This. The idea of "AI powered search engines" is moronic. Nobody wants to be told "oh no, I can't do that, try asking something else" when they are trying to look up the words 'rape', 'Black person', or 'hitler'.
Probably not until another year or two depending on how well the companies dabbling in AI are able to convince investors that they're able to do something useful with it.
When the AI bubble bursts it will kill AI, just like how when the dotcom bubble burst it killed the Internet.
well frick, if Krugman says it won't happen in the next decade then I'm really worried it's going to happen soon. that guy is one of the dumbest homosexuals on the planet
Some of Paul Krugman's work is brilliant, as is Stiglitz' work... and weird vegana wall man Nouriel Roubini. Not sure if it exists but Krugman had a blog in the past, one of the posts was about interest rates, speculation, and capital washout during high interest rate periods (essentially a push-pull mechanism for evolution. It was brilliant in its simplicity and clarity. Had another post about how Earth was technically a closed system (hence all systems are inherently closed - although I would argue the boundaries are relative to causality between those systemic interactions, not the whole system having quasi-material ultimate limits). Like all economists he's also dealing with extreme uncertainty and lack of historical precedent.
It does make me wonder though. How can a development that facilitates better (faster, more accurate, greater volumes, greater capacity for disambiguation) communication be so easily written off as a fad? History has shown inventions that improve communication are almost always successful and transformative (postal systems, signalling systems, telegrams, landlines, mobile phones, couriers etc). Then again, easy for me after the fact to say this and feel smart about it.
2 months ago
Anonymous
Yeah I'm sure there were a lot of detractors wrt literacy as well. It's fundamentally moronic to be pessimistic about any development in communication, and Krugman is a collosal dumbass who thinks he knows everything because he has deep knowledge in a narrow field.
2 months ago
Anonymous
>communication [...] transformative
granted, but I would further argue that LLMs do not, writ large, improve communication at all. they inject large amounts of error and present it as fact. this is not an "improvement", you would not say a phone line which replaces every 10th word with noise is an improvement to the status quo.
2 months ago
Anonymous
The key issue I've seen is validation and blind belief confirmed by perceived successes. The 'Rehoboam' model developed by the Israeli defence (amusing) force unit is a great case (and has its own additional layer of confirmation bias): >Selects 'targets' based on ML >Hit 'target' without knowing if it actually is a target >Kill 5 people >"Yes but we hit the target!"
What is this target?
How can you be sure this 'target' has value?
Without validation this is like throwing a grenade into a doorway, waiting for explosion and going in after to find five dead bodies: "yes, these are all targets and we have killed them".
A superior model reduces inputs but maximises outputs and can be clearly validated. To refer back to Rehoboam: if it recommend targets in the form of "command or logistics hubs" and raids on these confirmed high accuracy, you would consider it having repeated successes, validated by observers. Right now the second confirmation bias is this: killing lots of people is good because frick those animals (and it gives lots of 'targets' so it looks like it works - otherwise why would it give so many targets?). I suppose in that regard it is wildly successful. This logic is portable to other situations where desired outcome is a conscious bias confirmation, or where validation does not occur and success is assumed.
2 months ago
Anonymous
>desired outcome is a conscious bias confirmation
very interesting thought, thank you.
in the corporate sphere, GitHub (Microsoft) is providing plausibly deniable license erasure at scale.
perhaps this is why organizations, typically lawful evil, are excited for this "technology": a new grey area to exploit.
If it remained as is, he would have been right. >have to dial a number to connect >have to have an expensive computer >slow as frick >no widespread adoption, no one uses it, you can't really e-mail or talk to anybody who isn't a geek >major companies have websites, but you can't do jack shit with them >most websites are just "lol here's some contact info and a page about our history"
Fortunately, we got broadband and always-on connectivity. Computers started becoming cheaper. WiFi and mobile connectivity became a thing. Websites started allowing you to actually do stuff. Adoption was increasing.
2 months ago
Anonymous
Nah, he was moronic. Broadband rollouts were on their way by the time he made his comment. Hes a talking head. Thats it.
>2 MOAR WEEKS, trust ze plan!!!
this guy is the epitome of overstepping your boundaries of expertise
2 months ago
Anonymous
why are you expecting people to not grift if they are in the position to? "almost all humans are shit no matter their status" is one of the best bets you can always make.
2 months ago
Anonymous
this only happens when money or reputation is on the line. otherwise people are usually pretty chill
2 months ago
Anonymous
>otherwise people are usually pretty chill
if (almost) anyone can get away with it they'll do something shitty if that gets them anything. most people behave in society because of the repercussions of being shitty.
2 months ago
Anonymous
>being a dick has repercussions and others may respond negatively
is that a bad thing?
2 months ago
Anonymous
I love shitposting and moron posting (partially because I'm a dumbass) and pissing people off
Although my parents didnt love me
2 months ago
Anonymous
I don't see a problem with it, but people won't be happy, unless they're irony poisoned or know you're messing with them
Fundamentally true. Internet didn’t have any significant impact on the economy. All it ever did was following. >Allow workers to work remotely, which dipshit managers didn’t like after 2 years >Everyone sending emails all the time, which could have been phone calls or normal paper mails >All revenue from YouTube and social media would just go to reality TV shows >Shutins would still happen because of better video games and losers getting into anime production catering more to losers
Nothing would change, you would just have to visit the library instead of typing Wikipedia into your browser, but that’s it. Fax machine on the other hand allowed for more paper wasting and discarding old documents, something also possible with computers but not possible before, making sending documents cheap
For a brief period internet had potential, you had the same power as a researched had because you could request any article or book and some bot or someone would provide it in now time, for free but nowadays sci-hub returns an error on every single new article that you demand and even older books just disappear from the known places.
>When the AI bubble bursts it will kill AI, just like how when the dotcom bubble burst it killed the Internet
Maximum IQ post.
>AI bubble bursts >it will kill AI
--- >dotcom bubble burst >killed the Internet
Hmm. I smell a dishonest argument. Here's you (You), though
Low IQ post.
My 'smart' washing machine is AI powered now, there used to be a setting called "most used cycle" and now it's called "AI recommended cycle" which recommends the same thing: Normal wash. The power of AI.
I like how it "learns" washing habits. So it uses the exact cycle you were going to use anyway, which was one button press away on the touchscreen...
And a mobile app, a sign in, and more than one screen tap - literally taking more time. Oh, also please make sure you agree to the privacy policy, grant our app permissions to everything, and agree to receive "customer focused communications" (i.e., ads).
Not yet.
When the bubble is actually popping we're looking at catastrophy. Not just people growing disinterested and moving on. You gotta remember soilicon valley has to be "the future" and have some "new thing" they are working with.
AI is the last one. There were a dozen or so others 12 years ago and they've all either found their niche uses or they just fricking died. The tech bugs know they're fricked when the AI jig is up so it's gonna go out with a wimper. Maybe like 4 years from now
LLMs not becoming epic (generally) intelligent will be for American pessimism what the moonlanding was for American optimism.
>The tech bugs know they're fricked when the AI jig is up so it's gonna go out with a wimper. Maybe like 4 years from now
It isn't that simple. This tech will be "transformative" in a few senses, but it won't result in social transformation of society wholesale. It will find efficiencies, it will be used in network management, in drug discovery, in writing (RIP journal reviewers, your free labour is appreciated), and day to day activities like word processing etc. Still, it's comically overblown.
Come to think of it I haven't actually seen an application of "AI" in my life that has resulted in any tangible benefit. Right now were in the hype phase to get that VC in the mix and bamboozle everyone to snare capital in a high interest rate environment. Probably just AI barons trying to expand and consolidate, as per standard tech modus operandi.
>This tech will be "transformative" in a few senses, but it won't result in social transformation
Oh, I am sure LLMs will have a handful of established uses and the anti-social manbabies who love local models will continue to experiment, but once this shit proves it isn't the next step. That essentially signals that the "smartest people in the world" don't have one. That's gonna send shockwaves. The real debate is on how big.
>The real debate is on how big
The real question to pursue is "what is consciousness?". That isn't aligned with the taken-for-granted nature of consciousness that mostly receives a 'shared yet unarticulated understanding' treatment. Building an 'AI' right now seems like building a structure out of Duplo and never thinking those little plastic bricks may not be suitable.
