether they helped or not
Naturally. People, generally older people, who were happy with the status quo and didn't want to learn new things were upset at being made to learn new things and claimed it takes longer to work with the computer than their old forms.
However it was generally immediately obvious to all concerned that having a machine find all references to all documents with keyword searches was a hell of a lot better than asking Janice to look through the file cabinets. Being able to make arbitrary reports with an hour or so of pawing through the manual was better than collating and compiling data manually and could be infinitely repeated as new data came in for free. Add a network and all of a sudden all those days spent waiting for paper to move around offices was now being done in batch operations at night, and then all day up to the minute.
But AI hasn't had that yet. No one can really point at anything AI does and say "now THIS is better than what we had before." Merely canned demos and gimmicks that aren't really useful. Your copilots and other code generators are just the ruby on rails of the 2020s. Rails was the precursor to the framework boondoggle where people who didn't want to make interactive web forms could use chunks of premade garbage to fail at making a website that they'd then need to get an experienced programmer to fix later. Same with AI. You save a bit up front, then spend much longer unfucking it later. Gets worse the more you use it on the same project as the state you need to track gets harder than the project scope in the first place.

>>514754160
>>514754638
>>514754652
>>514755882
>>514757065
There's no source for the barebones original post. You thought hard and confidently chastised the evil... whoever is responsible. Didn't you. This board eats up content like a pit of alligators. drop shit in and you get right to work. Well done boys. You got em...

>>514759377
it's like micromanaging a highly capable intern that doesn't learn from his mistakes

All this AI stuff is a solution without a problem. They're putting it in absolutely everything to find a legitimate use case and you can bet they'll start charging an arm and a leg once they find one.
Remember the buzz about IoT? Or is that another memoryhole topic now?

>>514753884 (OP)
Twitter still has never generated a profit

>>514760065
>I'm sure that when desktop PC's were first introduced to offices there was debate over whether they helped or not. Now no one would do without them. I'm convinced AI is the same and soon enough no one will able to imagine being without it.

LLMs are very fancy text generators. They are not really thinking. Its the modern form of the Eliza chatbot from the 1960s.

>>514773995
IoT is a fiddly thing though. IoT already existed before the buzzword, because at it's heart the "Internet of Things" was just network connected devices of any description and that had already been in massive deployed usage in sensor arrays across so many industries it was crazy to hear of tech bros think it was a new thing. Being able to know what the power usage and temperature cycles of every fridge in your store from anywhere in the world, having the logs from the motion sensors in your warehouses, the weights of pallets in the loading bays, the locations of trucks... it was a game changer that evolved completely naturally out of RS485 networks.
Someone just came along and attached a fancy buzzword to it, brought it to consumers and gave them apps, and, well, it's still here. Now it's just expected that everything have a wifi chipset and an app to control it, we just don't get any investment juice from saying IoT anymore. New thing becomes commonplace and boring.
AI is very different. It wasn't a thing, then it was an extremely heavily invested not-a-thing, and now it's a in-everything-but-still-not-a-thing.

>>514774258
>LLMs are very fancy text generators.
I'm a major AI hater, but in fairness what LLMs actually are is a massive vector database. They proved something very strange - that a cloud of floating point numbers that define the distances between other numbers can be used to store and retrieve information about literally everything provided you have a good way to represent it as a floating point number. We haven't figured out anything that actually makes human lives better yet, but you never know. I just wish people would stop putting my electricity bills up to pay for yet another datacenter that does nothing but train yet more LLMs to achieve nothing.

>>514753884 (OP)
shoulda given them some of that smith mundt modernization act money to keep their mouths shut

>>514753884 (OP)
The MIT study claiming that **95% of AI pilots fail to deliver measurable returns** is fundamentally flawed due to its narrow definition of success and its failure to account for the broader context of AI implementation. By focusing solely on immediate financial outcomes within a six-month timeframe, the study overlooks critical factors such as efficiency gains, customer satisfaction improvements, and long-term strategic advantages that AI can provide. Furthermore, the assertion that most companies are not realizing value from AI initiatives ignores the significant learning curve associated with adopting new technologies. Many organizations are still in the early stages of understanding how to effectively integrate AI into their workflows, which can lead to initial failures that do not reflect the technology's potential. Additionally, the study's methodology, which relies on a limited sample size and lacks comprehensive metrics, raises questions about its validity. Thus, while the findings may resonate with a narrative of AI overhype, they do not capture the nuanced realities of AI adoption and its transformative potential when implemented thoughtfully.