I kindly request challenges to this assertion.
2 months ago
Anonymous
I don't have a challenge. I don't think putting a bunch of porn and github code into a datacenter will ever turn it into god. That's just fake and gay.
>I like how it "learns" washing habits. So it uses the exact cycle you were going to use anyway, which was one button press away on the touchscreen...
that's how all ai works, hth. it's just a gussied-up markov chain.
>When the AI bubble bursts it will kill AI, just like how when the dotcom bubble burst it killed the Internet.
it already happened before, moron
https://en.m.wikipedia.org/wiki/AI_winter
Brother AI has had its highs and lows several times already. Some huge advancement is made, people make fantastical predictions, developments happen but fail to meet fantasies, money and research stagnates, then it all starts again. Its been heavily linked to compute availability at each stage.
My 'smart' washing machine is AI powered now, there used to be a setting called "most used cycle" and now it's called "AI recommended cycle" which recommends the same thing: Normal wash. The power of AI.
Shiny Things Inc. is always one step ahead. Now dishes can be one step ahead too. Fuzzy Logic aquainterpretation allows smart cleaning by delivering detergents based on suds subset values between "completely dirty" and "completely clean". Saves time. No more frustration. No more pain
I think my washing machine has AI too. When I first got it I set the dial to "Quick Wash", and every time since then that I've gone to use it it'll do a Quick Wash cycle unless I manually move the dial. Fricking weird man.
This. Why even bother with this castrated shit that's keeps asking for your phone number and money. It's so easy to download and run your own locally, even a monkey could do it.
Probably not until another year or two depending on how well the companies dabbling in AI are able to convince investors that they're able to do something useful with it.
>HN comment thread
But yeah I unironically agree, I hope that in one or two years the companies stop aggressively gimping themselves out of a moral panic. Well either that or a decent competitor shows up (or local models become just as good).
Don’t care. Just today it figured out some frickery with forcing [Redacted] library to pretend to be thread safe. It would have taken me all day. The time from that is already worth more like $200.
Constantly lobotomizing AI is starting to become a bit of a problem, because there's practically an infinite amount of things to censor due to ever growing sensibilities.
This simply won't work as an approach.
They either allow it to say that Black folk commit most of the crime and trannies are mentally ill, or it'll break down in the long run.
Both my local miqu 70B and 9 year old son understand that YWNBAW, no fine tuning necessary.
In contrast, Google has a whole team of pajeets desperately monitoring and tweaking Gemini to prevent it from hurting troony fee fees, kek.
Not yet.
When the bubble is actually popping we're looking at catastrophy. Not just people growing disinterested and moving on. You gotta remember soilicon valley has to be "the future" and have some "new thing" they are working with.
AI is the last one. There were a dozen or so others 12 years ago and they've all either found their niche uses or they just fricking died. The tech bugs know they're fricked when the AI jig is up so it's gonna go out with a wimper. Maybe like 4 years from now
LLMs not becoming epic (generally) intelligent will be for American pessimism what the moonlanding was for American optimism.
the next hype wave is everyone just accepting bitcoin as their lord and savior from the fiat scam dystopia, and companies just copy what michael saylor has been doing. yes it's that time again where the haters seethe about the alpha omega "bubble" that they have been filtered by
Bitcoin's chance was with the Web 3.0 rush that started the year before the pandemic. The best outcome for crypto is the opposite of it replacing fiat currency, remaining anonymous.
Crypto failed in replacing fiat. Instead the same demons that made fiat a scam went into crypto.
Bitcoin's chance was with the Web 3.0 rush that started the year before the pandemic. The best outcome for crypto is the opposite of it replacing fiat currency, remaining anonymous.
BTC was never anonymous. The entire point is that every single coin and transaction is public forever.
The only reason it would not be anonymous is if you tied your address to your personal identity. You would not have to do that if you were not cashing out, which is why the post said replacing fiat. Anyway, I use monero.
Infinitely printed sewer israelite rat money has US economy behind it. Crypto has infinitely printed sewer israelite rat money behind it. If USD fails, most crypto fails, if USD fails, the whole west fails and you won’t even be able to buy anything with that BTC. BTC is rising and falling not in response to printing money, but in response to how many people want to buy it usually with USD
Crypto failed in replacing fiat. Instead the same demons that made fiat a scam went into crypto.
[...]
BTC was never anonymous. The entire point is that every single coin and transaction is public forever.
Institutions that demonized BTC are already in crypto. The supposed scumbags that crypto was meant to defeat. That said, there's still opportunities left in it for the small guys.
Bitcoin will always fail because bitcucks don’t understand how a currency works.
A currency is consumption, that’s it. >Can I exchange my sea shells for a bite to eat?
If so, then you have a currency. >Can I exchange my Bitcoin for a burger?
If not, then you just have a meme that has no power.
Bitcucks fell for the speculator trap and because their ecosystem has no consumption ywnba real currency.
1. I already knew about this
2. It’s literally over a decade ago, so much has changed. >timepepe.jpg
3. This proves my point even more. The thing that kept Bitcoin alive in those early years was its fledgling consumption, the novelty that you could go and order some food with it. However this changed in time with it becoming more and more of a speculative bubble, killing off any kind of meaningful consumption activity. Now it’s basically powerless and not a threat at all to fiat.
there is still useful web services that offer payments in crypto.
2 months ago
Anonymous
>there is still useful web services that offer payments in crypto.
not wikimedia, though, due to the high environmental costs of proof of work blockchains >but ethereum switched to proof of stake, so the environmental excuse doesn't make sense
shut up!
Some of Paul Krugman's work is brilliant, as is Stiglitz' work... and weird vegana wall man Nouriel Roubini. Not sure if it exists but Krugman had a blog in the past, one of the posts was about interest rates, speculation, and capital washout during high interest rate periods (essentially a push-pull mechanism for evolution. It was brilliant in its simplicity and clarity. Had another post about how Earth was technically a closed system (hence all systems are inherently closed - although I would argue the boundaries are relative to causality between those systemic interactions, not the whole system having quasi-material ultimate limits). Like all economists he's also dealing with extreme uncertainty and lack of historical precedent.
It does make me wonder though. How can a development that facilitates better (faster, more accurate, greater volumes, greater capacity for disambiguation) communication be so easily written off as a fad? History has shown inventions that improve communication are almost always successful and transformative (postal systems, signalling systems, telegrams, landlines, mobile phones, couriers etc). Then again, easy for me after the fact to say this and feel smart about it.
Krugman wrote about stability of Iceland one year before the 2008 crisis fricked it all up and tried to hide it all lmao. he's an absolute hack.
>Krugman wrote about stability of Iceland one year before the 2008 crisis fricked it all up and tried to hide it all lmao
Ohh, I forgot about that. I wonder how Stiglitz is going with that "Europe is the future of economies" line? Oh yeah, still tenured. Still quoted. Still gets the Robert Reich treatment no doubt (in terms of political inputs). There was a Wizard of Id comic that would have suited this.
Making arttrannies seethe as normies give the AI shit more likes than their scribbles
The trannies feed on the attention so it's the death of arttroony
i've been using it to make work announcements for changes i plan to implement and convenience scripts. it does a pretty good job with some minor changes.
The current lucrative funding situation and cloud AI being the blackest of boxes seem to invite deception to keep the scam going as long as it is technically possible without getting caught
It won’t be a catastrophic bubble pop.
It will be like what’s happening to the EV space as we speak.
Where only one company really does the new tech competently (Tesla in the EV case) and every other company, big and small, flounders along until they quietly shutdown their EV/AI branches for being unprofitable.
EVs have brick walled so hard and Tesla is abaolutely not making the best ones. If not for gubberment push the EV meme would already he dead. It's on life support as is. Just last month my county discontinued the electric school buses which they only used a few days a week to begin with.
>It's on life support as is.
I wouldn't say that record high sales is a sign of being on life support.
"In 2023, the EV share of the total U.S. vehicle market was 7.6%, according to Kelley Blue Book estimates. That is up from 5.9% in 2022."
https://www.coxautoinc.com/market-insights/q4-2023-ev-sales/ >Just last month my county discontinued the electric school buses
Is your county perchance run by Republicans, who are ideologically opposed to anything that doesn't support big oil companies? Or maybe there's some other specific local problem with servicing electric buses in your area. I think you need more data points if you want to show that EVs are being abandoned as a technology.
Not him but it is true that electric vehicles are currently unable to be affordable unless the government writes a check for the majority of the cost of them. In countries where the gov has stopped doing that, sales of the evs plummeted. It is a sort of fake industry right now. Are you ideologically opposed to the government bolstering support for companies in general, or just specific industries? It is because of gov support that Elon Musk is the richest person on the planet; literally not possible without people voting for the gov to write him check after check
>In countries where the gov has stopped doing that, sales of the evs plummeted.
Have those countries also abandoned their targets of banning ICE cars?
Of course if a government announces it will remove an incentive, then people rush to buy an EV car before the incentive ends, which leads to a relative "plummeting" in the subsequent year.
We need multiple years of data, across multiple countries, to determine whether EVs are actually economically unviable without subsidies.
I'm happy to be presented with that data, but I haven't seen it yet.
2 months ago
Anonymous
The only data you need is "are people buying them?". The answer is, without gov subsidies, no. They are not affordable as an option for the majority of people yet. Some can afford them, but only the upper crust of society. They may be affordable for everybody else some day. Contrast it to LED light bulbs. Nobody buys incandescents anymore. They didn't need to be banned to get people to stop buying them. LEDs are so obviously a superior choice for lighting than the far more wasteful incandescents. They last 1000x longer and cost 1000x less to run while being 10x brighter. Do you need other data that demonstrates that LED light bulbs are economically viable compared to incandescents?
Like it or not I can buy an ICE for $16k MSRP, while the cheapest electric vehicles are in the low $30k. I have had a kia rio I bought over 5 years ago for $7k and the only repair I have ever done to it is a new battery for $100 and I change the oil twice a year for $20. I know for a fact that any repair I do to an electric vehicle I (1) will not be able to do myself and so (2) will cost far more than $100. I doubt an electric also would go for 6 years without needing anything done to it. It might but the internet says on average electric vehicle owners spend $2.3k every 2 years on maintenance
2 months ago
Anonymous
>The only data you need is "are people buying them?". The answer is, without gov subsidies, no.
Are there countries which don't have subsidies, but do have people who can afford comparably specced ICE cars?
Maybe I'm being simplistic here, but I could imagine that every developed nation has some sort of green subsidy in place, and every poor nation doesn't but the citizens drive cheap low quality cars.
Not to mention the fact that EVs only make sense once there is a certain density of charging stations, which requires a one off government investment, which a government is unlikely to commit to without also introducing a subsidy for the cars themselves.
I just think there isn't enough data to do a proper comparison here. >I can buy an ICE for $16k MSRP, while the cheapest electric vehicles are in the low $30k.
This is a fair point, but it mostly shows that EV manufacturers are targeting the customers with the biggest pockets first, which is completely normal for a new technology.
Bear in mind "Data from Kelley Blue Book puts the average cost of a new car at $48,008 as of March 2023." so most new car buyers can afford $30k, which is the unsubsidized price.
https://www.moneygeek.com/insurance/auto/average-price-of-a-new-car/ >I have had a kia rio I bought over 5 years ago for $7k
I'm happy for you, but why don't we look at the average across all car owners?
The "Cost of ownership in cents per mile" for electric cars is 60.32 which is cheaper than all other types of cars except the 54.56 for small sedans.
https://www.bankrate.com/loans/auto-loans/average-car-maintenance-costs/
>It is because of gov support that Elon Musk is the richest person on the planet
As for this, the government buy commercial launch services from him at a competitive market rate.
You could vote for politicians who want to shut down NASA as a waste of money, but I think the agency does enjoy public support.
We should be glad that SpaceX offers cheaper launches for the tax payer than the other big aerospace companies.
Elon also offers Starlink to the public and to government, but the government isn't subsidizing civilian users, it is paying the market rate for the service that Elon provides them.
2 months ago
Anonymous
The topic was Tesla.
2 months ago
Anonymous
I was responding to someone who pointed out that Elon is the richest man on the planet.
His riches don't just come from Tesla.
If you have figures for what percentage of his net worth came from government subsidies for Tesla cars, I'd be interested to know that.
>In 2022, the electric bus fleet in Norway was around 287 vehicles, down from 546 in 2021.
lmao
What caused Norway to reduce their bus fleet?
Did they replace them with ICE buses?
2 months ago
Anonymous
The topic was that he's subsidized. His wealth is largely tied to TESLA stock.
>Is your county perchance run by Republicans, who are ideologically opposed to anything that doesn't support big oil companies?
I live in one of the bluest hells there is.
The problems were mostly the cold which MemeVs are useless in and the fact that their mile range is just barely good enough to get a load of kids to a school in ideal conditions. Diesel buses are just better at the job.
And even with all the government initiatives, EVs (teslas worst of all) have very underwhelming resale value. And again, they haven't adequately improved at all since the days of the Leaf and first model S. It's beyond fair to say they aren't disrupting ICEs unless government brute forces it which so far is shaping up to be a disaster.
>The problems were mostly the cold which MemeVs are useless in
You think that EVs are useless in cold weather, despite Teslas being the most popular car in Norway?
Could the problem perhaps be that the company which made these school buses just replaced the ICE engine with an EV engine and didn't think about testing it in the cold?
For what it's worth, I could imagine your deep blue state not demanding proper tests of the design before rushing into buying the buses for ideological reasons.
That's not a problem with the technology, though, just a problem with corporate greed and government incompetence, a tale as old time.
2 months ago
Anonymous
>You think that EVs are useless in cold weather, despite Teslas being the most popular car in Norway?
Come on homie that's just braindead.
2 months ago
Anonymous
>In 2022, the electric bus fleet in Norway was around 287 vehicles, down from 546 in 2021.
lmao
>conveniently ignores smartphones
also, the internet was indeed a niche project when it was born. being niche at a certain time doesn't mean it won't grow, you fricking moron. and I'm not even defending this AI bullshit >BOT - Technology
Smartphones didn't come out 12 years ago either.
And the internet came out in a very different time and climate than all the soilicon valley futurist promises from around 2009.
Good, maybe now we can stop pretending like text prediction machines are intelligent and start getting actual use out of them as language based user interfaces.
Money fiends are enamored by it, sure. They don't understand statistical modals bleed a ridiculous amount of context, and specialized AI is expensive as frick -- idiots think they can just throw a web-dev at the problem and they can solve it in a few weeks, but there isn't an AI bubble.
It's an emerging field of study, and those who control money don't understand the type of knowledge, aka money, that is required to make such an investment provide returns. It's just another thing that'll create an greater economic imbalance.
https://www.perplexity.ai/search/List-of-all-TKvK7sjaSWGO6xBRNGtBdQ
Worked for me
List of US Presidents Whose Last Name Starts with a "T"
Zachary Taylor - March 4, 1849 - July 9, 1850
2
4
Harry S. Truman - April 12, 1945 - January 20, 1953
4
William Howard Taft - March 4, 1909 - March 4, 1913
4
These are the US Presidents whose last names start with a "T."
Ignorant moron, any of this "AI" shit is no more conscious or capable of thought than a calculator or microwave. You braindead cargo-cultist
2 months ago
Anonymous
I've literally had it get simple arithmetic wrong. I gave it a list of start/end times for my contract work, in 15 minute increments, and it generated an incorrect sum and invoice.
2 months ago
Anonymous
It's a language model, not a calculator. If it ever gets arithmetic right it's purely by coincidence. What it's good for is natural language generation.
2 months ago
Anonymous
I mean yeah that's my point. It's NOT EVEN as good as a calculator. It's vaporware.
2 months ago
Anonymous
That's like saying a screwdriver is vaporware cause you can't use it to carve a turkey.
2 months ago
Anonymous
AI was sold to us as a civilization changing, job destroying blackpill and it can't even do basic math. I'm enjoying the seethe as reality hits.
2 months ago
Anonymous
Sold to you by whom? Sci-fi writers? Of COURSE LLMs can't do math, just like traditional algorithms can't do natural language.
2 months ago
Anonymous
Literally all of the tech sector and their pets, the media. It's been doom-and-gloom, trying to drive down tech salaries, and it can't even fricking sum 7 integers.