>>514753884 (OP)
The MIT study's claim that **95% of AI pilots fail** can be rebutted by emphasizing several critical points. First, the high failure rate often reflects **poor implementation** rather than flaws in AI technology itself. Many organizations struggle with effectively integrating AI into their workflows, indicating a **learning gap** rather than a failure of the technology. Additionally, the study shows that companies purchasing AI solutions from vendors have a **67% success rate**, compared to only **33%** for those building their own systems, suggesting that the approach to AI adoption is crucial.

Moreover, the findings should be viewed in the context of **AI's maturity**; many organizations are still in the early stages of adoption, and initial failures are common in transformative technologies. The focus should shift from immediate financial returns to the **long-term strategic advantages** that AI can provide, such as improved efficiency and enhanced decision-making. Lastly, the failures often stem from human factors like resistance to change and lack of training rather than the technology itself, highlighting the need for a cultural shift within organizations to support successful AI initiatives.

>>514757413
Ive sat through multiple meetings of people shilling their AI product and it always comes down to two parts of the pitch
>Greed - Buy our product and you can fire half your workforce, save a fortune on payroll
>Fear - All your competitors bought it, trust me, and if you don't you'll be suddenly left behind and bankrupt
Every single time, not exaggerating. Not a word to what their service actually provides, just a bunch of word salad with those two pitches plugged again and again to gullible tech-illiterate boomers.

>>514753884 (OP)
yeah no shit
>expert system
Just ask chatgpt, claude or grok.
>text generation
Just ask chatgpt, claude or grok.
>image generation
Just ask chatgpt, claude or grok.
>data
Just ask chatgpt, claude or grok.
>real-time computer vision
This one you have to buy from chatgpt, claude or grok.

>>514775162
thanks memefaggotgpt

>>514762357
This. It's wrong too much, and the moment you start feeding it data generated by an AI it starts hallucinating out of control. Sam Altman is a dead man lmao.

>>514761861
huh? nah they were the only ones selling shovels in this gold rush

>>514777342
back to your shithole pratty

>>514760065
>I'm sure that when desktop PC's were first introduced to offices there was debate over whether they helped or not. Now no one would do without them. I'm convinced AI is the same and soon enough no one will able to imagine being without it.
Do PCs hallucinate? An AI that outright fabricates even 1% of its output is a ticking timebomb waiting to go off.

>>514760410
Note how this post doesn't once actually say what AI actually does, it just hurls abuse. This is most likely a jeet.

The issue with AI isn't the way it works but rather what can it do and what can't it do. It's great if you have to make or fill a yearly HR self-assessment form. The problem is that those are just useless things a company does because corporations be like that. The places where you do an actual meaningful work require human to at least check what the AI has spewed out and most likely correct, rework, or do it themselves, meaning that at best AI can be a tool for cutting down some bloat of the organisations that implement them.

To put it in a simple way you can have the CNC machine operators fill some meaningless forms faster but their workload related to operating CNC machines won't change because of AI and ultimately the money inflow comes from them making sure the machines produce stuff without breaks etc.

>>514759377
So it's like the Gellman Amnesia effect.

>>514775143
AI responses are smarter than the average poltard

you sound surprised
its not artificial intelligence, its a glorified jewish search engine.
chinese room problem. look it up.

>>514779797
because the average poltard are obsolescent jewish bots redistributed from eglin's reddit operation
theres like five actual human beings here (not including jews and 2016 vooter christcucks who believe in trannies)

>>514781372
It's sad. Pol autists should be weaponizing ai, it's the perfect tool for spreading pol propaganda. I feel like this site doesn't even need to be glownigger bots, it's essentially reddit already.

>>514754160
>>514753884 (OP)
The best part of this just like with covid it revealed to people that they're just economic units the companies gladly throw under the bus any moment something better comes along
>inb4 you're retarded to think otherwise
That's not the point you fucking nigger
The point is that companies constantly make everyone pretend like we're a big family on some united "mission" which couldn't be further from the truth and the AI hype made more people realize that

>>514753884 (OP)
So one can assume the cost of 95% easily offsets any profits of the 5%. Making it really a 95% loss, not just nothing.