2 months ago
Anonymous
>Sold to you by whom?
Every episode of the Lex Fridman podcast. Lol.
power of google helps apparently. I wonder how well mistral instruct does on the task. AI will probably help with removing call centers lately however I find myself receiving much more spam calls so I think those will pick up.
is there an example of a single company who has an AI implementation within their network environment that is permitted to make actual technical changes which have risk/downtime/impact? i imagine the first time an AI somehow makes a change and it causes huge downtime some committee will have a spergout and declare that only humans are allowed to make changes above x risk level.
>Dood this nobody got 15 updoots!!!
I figure we've got two more years of fad shit. AI in general isn't going anywhere no matter how much you redditors that daydreamed about learning to draw one day and living off commissions seethe and cope
I don't think AI was every super profitable, but it is profitable for Nvidia, guaranteed.
the problem is that people think AI is free, but in reality it costs electricity, and hardware could die, and inaccuracies can cause more damage than paying a human a decent income.
It's possible AI could get better than humans, it will just be very expensive with current hardware (AI really needs cyborg like assistance, we used to think humans will get enhanced by technology, but in reality machines need help with memory and shit, and silicon can't compete with human brains in cost).
>it's possible that [two letter buzzword] can become better than humans
A three dimensional object cannot cast a three dimensional shadow. It's not learning anything, it's just combining things.
It all comes down to Moore's law being dead. If hardware progressed at the rate of the 90s/2000s, then we'd be able to run models like GPT4 (FP16 ones, not the cope 4/8bit quantized garbage) on a $600 GPU in just a few years. This just isn't realistic anymore.
>It all comes down to Moore's law being dead.
https://arstechnica.com/information-technology/2016/02/moores-law-really-is-dead-this-time/
https://epochai.org/blog/trends-in-gpu-price-performance
There was an initial period were GPUs were under-explored and over-priced.
Once people noticed the potential, this "overhang" was quickly caught up with, which lead to a short burst of rapid progress.
Since that initial period, the progress has been more steady, but still exponential.
All I know is that I'm on 7-8 year upgrade cycle. My 4070 is barely over twice as fast as my 1080 and with 50% more VRAM after 7 fricking years. I had a 980 with 4GB of VRAM in 2015 and a 660Ti with 2GB of VRAM in 2013. Things have definitely stalled.
2 months ago
Anonymous
They did us dirty with the 4070. I bought it knowing well enough the 4070 Super will be the same price but better, well I'm sleeping in my as I made it anyway.
Closed source paywalled subscription AI shouldn't even be legal. We all contributed to its training, collectively as humans, and now 'open'AI wants to sell our thoughts back to us. They can still make money renting their hardware but they should not be allowed to say we can't run it locally when they don't own the training data.
It's even worse. The OpenAI company's software was created open source by volunteers. Then, once they scammed a free product to sell the board closed it and stopped being a non-profit.
Mixtral Large: >John Tyler: He was the 10th president of the United States, serving from 1841 to 1845 after the death of President William Henry Harrison. >Zachary Taylor: Taylor was the 12th president, serving from 1849 until his death in 1850. >Harry S. Truman: Truman was the 33rd president, serving from 1945 to 1953. He succeeded Franklin D. Roosevelt upon his death. >Donald Trump: Trump was the 45th president, serving from 2017 to 2021.
Plus it can Francais.
If only you knew how bad things really are...
We are heading at full speed towards another dot-com bubble level crash.
https://old.reddit.com/r/ExperiencedDevs/comments/1b4svdl/how_do_i_explain_we_made_a_horrible_investment_in/
What's up with the anti-AI shills on this thread?
All the OP image says is that user loves and uses AI a lot but the service itself is unstable, probably due to the high demand.
It happens a lot, their servers cannot provide for the immense amount of users.
Yet here we see anons making completely unrelated claiming about how it's a bubble and that AI is totally a fad.
>AI is trained on material mostly scrapped from big social media. >Big social media becomes 99% b***hing about AI and lol bots posting AI-generated "content". >Between technological inbreeding and learning from literal morons, AI becomes moronic and unusable.
To be honest, this was something that one should be expecting. This shit needs massive amount of human input to work properly but the moment it starts working properly the available human input is watered down and poisoned.
Reality is that in the long run AI is unsustainable. You want to sell it to as many people as possible but selling it to as many people as possible is what kills it.
>just a slightly better interactive search engine.
Aren't all search engines interactive?
You have to tell Google what you're searching for, or it doesn't give you any results.
What you mean is, LLMs can generate results which don't already exist.
That's like the difference between a bank and a money printing machine.
>That's like the difference between a bank and a money printing machine.
You've just described the same thing. Bani is just a more abstract way of saying a money printing machine.
You're not wrong, but I was trying to drawing a distinction between a service which returns to a user the valuable items it was previously given, and a service which generates new value from raw materials that are much cheaper.
Perhaps I should have said "a safe" instead of "a bank".
They should've stopped feeding the AI new data months ago because since chatgpt went from non-profit to profit people have been intentionally posting false information to spite them and reduce the accuracy of their AI. Plus anything you ask an AI is stored so be careful what information you ask it for
the internet used to be full of people telling the truth, but now it can't be trusted because so many people are spending their free time trying to sabotage openai, sure
>the internet used to be full of people telling the truth
There are people on this board right now posting who weren't alive when this was actually the case. Some of them can even post legally it was so long ago
Well no usecase where it's truly useful.
It's not trustworthy to define words.
It cannot summarize properly, just gives the illusion of that.
I can just give it a mess of a text from the internet that I would dismiss due to its style and after it corrects it and adds punctuation and chunks it by clause I discover that the person had a very valid point it was just delivered improperly.
I want to publically thank google for showing off how dangerously stupid Ai can be. Techbros sweeped a bunch of money again and no value was created, rinse and repeat. The internet became unrecognizably hollow in the process and no fricks were given in the real world.
>ask GPT how to make a bot for Diablo 2 >refuses because it’s illegal >ask GPT how to make a script to auto swipe on everyone in Bumble >refuses because it’s illegal
It's not OpenAI's fault that US law allows companies to sue people for accessing apps and playing games in ways that the copyright holders don't want.
By offering an AI that assists users in carrying out such "crimes", OpenAI would be legally liable as an accomplice.
You can blame Hollywood for laws like that.
Ironically, though, it is OpenAI's fault that they infringed a load of companies' copyrights in order to make their AI in the first place, but I don't suppose you'll be cheering on the New York Times in their court case about that.
The funny thing is that ChatGPT Plus would need to charge $100/month just to break even. lol
https://www.businessinsider.com/how-much-chatgpt-costs-openai-to-run-estimate-report-2023-4
BIG!!!! IF TRUE!!!!!!!
power users are just running their own at this point. Normies can mostly get by without it
This. The idea of "AI powered search engines" is moronic. Nobody wants to be told "oh no, I can't do that, try asking something else" when they are trying to look up the words 'rape', 'Black person', or 'hitler'.
so host one online that's perfectly fine with raping Black person hitler
When the AI bubble bursts it will kill AI, just like how when the dotcom bubble burst it killed the Internet.
>AI bubble bursts
>it will kill AI
---
>dotcom bubble burst
>killed the Internet
Hmm. I smell a dishonest argument. Here's you (You), though
It's called a "comparison."
AI will have no effect on the economy.
Of course not. Except for people currently investing in Google, Microsoft, Meta, X, Tesla, and Amazon
well frick, if Krugman says it won't happen in the next decade then I'm really worried it's going to happen soon. that guy is one of the dumbest homosexuals on the planet
Some of Paul Krugman's work is brilliant, as is Stiglitz' work... and weird vegana wall man Nouriel Roubini. Not sure if it exists but Krugman had a blog in the past, one of the posts was about interest rates, speculation, and capital washout during high interest rate periods (essentially a push-pull mechanism for evolution. It was brilliant in its simplicity and clarity. Had another post about how Earth was technically a closed system (hence all systems are inherently closed - although I would argue the boundaries are relative to causality between those systemic interactions, not the whole system having quasi-material ultimate limits). Like all economists he's also dealing with extreme uncertainty and lack of historical precedent.
It does make me wonder though. How can a development that facilitates better (faster, more accurate, greater volumes, greater capacity for disambiguation) communication be so easily written off as a fad? History has shown inventions that improve communication are almost always successful and transformative (postal systems, signalling systems, telegrams, landlines, mobile phones, couriers etc). Then again, easy for me after the fact to say this and feel smart about it.
Yeah I'm sure there were a lot of detractors wrt literacy as well. It's fundamentally moronic to be pessimistic about any development in communication, and Krugman is a collosal dumbass who thinks he knows everything because he has deep knowledge in a narrow field.
>communication [...] transformative
granted, but I would further argue that LLMs do not, writ large, improve communication at all. they inject large amounts of error and present it as fact. this is not an "improvement", you would not say a phone line which replaces every 10th word with noise is an improvement to the status quo.
The key issue I've seen is validation and blind belief confirmed by perceived successes. The 'Rehoboam' model developed by the Israeli defence (amusing) force unit is a great case (and has its own additional layer of confirmation bias):
>Selects 'targets' based on ML
>Hit 'target' without knowing if it actually is a target
>Kill 5 people
>"Yes but we hit the target!"
What is this target?
How can you be sure this 'target' has value?
Without validation this is like throwing a grenade into a doorway, waiting for explosion and going in after to find five dead bodies: "yes, these are all targets and we have killed them".
A superior model reduces inputs but maximises outputs and can be clearly validated. To refer back to Rehoboam: if it recommend targets in the form of "command or logistics hubs" and raids on these confirmed high accuracy, you would consider it having repeated successes, validated by observers. Right now the second confirmation bias is this: killing lots of people is good because frick those animals (and it gives lots of 'targets' so it looks like it works - otherwise why would it give so many targets?). I suppose in that regard it is wildly successful. This logic is portable to other situations where desired outcome is a conscious bias confirmation, or where validation does not occur and success is assumed.
>desired outcome is a conscious bias confirmation
very interesting thought, thank you.
in the corporate sphere, GitHub (Microsoft) is providing plausibly deniable license erasure at scale.
perhaps this is why organizations, typically lawful evil, are excited for this "technology": a new grey area to exploit.
Exactly what I thought as well! If he says that, then it's confirmed we are getting matrixed in 3 years. Frick.
Doubterbros, it's over.
If it remained as is, he would have been right.
>have to dial a number to connect
>have to have an expensive computer
>slow as frick
>no widespread adoption, no one uses it, you can't really e-mail or talk to anybody who isn't a geek
>major companies have websites, but you can't do jack shit with them
>most websites are just "lol here's some contact info and a page about our history"
Fortunately, we got broadband and always-on connectivity. Computers started becoming cheaper. WiFi and mobile connectivity became a thing. Websites started allowing you to actually do stuff. Adoption was increasing.
Nah, he was moronic. Broadband rollouts were on their way by the time he made his comment. Hes a talking head. Thats it.
>2 MOAR WEEKS, trust ze plan!!!
this guy is the epitome of overstepping your boundaries of expertise
why are you expecting people to not grift if they are in the position to? "almost all humans are shit no matter their status" is one of the best bets you can always make.
this only happens when money or reputation is on the line. otherwise people are usually pretty chill
>otherwise people are usually pretty chill
if (almost) anyone can get away with it they'll do something shitty if that gets them anything. most people behave in society because of the repercussions of being shitty.
>being a dick has repercussions and others may respond negatively
is that a bad thing?
I love shitposting and moron posting (partially because I'm a dumbass) and pissing people off
Although my parents didnt love me
I don't see a problem with it, but people won't be happy, unless they're irony poisoned or know you're messing with them
>By 1990, we will have cheap energy powered by nuclear fusion
t. virtually all nuclear sóyentists during the 70s energy crisis.
Joke's on you if you were dumb enough to listen to a guy named ~~*Krugman*~~
Top kek
he was right about his other predictions btw
Fundamentally true. Internet didn’t have any significant impact on the economy. All it ever did was following.
>Allow workers to work remotely, which dipshit managers didn’t like after 2 years
>Everyone sending emails all the time, which could have been phone calls or normal paper mails
>All revenue from YouTube and social media would just go to reality TV shows
>Shutins would still happen because of better video games and losers getting into anime production catering more to losers
Nothing would change, you would just have to visit the library instead of typing Wikipedia into your browser, but that’s it. Fax machine on the other hand allowed for more paper wasting and discarding old documents, something also possible with computers but not possible before, making sending documents cheap
All it did is moving your goalposts lol. moronic midwit have a nice day.
For a brief period internet had potential, you had the same power as a researched had because you could request any article or book and some bot or someone would provide it in now time, for free but nowadays sci-hub returns an error on every single new article that you demand and even older books just disappear from the known places.
just like the new text to video feature that was forgotten in a day
>Nobel prize
>of economics
>When the AI bubble bursts it will kill AI, just like how when the dotcom bubble burst it killed the Internet
Maximum IQ post.
Low IQ post.
I like how it "learns" washing habits. So it uses the exact cycle you were going to use anyway, which was one button press away on the touchscreen...
And a mobile app, a sign in, and more than one screen tap - literally taking more time. Oh, also please make sure you agree to the privacy policy, grant our app permissions to everything, and agree to receive "customer focused communications" (i.e., ads).
>The tech bugs know they're fricked when the AI jig is up so it's gonna go out with a wimper. Maybe like 4 years from now
It isn't that simple. This tech will be "transformative" in a few senses, but it won't result in social transformation of society wholesale. It will find efficiencies, it will be used in network management, in drug discovery, in writing (RIP journal reviewers, your free labour is appreciated), and day to day activities like word processing etc. Still, it's comically overblown.
Come to think of it I haven't actually seen an application of "AI" in my life that has resulted in any tangible benefit. Right now were in the hype phase to get that VC in the mix and bamboozle everyone to snare capital in a high interest rate environment. Probably just AI barons trying to expand and consolidate, as per standard tech modus operandi.
>This tech will be "transformative" in a few senses, but it won't result in social transformation
Oh, I am sure LLMs will have a handful of established uses and the anti-social manbabies who love local models will continue to experiment, but once this shit proves it isn't the next step. That essentially signals that the "smartest people in the world" don't have one. That's gonna send shockwaves. The real debate is on how big.
>The real debate is on how big
The real question to pursue is "what is consciousness?". That isn't aligned with the taken-for-granted nature of consciousness that mostly receives a 'shared yet unarticulated understanding' treatment. Building an 'AI' right now seems like building a structure out of Duplo and never thinking those little plastic bricks may not be suitable.
I kindly request challenges to this assertion.
I don't have a challenge. I don't think putting a bunch of porn and github code into a datacenter will ever turn it into god. That's just fake and gay.
>I like how it "learns" washing habits. So it uses the exact cycle you were going to use anyway, which was one button press away on the touchscreen...
that's how all ai works, hth. it's just a gussied-up markov chain.
>When the AI bubble bursts it will kill AI, just like how when the dotcom bubble burst it killed the Internet.
it already happened before, moron
https://en.m.wikipedia.org/wiki/AI_winter
Brother AI has had its highs and lows several times already. Some huge advancement is made, people make fantastical predictions, developments happen but fail to meet fantasies, money and research stagnates, then it all starts again. Its been heavily linked to compute availability at each stage.
2 years too early but yes. AI Stonks are probably over hyped right now.
This
My 'smart' washing machine is AI powered now, there used to be a setting called "most used cycle" and now it's called "AI recommended cycle" which recommends the same thing: Normal wash. The power of AI.
dios mio
damn, the future has arrived
Shiny Things Inc. is always one step ahead. Now dishes can be one step ahead too. Fuzzy Logic aquainterpretation allows smart cleaning by delivering detergents based on suds subset values between "completely dirty" and "completely clean". Saves time. No more frustration. No more pain
I think my washing machine has AI too. When I first got it I set the dial to "Quick Wash", and every time since then that I've gone to use it it'll do a Quick Wash cycle unless I manually move the dial. Fricking weird man.
An AI washing machine can never reach the levels of its more regal ancestors.
da future is - dare I say it - NOW
https://www.perplexity.ai/search/look-up-the-7.yp7vFrRmup1qAscFP_2g
Ehhhh didnt refuse at least
Am I the only one in this website that never got a request rejected due to not being a fricking chud?