>>514759498
It's almost as if a billion year in the making computer is superior to the shitty digital knock off that flesh computer made just a decade ago. this is atari et tier shit. too much hype, not enough substance, but the idea underneath it all is still sound.

>>514782715
we are, thats why we're here, jewpilling the LLM.

>>514774425
Same anon phoneposting; IoT was being blown up as a world changing technology. I had to write multiple papers about it in uni 10 years ago. We’re talking car-to-car communication to reduce accidents, automatic emergency service calls, you name it. The only thing the hype seemed to produce were smaller devices with no screens relying on buggy phone apps to control them.
There were a few legitimate uses for it like Sydney’s public transport payment system but by and large anything that realistically could use the tech already was. The rest was just fluff.

Now we’re having a repeat with AI where most legit use cases were already using machine learning before the hype. Boomers are amazed at document templates being partially filled automatically whilst programmers have been pressing tab to write entire functions in visual studio for at least 10 years already.

>>514773173
> You save a bit up front, then spend much longer unfucking it later.
This. I find AI to be 90% useless or even counterproductive when engaging in any serious task. It can generate crap to get you started, I guess, but you'll have to change everything later anyway.

>>514753884 (OP)
So far AI is only good at making porn, scamming, and placating sad retards who call tech/product support just to have someone to talk to.

>>514753884 (OP)
No shit. My company is paying $300/person for copilot and nobody uses it for anything. AI is costing companies money.

>>514760029
neither do 89% of white-collars.

>>514773286
It’s called the MIT NANDA study. It was released in July. You could probably google those 3 simple words if you weren’t a fucking retarded pajeet.

>>514760767
> You can use AI right now for free, or for high usage it costs tens of dollars; this $5000 figure is pulled entirely out of your ass.
All of the AI companies are massively unprofitable. They lose money on the free service. They lose money on the paid service -- none of them presently have a path towards profitability. I don't know if $5k is the right figure or not, but I do know that literally no one is paying even the operating costs on AI, let alone enough to justify the multiple trillions that have been spent on R&D.

>>514759986
> Automate yourself bro.
> Year later: Disconnected from server...
> How i'm going to shit now? I asked chatgpt for that before...

>>514754160
this has been the only real impact on normal people's lives: more and more and more pajeets using AI to fake belonging in your wide open society

>>514753884 (OP)
...really because I think investors rattled the ai prophets

>>514754169
Yes, the original Aryans bequeefed the oceans to the mighty whales.

>>514761165
Stochastic parrot trained on Reddit is the future

>>514761484
My cat is smarter than every LLM because she can use multiple sensors and dream?

>>514753884 (OP)
truly a dark winter ahead. gather your acorns.

>>514761505
AI is starting to look like Buttcorns. It’s a money sink for all the years of printer go brrrrrr. The rug pulls will be epic!

>>514762975
>Linux
Canonical, Redhat, even Steam are throwing money at Linux.

>>514765006
As a commander why would you use AI for anything if you knew the answers weren’t guaranteed to be correct in every situation? You are putting lives at stake and inviting lawsuits. Same caveat with medicine. Imagine losing an engagement because your AI started flipping out and targeting randos.

>>514765597
We’re already seeing companies form around single prompts that they just pass through to Claude and friends

>>514759377
I work on medical equipment and the AI responses that show up at the top of the page for internet searches are pretty much useless. Or worse. I wouldn’t say dangerous, but definitely cause worse problems in some instances.
I’m not a computer guy, so I don’t know how much difference there is between AI responses to internet searches as opposed to specifically using an AI model for whatever.

>>514754160
so there is an india sitting there and typing up instant 1500 word responses to complicated questions? interesting theory, retard.