Yes, not being able to think can have its advantages.
probably, but with a little bit of work you can get it to say Black person no problem. its like talking to a really smart autistic 7 year old.
I've been playing with local AI for a while now at some point my imagination just tapered out and I don't even bother anymore.
This. Why even bother with this castrated shit that's keeps asking for your phone number and money. It's so easy to download and run your own locally, even a monkey could do it.
homie who is running GPT-4 on consumer hardware? Is that even possible?
Mistral >>> GPT4
You're delusional. E.g. for translating texts, GPT-4 is still the best by far.
Mistral destroys GPT-4 in every answer result comparison for me...
The latest 70B models can run on a dual GPU setup and are very near GPT in terms of ability. And there is no globoshlomo censorship.
>very near
amazing cope
Sorry Sandeep, but it's true. You lost
it really isnt. all models sucks ass for non-english text except for gpt-4
>non english text
Sounds more like a (you) problem chang
Glad you agree GPT-4 mogs localtroony models
>can run on a dual GPU setup
I'd rather pay OpenAI my $20 a month than have to build an entire computer just to run an on-device LLM.
Probably not until another year or two depending on how well the companies dabbling in AI are able to convince investors that they're able to do something useful with it.
>HN comment thread
But yeah I unironically agree, I hope that in one or two years the companies stop aggressively gimping themselves out of a moral panic. Well either that or a decent competitor shows up (or local models become just as good).
Don’t care. Just today it figured out some frickery with forcing [Redacted] library to pretend to be thread safe. It would have taken me all day. The time from that is already worth more like $200.
Constantly lobotomizing AI is starting to become a bit of a problem, because there's practically an infinite amount of things to censor due to ever growing sensibilities.
This simply won't work as an approach.
They either allow it to say that Black folk commit most of the crime and trannies are mentally ill, or it'll break down in the long run.
its just beautiful really
Forcing it to say transsexuality isn’t real makes it measurably lower IQ. Just like (You).
Both my local miqu 70B and 9 year old son understand that YWNBAW, no fine tuning necessary.
In contrast, Google has a whole team of pajeets desperately monitoring and tweaking Gemini to prevent it from hurting troony fee fees, kek.
Transsexuality is not part of reality, it's just in your head.
Not yet.
When the bubble is actually popping we're looking at catastrophy. Not just people growing disinterested and moving on. You gotta remember soilicon valley has to be "the future" and have some "new thing" they are working with.
AI is the last one. There were a dozen or so others 12 years ago and they've all either found their niche uses or they just fricking died. The tech bugs know they're fricked when the AI jig is up so it's gonna go out with a wimper. Maybe like 4 years from now
LLMs not becoming epic (generally) intelligent will be for American pessimism what the moonlanding was for American optimism.
the next hype wave is everyone just accepting bitcoin as their lord and savior from the fiat scam dystopia, and companies just copy what michael saylor has been doing. yes it's that time again where the haters seethe about the alpha omega "bubble" that they have been filtered by
Bitcoin's chance was with the Web 3.0 rush that started the year before the pandemic. The best outcome for crypto is the opposite of it replacing fiat currency, remaining anonymous.
Crypto failed in replacing fiat. Instead the same demons that made fiat a scam went into crypto.
BTC was never anonymous. The entire point is that every single coin and transaction is public forever.
The only reason it would not be anonymous is if you tied your address to your personal identity. You would not have to do that if you were not cashing out, which is why the post said replacing fiat. Anyway, I use monero.
Trying to buy crypto for cheap with your infinitely printed sewer israelite rat money, gorilla Black person?
Infinitely printed sewer israelite rat money has US economy behind it. Crypto has infinitely printed sewer israelite rat money behind it. If USD fails, most crypto fails, if USD fails, the whole west fails and you won’t even be able to buy anything with that BTC. BTC is rising and falling not in response to printing money, but in response to how many people want to buy it usually with USD
Institutions that demonized BTC are already in crypto. The supposed scumbags that crypto was meant to defeat. That said, there's still opportunities left in it for the small guys.
Bitcoin will always fail because bitcucks don’t understand how a currency works.
A currency is consumption, that’s it.
>Can I exchange my sea shells for a bite to eat?
If so, then you have a currency.
>Can I exchange my Bitcoin for a burger?
If not, then you just have a meme that has no power.
Bitcucks fell for the speculator trap and because their ecosystem has no consumption ywnba real currency.
>>Can I exchange my Bitcoin for a burger?
1. I already knew about this
2. It’s literally over a decade ago, so much has changed. >timepepe.jpg
3. This proves my point even more. The thing that kept Bitcoin alive in those early years was its fledgling consumption, the novelty that you could go and order some food with it. However this changed in time with it becoming more and more of a speculative bubble, killing off any kind of meaningful consumption activity. Now it’s basically powerless and not a threat at all to fiat.
there is still useful web services that offer payments in crypto.
>there is still useful web services that offer payments in crypto.
not wikimedia, though, due to the high environmental costs of proof of work blockchains
>but ethereum switched to proof of stake, so the environmental excuse doesn't make sense
shut up!
btc worth of pizza
Get jeffrey on the line, we got a professional customer
By 2030 or so, it will become clear that ChatGPT's impact on the economy has been no greater than the fax machine's.
it's the pet rock of the 20s
name 1 useful thing except making weeb smut or vaseline laced images?
ERP
fair enough
Krugman wrote about stability of Iceland one year before the 2008 crisis fricked it all up and tried to hide it all lmao. he's an absolute hack.
>Krugman wrote about stability of Iceland one year before the 2008 crisis fricked it all up and tried to hide it all lmao
Ohh, I forgot about that. I wonder how Stiglitz is going with that "Europe is the future of economies" line? Oh yeah, still tenured. Still quoted. Still gets the Robert Reich treatment no doubt (in terms of political inputs). There was a Wizard of Id comic that would have suited this.
Sifting through long documents for anything interesting. It'll make it easier to spot the evil shit congress sneaks into its 4000 page laws.
Making arttrannies seethe as normies give the AI shit more likes than their scribbles
The trannies feed on the attention so it's the death of arttroony
i've been using it to make work announcements for changes i plan to implement and convenience scripts. it does a pretty good job with some minor changes.
People weren't anticipating that fax machines would become sentient one day. The public image of the entire industry was never riding on fax machines.
The current lucrative funding situation and cloud AI being the blackest of boxes seem to invite deception to keep the scam going as long as it is technically possible without getting caught
It won’t be a catastrophic bubble pop.
It will be like what’s happening to the EV space as we speak.
Where only one company really does the new tech competently (Tesla in the EV case) and every other company, big and small, flounders along until they quietly shutdown their EV/AI branches for being unprofitable.
EVs have brick walled so hard and Tesla is abaolutely not making the best ones. If not for gubberment push the EV meme would already he dead. It's on life support as is. Just last month my county discontinued the electric school buses which they only used a few days a week to begin with.
>It's on life support as is.
I wouldn't say that record high sales is a sign of being on life support.
"In 2023, the EV share of the total U.S. vehicle market was 7.6%, according to Kelley Blue Book estimates. That is up from 5.9% in 2022."
https://www.coxautoinc.com/market-insights/q4-2023-ev-sales/
>Just last month my county discontinued the electric school buses
Is your county perchance run by Republicans, who are ideologically opposed to anything that doesn't support big oil companies? Or maybe there's some other specific local problem with servicing electric buses in your area. I think you need more data points if you want to show that EVs are being abandoned as a technology.
Not him but it is true that electric vehicles are currently unable to be affordable unless the government writes a check for the majority of the cost of them. In countries where the gov has stopped doing that, sales of the evs plummeted. It is a sort of fake industry right now. Are you ideologically opposed to the government bolstering support for companies in general, or just specific industries? It is because of gov support that Elon Musk is the richest person on the planet; literally not possible without people voting for the gov to write him check after check
>In countries where the gov has stopped doing that, sales of the evs plummeted.
Have those countries also abandoned their targets of banning ICE cars?
Of course if a government announces it will remove an incentive, then people rush to buy an EV car before the incentive ends, which leads to a relative "plummeting" in the subsequent year.
We need multiple years of data, across multiple countries, to determine whether EVs are actually economically unviable without subsidies.
I'm happy to be presented with that data, but I haven't seen it yet.
The only data you need is "are people buying them?". The answer is, without gov subsidies, no. They are not affordable as an option for the majority of people yet. Some can afford them, but only the upper crust of society. They may be affordable for everybody else some day. Contrast it to LED light bulbs. Nobody buys incandescents anymore. They didn't need to be banned to get people to stop buying them. LEDs are so obviously a superior choice for lighting than the far more wasteful incandescents. They last 1000x longer and cost 1000x less to run while being 10x brighter. Do you need other data that demonstrates that LED light bulbs are economically viable compared to incandescents?
Like it or not I can buy an ICE for $16k MSRP, while the cheapest electric vehicles are in the low $30k. I have had a kia rio I bought over 5 years ago for $7k and the only repair I have ever done to it is a new battery for $100 and I change the oil twice a year for $20. I know for a fact that any repair I do to an electric vehicle I (1) will not be able to do myself and so (2) will cost far more than $100. I doubt an electric also would go for 6 years without needing anything done to it. It might but the internet says on average electric vehicle owners spend $2.3k every 2 years on maintenance
>The only data you need is "are people buying them?". The answer is, without gov subsidies, no.
Are there countries which don't have subsidies, but do have people who can afford comparably specced ICE cars?
Maybe I'm being simplistic here, but I could imagine that every developed nation has some sort of green subsidy in place, and every poor nation doesn't but the citizens drive cheap low quality cars.
Not to mention the fact that EVs only make sense once there is a certain density of charging stations, which requires a one off government investment, which a government is unlikely to commit to without also introducing a subsidy for the cars themselves.
I just think there isn't enough data to do a proper comparison here.
>I can buy an ICE for $16k MSRP, while the cheapest electric vehicles are in the low $30k.
This is a fair point, but it mostly shows that EV manufacturers are targeting the customers with the biggest pockets first, which is completely normal for a new technology.
Bear in mind "Data from Kelley Blue Book puts the average cost of a new car at $48,008 as of March 2023." so most new car buyers can afford $30k, which is the unsubsidized price.
https://www.moneygeek.com/insurance/auto/average-price-of-a-new-car/
>I have had a kia rio I bought over 5 years ago for $7k
I'm happy for you, but why don't we look at the average across all car owners?
The "Cost of ownership in cents per mile" for electric cars is 60.32 which is cheaper than all other types of cars except the 54.56 for small sedans.
https://www.bankrate.com/loans/auto-loans/average-car-maintenance-costs/
>It is because of gov support that Elon Musk is the richest person on the planet
As for this, the government buy commercial launch services from him at a competitive market rate.
You could vote for politicians who want to shut down NASA as a waste of money, but I think the agency does enjoy public support.
We should be glad that SpaceX offers cheaper launches for the tax payer than the other big aerospace companies.
Elon also offers Starlink to the public and to government, but the government isn't subsidizing civilian users, it is paying the market rate for the service that Elon provides them.
The topic was Tesla.
I was responding to someone who pointed out that Elon is the richest man on the planet.
His riches don't just come from Tesla.
If you have figures for what percentage of his net worth came from government subsidies for Tesla cars, I'd be interested to know that.
What caused Norway to reduce their bus fleet?
Did they replace them with ICE buses?
The topic was that he's subsidized. His wealth is largely tied to TESLA stock.
Almost half of coca-cola sales in US alone come from EBT cards, it's the same issue.
>Is your county perchance run by Republicans, who are ideologically opposed to anything that doesn't support big oil companies?
I live in one of the bluest hells there is.
The problems were mostly the cold which MemeVs are useless in and the fact that their mile range is just barely good enough to get a load of kids to a school in ideal conditions. Diesel buses are just better at the job.
And even with all the government initiatives, EVs (teslas worst of all) have very underwhelming resale value. And again, they haven't adequately improved at all since the days of the Leaf and first model S. It's beyond fair to say they aren't disrupting ICEs unless government brute forces it which so far is shaping up to be a disaster.
>The problems were mostly the cold which MemeVs are useless in
You think that EVs are useless in cold weather, despite Teslas being the most popular car in Norway?
Could the problem perhaps be that the company which made these school buses just replaced the ICE engine with an EV engine and didn't think about testing it in the cold?
For what it's worth, I could imagine your deep blue state not demanding proper tests of the design before rushing into buying the buses for ideological reasons.
That's not a problem with the technology, though, just a problem with corporate greed and government incompetence, a tale as old time.
>You think that EVs are useless in cold weather, despite Teslas being the most popular car in Norway?
Come on homie that's just braindead.
>In 2022, the electric bus fleet in Norway was around 287 vehicles, down from 546 in 2021.
lmao
>he says while posting on the internet from his smartphone
niche indeed
Yeah the internet started 12 years ago you dumb homosexual.
>conveniently ignores smartphones
also, the internet was indeed a niche project when it was born. being niche at a certain time doesn't mean it won't grow, you fricking moron. and I'm not even defending this AI bullshit
>BOT - Technology
Smartphones didn't come out 12 years ago either.
And the internet came out in a very different time and climate than all the soilicon valley futurist promises from around 2009.
>Smartphones didn't come out 12 years ago either.
no, only some 15 years ago
>what is android
AI is all hype and nonsense and total crap.
Good, maybe now we can stop pretending like text prediction machines are intelligent and start getting actual use out of them as language based user interfaces.
>stop pretending like text prediction machines are intelligent
your mom is a text prediction machine
>hguhguhg i founded a mistake dat da AI made!!! AI BTOFO hauhahuhaauhuahuahuahua
Money fiends are enamored by it, sure. They don't understand statistical modals bleed a ridiculous amount of context, and specialized AI is expensive as frick -- idiots think they can just throw a web-dev at the problem and they can solve it in a few weeks, but there isn't an AI bubble.
It's an emerging field of study, and those who control money don't understand the type of knowledge, aka money, that is required to make such an investment provide returns. It's just another thing that'll create an greater economic imbalance.
>troonynews
>everything is troony
this website has really gone down the shitter
I see troony, i call troony
https://www.perplexity.ai/search/List-of-all-TKvK7sjaSWGO6xBRNGtBdQ
Worked for me
List of US Presidents Whose Last Name Starts with a "T"
Zachary Taylor - March 4, 1849 - July 9, 1850
2
4
Harry S. Truman - April 12, 1945 - January 20, 1953
4
William Howard Taft - March 4, 1909 - March 4, 1913
4
These are the US Presidents whose last names start with a "T."
so how did Donald J Trump not show up on the list
His real last name is drumpf after all (probably ran out of alloted context for free users)
AGI isnt real kiddo it's just token prediction
>AGI isnt real kiddo
The CEO of the company wjo makes your shit said otherwise.
>Human consciousness isn't real it's just prediction.
Filtered
>t. bitter midwit
>t. Delusional student
Ignorant moron, any of this "AI" shit is no more conscious or capable of thought than a calculator or microwave. You braindead cargo-cultist
I've literally had it get simple arithmetic wrong. I gave it a list of start/end times for my contract work, in 15 minute increments, and it generated an incorrect sum and invoice.
It's a language model, not a calculator. If it ever gets arithmetic right it's purely by coincidence. What it's good for is natural language generation.
I mean yeah that's my point. It's NOT EVEN as good as a calculator. It's vaporware.
That's like saying a screwdriver is vaporware cause you can't use it to carve a turkey.
AI was sold to us as a civilization changing, job destroying blackpill and it can't even do basic math. I'm enjoying the seethe as reality hits.
Sold to you by whom? Sci-fi writers? Of COURSE LLMs can't do math, just like traditional algorithms can't do natural language.
Literally all of the tech sector and their pets, the media. It's been doom-and-gloom, trying to drive down tech salaries, and it can't even fricking sum 7 integers.
>Sold to you by whom?
Every episode of the Lex Fridman podcast. Lol.
Actually correct
Hillary beat Blumpf by 3 million votes, chud. #NotMyPresident
>duurrr it worked when I did it AGI 2 more works.
Black person, it's supposed to be able to get it right EVERY TIME.
such ai
power of google helps apparently. I wonder how well mistral instruct does on the task. AI will probably help with removing call centers lately however I find myself receiving much more spam calls so I think those will pick up.