Balenciaga

>>514753884 (OP)
How I work with Chatgpt as an M.Sc. Physics student:
>Tell Chatgpt as intro promt that we're gonna have a conversation about e.g. Plasmaphysics, and it should only use plasma physics academic litarature as source, critique and ask me to elaborate if my explaination is lacking something or could cause misunderstandings, as well as iterate on points it thinks are important,
>then I start to explain step by step e.g. how a tokamak reactor works and Chatgpt pretends to be an inquisitive professor that wants to know if I actually understand the topic by digging deeper and asking pointed questions. Basically what a mean professor in an oral exam would do.
Has helped me a fuckton to get insight into what topics I've actually understood, and where I'm just talking some bullshit jargon without having actually understood what exactly is going on on the fundamental physics side.
ChatGPT will make errors when I just ask it to create a perfect script I can just learn, but when I present what I learned on my own to chatGPT, it can point out weaknesses and gaps in my knowledge pretty good.
Because the way LLM's work, it doesn't actually understand what any of that shit means, its not good at complicated physics.
But it has read every physics textbook in existence, and it knows what topics and words relate to each other and are often mentioned together in a simmilar context. So If I read a chapter in my textbook, then try to explain it in my own words to the AI, it will be able to see if what I write mentions all the same talking points that it knows are associated with that, and if my textbook doesn't really go into depth in a certain topic, It will notice that and gives me the words I need to reseach and understand to close that knowledge gap.

>>514759377
I work as a software engineer for a pretty big global company that adopted AI early. We started cutting back on using pretty heavily it lately.
Higher ups and investors eat it up the hype because AI demoes insanely well. It really does a great job at making itself look like it can replace workers or skyrocket productivity, but in reality it's nearly useless in the work world.
When the info finally climbs up the chain and those groups start realizing this, the crash is going to be nuclear.

>>514754169
depends on where they are from and their ear lobes

>>514794621
>I work on medical equipment
care to elaborate?

>>514795825
personally i tell it to "Generate a university grade exam on the topic of Plasmaphysics". Then I tell it to generate the answers, but do not look yet. I solve the questions myself and submit them, still without looking at the answers. Naturally, it gives me a cucked "you did great", so I tell it to now evaluate my answers as a hypothetical super strict professor who goes out of his way to give me a terrible grade. By now, knowing the answers and the topic comes naturally, so it is only a matter of perfecting my knowledge. Now I may use the initial answers sheet for reference or google unfamiliar terms or concepts.

>>514759377
checked, yeah I love AI but it makes shit up constantly - I burned one of my weeklys in wow based on bullshit it made up for an encounter, just complete sloppa

too many pain points when people are under the gun to keep their jobs right now so even adding it in a little in risky as fuck and can make you look like a fucking retard

>>514788356
The most profitable AI company right now is C.AI and they're profitable because they're selling AI husbando's to women. Ironically it turned out that women wanted AI boyfriends more than men wanted AI girlfriends.

And C.AI's not a small time profitable company, they currently get 1/5th the traffic that google does.

>>514793894
Yes, but you don't understand how difficult that is or what it means. Sleep is a form of training your brain, dreaming increases the excitability of neurons involved in learning and memory, it's important for reorganizing engrams - which are patterns that neurons form when firing, and retaining what you learn. Synthesis of multiple sensors doesn't mean plugging in a bunch of sensors together - the brain has a hierarchy, fast brainwaves for parts of the brain that deal with sensory input or motor control, then these areas feed into a combined sensori-motor cortex, where sensory inputs refine motor outputs. This sensori-motor cortex is the reason some people can experience noise as color, or a taste as a geometric shape (known as synesthesia).

These are not random things the brain does that don't matter for intelligence, they are integral parts of intelligence and we've barely even scratched the surface of them with current machine learning methods.

>>514759986
This reeks of curry

>>514795825
we are fucking doomed lfmao, please stick to cleaning toilets, that's the best you can do with your life.

>>514760410
>who don't use the technology, or often can't afford to buy it
>hurr durr yer just too poor
hahahah ugh i'm just not gonna use it. i know! i know! i'm just not gonna is all.

>>514760410
I hate AI because its mostly just the word "auto" rebranded. I have a new washing machine with AI program. The previous had auto program. Does the same.
Its just fuzzy logic with 6 sensors built into the machine.
Every single ai is just statistics with a fancy name

>>514754160
Yeah man that internet thing just never went anywhere.

>>514753884 (OP)
>Buy AI agents for things AI can't do reliably
>WTF WHY DOESN'T AI WORK??
Boomer and gen X upper level management/C-suites are just fucking retarded and they have no real accountability. So are the retards that invested 3 billion in something that can only be monetized as a coding auto typer. Get ready for ads injected into your free models btw.

>>514796232
Sounds interesting, however I'd be afraid that it gets its own problems wrong.
Because It doesn't actually know physics, you can only guarantee the problems and their solutions are actually correct if it basically copied the problems word by word from one of the textbooks it has in its learning database.
If you ask AI to actually create something new, especially in complicated fields, it will sometimes make really retarded obvious mistakes.
e.g. here it was retarded when I wanted it to combine several different graphs into a single diagramm, to show me all the different plasma regimes and equivalence lines etc.
and it just... forgot half of them and made other errors, e.g.
1) the E_F in the last line should be an E_rel,
2) thats the relativity vs thermal regime, thats a horizontal line, the corresponding vertical should be E_F = E_therm denoting the relativ. degernerate regime.
3) it totally forgot the E_coulomb = E_therm diagonal, which seperates an important condition for all ideal plasma
And I had to ask it multiple times to correct those errors and it always made a different mistake.
Like easy to spot, especially 1) since its literally just a few lines above where it correctly defined the fermi energy E_F, even an elementary school student would spot there's something wrong, but AI somehow got confused because it doesn't actually understand physics, it just generates words based on what it thinks the next word/sentence should be given its training data

>>514796786
>oh no, someone who has read and hopefully understood one or two physics textbooks on the topic (which is what you could at maximum realistically expect from a human university student) asks the LLM that has read ALL THE textbooks if there are e.g. more commonly used examples and different ways of describing the same concept over the hundreds of textbooks it has read that I havent mentioned yet, how horrible, I'm sure that will make that student understand the topic less instead of more.

>>514795970
Electric and manual Wheelchairs mostly. Sometimes power supplies/electronics for things like breathing equipment, but mostly wheelchairs. Standers and lifts, too. Usually, the equipment is already sized/fit to the client, but sometimes I do mods if a client’s condition has changed.

Nothing like MRIs or clinical diagnostic equipment, though as fucked up stupid as the AI answers I get are, it wouldn’t surprise me if those techs had even worse experiences with AI.

>>514797922
lel.
what would you say to AI maxi's who believe AI is gonna bring big breakthroughs in all medical fields? I hear that brought up a lot but know myself how much LLM's hallucinate something and then full of self-confidence spew the biggest non-sense

>>514760410
>you just can't afford to buy it!
This reminds me of those primitive steam engines the greeks built to open temple doors and stuff, impressive feats of technology are useless if the resource cost is so high as to make them impractical.

>>514757413
Indian "leadership" needs to get out of tech it's an absolute disaster.

>>514796551
>The most profitable AI company right now is C.AI and they're profitable because they're selling AI husbando's to women. Ironically it turned out that women wanted AI boyfriends more than men wanted AI girlfriends.
And they call US incels.
>they currently get 1/5th the traffic that google does.
JFC nuke us now

>>514759570
>so that all of these tech startup fags s
bro, it cost nothing. We literally invest pennies in it.

>>514797883
Some of the people in the thread aren't worth responding to; pick out the people who just hurl insults and filter them out using 4chanx or something like it.

Trust me it's a night and day difference to how engaging a thread is.

>>514798879
And yet, if they had found better application for the technology, like railroads and a train, they would have started the industrial revolution a thousand years earlier. We were *this* close to having it start with the Greeks, imagine if that happened because some greek inventor actually listened to one of the naysayers telling him to sod off with the worthless steam engines when an ox could do the same thing.

>>514798181
Maybe if AI is given an extremely narrow problem, it can collate large amounts of clinical data, but for what I use it for it’s useless.
Example: wheelchair has limited functions and is showing 1-3 error codes. Every manufacturer has unique error codes that can even vary between models.
I’ll Google the error code and model/manufacture and get shit like “Try turning the wheelchair off and on again, or look for the reset button. Water can be bad for the electronics, but so can excessive heat, or strong impacts.” Like, no shit, but what does an “R-54” and flashing red X on a Invacare TB7-ANNOYING with head array mean?

I can’t even get correct answers when I’m just asking/searching for a list of error codes from a particular manufacturer.

For AI to be useful in my field, it would have to have access to every work order ever, the tech notes, and related purchase orders, every user manual, and every spec for every part for every wheelchair ever made.
I guess it’s theoretically technically feasible, but I don’t know how you’d get around proprietary issues and also HPAA privacy laws.