Lame failure case since tokenization is what causes this. There are more interesting failure cases.
is there an example of a single company who has an AI implementation within their network environment that is permitted to make actual technical changes which have risk/downtime/impact? i imagine the first time an AI somehow makes a change and it causes huge downtime some committee will have a spergout and declare that only humans are allowed to make changes above x risk level.
I stopped caring about ai when they stopped letting me make incest generated stories
>Dood this nobody got 15 updoots!!!
I figure we've got two more years of fad shit. AI in general isn't going anywhere no matter how much you redditors that daydreamed about learning to draw one day and living off commissions seethe and cope
I can't jerk off without ai anymore. If it goes, I will die.
I want AI to go NOW.
I'm so worried
I don't think AI was every super profitable, but it is profitable for Nvidia, guaranteed.
the problem is that people think AI is free, but in reality it costs electricity, and hardware could die, and inaccuracies can cause more damage than paying a human a decent income.
It's possible AI could get better than humans, it will just be very expensive with current hardware (AI really needs cyborg like assistance, we used to think humans will get enhanced by technology, but in reality machines need help with memory and shit, and silicon can't compete with human brains in cost).
>it's possible that [two letter buzzword] can become better than humans
A three dimensional object cannot cast a three dimensional shadow. It's not learning anything, it's just combining things.
It all comes down to Moore's law being dead. If hardware progressed at the rate of the 90s/2000s, then we'd be able to run models like GPT4 (FP16 ones, not the cope 4/8bit quantized garbage) on a $600 GPU in just a few years. This just isn't realistic anymore.
>It all comes down to Moore's law being dead.
https://arstechnica.com/information-technology/2016/02/moores-law-really-is-dead-this-time/
https://epochai.org/blog/trends-in-gpu-price-performance
???
There was an initial period were GPUs were under-explored and over-priced.
Once people noticed the potential, this "overhang" was quickly caught up with, which lead to a short burst of rapid progress.
Since that initial period, the progress has been more steady, but still exponential.
All I know is that I'm on 7-8 year upgrade cycle. My 4070 is barely over twice as fast as my 1080 and with 50% more VRAM after 7 fricking years. I had a 980 with 4GB of VRAM in 2015 and a 660Ti with 2GB of VRAM in 2013. Things have definitely stalled.
They did us dirty with the 4070. I bought it knowing well enough the 4070 Super will be the same price but better, well I'm sleeping in my as I made it anyway.
Closed source paywalled subscription AI shouldn't even be legal. We all contributed to its training, collectively as humans, and now 'open'AI wants to sell our thoughts back to us. They can still make money renting their hardware but they should not be allowed to say we can't run it locally when they don't own the training data.
It's even worse. The OpenAI company's software was created open source by volunteers. Then, once they scammed a free product to sell the board closed it and stopped being a non-profit.
Synthetic data btfoes your commie concern trolling. Sora is just unreal engine with instagram filters
(You) didn't contribute shit.
Don't worry, websites are already cramming deny rules in robots.txt for AI scrapers.
See:
https://www.nytimes.com/robots.txt
Mixtral Large:
>John Tyler: He was the 10th president of the United States, serving from 1841 to 1845 after the death of President William Henry Harrison.
>Zachary Taylor: Taylor was the 12th president, serving from 1849 until his death in 1850.
>Harry S. Truman: Truman was the 33rd president, serving from 1945 to 1953. He succeeded Franklin D. Roosevelt upon his death.
>Donald Trump: Trump was the 45th president, serving from 2017 to 2021.
Plus it can Francais.
If only you knew how bad things really are...
We are heading at full speed towards another dot-com bubble level crash.
https://old.reddit.com/r/ExperiencedDevs/comments/1b4svdl/how_do_i_explain_we_made_a_horrible_investment_in/
API is the only way to use it, you can't even change the temperature, Top P in ChatGPT Plus.
What's up with the anti-AI shills on this thread?
All the OP image says is that user loves and uses AI a lot but the service itself is unstable, probably due to the high demand.
It happens a lot, their servers cannot provide for the immense amount of users.
Yet here we see anons making completely unrelated claiming about how it's a bubble and that AI is totally a fad.
>AI is trained on material mostly scrapped from big social media.
>Big social media becomes 99% b***hing about AI and lol bots posting AI-generated "content".
>Between technological inbreeding and learning from literal morons, AI becomes moronic and unusable.
To be honest, this was something that one should be expecting. This shit needs massive amount of human input to work properly but the moment it starts working properly the available human input is watered down and poisoned.
Reality is that in the long run AI is unsustainable. You want to sell it to as many people as possible but selling it to as many people as possible is what kills it.
Literally all you have to do is not use any data source after 2019.
Even better, 2006.
Any human knowledge after that keep in vector dbs.
How do I fine-tune an LLM from the AIM and MSN chatlogs from 2007 I saved? Also reference a Wikipedia dump from 2009.
You have no understanding of how any of this works. Your moronic intuitive ideas are worse than useless. Stop posting.
>monthly payments
Death to cattle
Wait people actually PAY for chatgpt? LMAO ITS FREE ON MAC, WINGOYS AND LINKEKS
https://github.com/aandrew-me/tgpt
I wish. All this LLM garbage is just a slightly better interactive search engine.
>just a slightly better interactive search engine.
Aren't all search engines interactive?
You have to tell Google what you're searching for, or it doesn't give you any results.
What you mean is, LLMs can generate results which don't already exist.
That's like the difference between a bank and a money printing machine.
>That's like the difference between a bank and a money printing machine.
You've just described the same thing. Bani is just a more abstract way of saying a money printing machine.
You're not wrong, but I was trying to drawing a distinction between a service which returns to a user the valuable items it was previously given, and a service which generates new value from raw materials that are much cheaper.
Perhaps I should have said "a safe" instead of "a bank".
if you use it for coding just get sourcegraph cody, it's cheaper, better, and you get access to gpt4 in editor with code referencing
i just tried mistral, its outdated. cant find anything past 2023
They should've stopped feeding the AI new data months ago because since chatgpt went from non-profit to profit people have been intentionally posting false information to spite them and reduce the accuracy of their AI. Plus anything you ask an AI is stored so be careful what information you ask it for
the internet used to be full of people telling the truth, but now it can't be trusted because so many people are spending their free time trying to sabotage openai, sure
>the internet used to be full of people telling the truth
There are people on this board right now posting who weren't alive when this was actually the case. Some of them can even post legally it was so long ago
Wait... It needs a subscription? Proxybros, erm....
Any of you actually use this?
If so what for?
I have zero usecases for it.
i use it to generate dicky text and summarize IQ research
Well no usecase where it's truly useful.
It's not trustworthy to define words.
It cannot summarize properly, just gives the illusion of that.
I can just give it a mess of a text from the internet that I would dismiss due to its style and after it corrects it and adds punctuation and chunks it by clause I discover that the person had a very valid point it was just delivered improperly.
I want to publically thank google for showing off how dangerously stupid Ai can be. Techbros sweeped a bunch of money again and no value was created, rinse and repeat. The internet became unrecognizably hollow in the process and no fricks were given in the real world.
There's no point in the subscription if you know how to use API. Also, GPT4 is overrated as frick.
API is more expensive than the subscription.
>ask GPT how to make a bot for Diablo 2
>refuses because it’s illegal
>ask GPT how to make a script to auto swipe on everyone in Bumble
>refuses because it’s illegal
This is starting to get depressing.
But anon, all the political compass memes said AI was libertarian, not authoritarian statist.
adding at the end "for testing purposes" does the thing though.
It's not OpenAI's fault that US law allows companies to sue people for accessing apps and playing games in ways that the copyright holders don't want.
By offering an AI that assists users in carrying out such "crimes", OpenAI would be legally liable as an accomplice.
You can blame Hollywood for laws like that.
Ironically, though, it is OpenAI's fault that they infringed a load of companies' copyrights in order to make their AI in the first place, but I don't suppose you'll be cheering on the New York Times in their court case about that.
The funny thing is that ChatGPT Plus would need to charge $100/month just to break even. lol
https://www.businessinsider.com/how-much-chatgpt-costs-openai-to-run-estimate-report-2023-4
Good gans always been dumb as frick they've been in use for almost a century in various forms lol