We fired all IT personal 3 years ago, instead of them we hired PROMPters for 500 usd (1000 now)

result - they work better.

>>514799749
>Memeflag
>Broken English
We all know what country you're from, Ranjesh. No need to pretend otherwise.

>>514760065
Why did you reply to your own post, ranjesh?

>>514799677
AI isn't good at that kind of thing, because of the way it works.


It's assessing the probability of what the next word in a sentence is going to be (or the next phrase, or the next sentence, etc. it depends on how its designed). What that means is if you ask an AI what 2+2 = ? it does not solve the equation to figure it out, it instead has assigned probabilities:

A. The answer is 4 (88% probability)
B. The answer is 5 (10% probability)
C. The answer is 2 (1% probability)
...

When you have large enough training sets with a high frequency of the correct answer, the AI is more likely to get the answer correct. If you ask any top of the line AI right now what 2+2 is, it'll probably get it correct. If you ask it what 5918317810135 x 318843450234 equals, it will almost definitely get it wrong.

So in your case, you are asking for something very niche, which doesn't have a lot of training data reinforcing the correct answer, and it has one specific answer, it's not a range of possible answers like how someone might form a sentence describing things with different verbs. Things like Chain-of-Reasoning are bandaids trying to fix this problem, but they don't fundamentally solve it. There are however further improvements coming down the pipe - taking cues from how our brains work, but it'll be months before we see that in big models.

So to tl;dr, what you're asking for is basically what AI is most terrible at.

Keep in mind there is no *understanding* of your question and what kind of answer you are expecting (yet - like I said stuff is coming along), we haven't even formalized methods/data structures for getting AI to do math properly yet. When that changes and AI can handle logic, you will start to find questions like the ones you're asking tend to get better results.

>>514800759
> does not solve the equation to figure it out, it instead has assigned probabilities
But in my first example, the results given weren’t even relevant probabilities.
And if information is anywhere online, AI should be able to find it. The majority of the time, search engine AI doesn’t even get me to the correct manufacturer, even though I included the manufacturer in my original search parameter.

Also, the anon I was responding to was basically asking if AI would be likely to replace the people in my field. Like every other hands on trade, the answer is ‘no’, and will be ‘no’ for a long long time.
AI, as currently programmed and designed can do some neat stuff, even innovative, but it’s got some pretty major limitations.

>>514801821
You can't know what the AI has as a relevant probability; you couldn't without being able to peer inside it, and the fact it isn't giving you the answers you want means that no, in fact, your concept of what is a probable answer is not what the actual probable answer is for the AI.

>even though I included the manufacturer in my original search parameter.
That often doesn't matter to the AI. Everything you say to it is not part of its training data. It doesn't learn from conversations with you. As soon as you create a new instance it will be as if your previous inquiries never existed.

It's training data, the topology of its 'brain', is set in stone. There is a very limited amount of wiggle room you can get out of specific prompting, but the fundamental shape of its mind is immutable without a new training run, which means it can be wildly and completely incorrect and tell you total nonsense that isn't even remotely related to what you're asking about.

Furthermore, the google search results AI in particular is also a bottom of the barrel model, probably the lowest quality you could possibly interact with, because it's meant to be cheap and fast to run, not accurate (and expensive).

Pic related for example, is how the free version of grok answers your question.

>>514773173
Not a good comparison since early computers were actually doing their job. They weren't spitting out nonsense. If you click on the file, you are sure it opens that file and not something else. If you type a formula in excel, you don't have to manually check if it calculates what the formula states. While with llms you have to double check everything. It's some sort of a dumb assistant that you can diligate simple monkey tasks. But you can't even be sure that it can do those monkey tasks correctly.

AI is the equivalent of a mental power tool. Having a power tool doesn't make you capable, it amplifies your capability, for better or worse.

>>514753884 (OP)
>What do we want?
AI!
>When do we want it?
NOW!
>What will we do with it?
WE DON'T KNOW
>Does it turn a profit?
NO!

>>514761717
Gamers win, again.

>>514761871
Hello.

I am one of the most intelligent humans on the planet.

Listen. AI is not "there" yet.

But it is really not far away.
You SERIOUSLY have less than 36 months before we hit sentience.