About a quarter of South Koreans already pay for AI services, compared to just 2% of Americans
South Korea is preparing to offer free generative AI services to its entire population, using the technology to help people handle everyday tasks such as booking medical appointments, finding housing, and sorting out taxes.
The program is intended to expand access to homegrown AI systems while giving the government a larger role in how they are used for everyday services.
I think it will be interesting, to see how countries with high AI usage fair in a few years.
why wait, the results are already in for people, studies of HS students/college using AI to do all thier HW, or thinking have come out with worst reading, writhing and math competence, than the older generations.
Lol they are pumping hard.
RIP Korean students in formative years learning critical thinking and social skills.
Govt-funded Chatbot about to gaslight them that every idea and question is genius-level, and that social connections are something a computer can fulfil.
What students? They’ve been shutting down school because they have no students to teach. They are on the leading edge of a demographic trend the ‘developed’ world will soon see.
Care to explain? Sounds interesting, my last impression was that SK was a society with massive competitive pressure in education and career. What happened?
There aren’t any children being made. Therefore no students to teach. Same is happening all over the world but way higher pace in SK. It’s easy to forget how shitty their society can be with the glitz of k-pop culture distracting us. I don’t mean that as an insult specific to Korea. Many other countries look great from the outside but is never without issues.
Younger person looking at their own society collapse around them because having children is too expensive: at least this old generation will take their stupid dogma with them along with their institutional knowledge of machining, woodworking, trades, etc.
Older person looking at their own society collapse around them because having children is too expensive: at least we made the dogma stick!
It’s all related. Massive societal pressure = no time for children. No time for children means birth rates are significantly below replacement rate. At first the issue was ignorable. Now it’s not. Cracks are starting to show. With so few children, nobody is “picking up the torch” for virtually very industry, and whatever workers are left are dwindling as they retire.
You see a similar thing happening in Japan, too. And a similar issue is occurring in America and Europe, though their situation is not as bad in comparison. It’s speculated that all advanced economies will eventually face this challenge, but South Korea is really trying to speedrun total societal collapse
Declining birth rates.
There’s nothing inherently wrong with declining birth rates when your population is so oversized. If you try the “there’s nobody to look after the old people” argument, that’s infrastructure and priorities, not an actual resource shortage.
I wasn’t suggesting it was wrong or anything, just offering an explanation. Personally, I reckon there too many fucking people in the world.
You know that you can use ChatGPT for free, right? Also, paid plans aren’t so expensive that a student couldn’t afford one.
If you think free access is the only thing between students using AI and learning traditionally, I’m afraid you will be disappointed.
Free AI won’t change much for most people, except that the government can choose what models the citizens will use. This way they can be independent of US companies dominating the market.
This is home.grown south-korean AI we’re talkong about here. It will never be satisfied with anything you do and will always criticize you no matter how good you are.
The initiative is part of South Korea’s effort to build a stronger domestic AI industry and reduce its reliance on systems developed in the US and China.
I get it. And even though this arrangement gives me some major government-corporate complex ick, I think it’s better for all of us to limit the global reach of the US and Chinese mass surveillance operations.
The entire SK economy is built upon major government-corporate complex ick
Risky click of the day.
Btw, were we already giving -gate suffixes to everything in 2014?
Remember tan suit-gate? Mustardgate? Of course we were
Oh man, my memory from 2010s are fading away already. Thanks for the reminder.
I’m still waiting to see who implements cyberpunk first: the Sovereign Chaebol of Samsung or the United State of FAANG.
The technology isn’t useful enough in the short term to afford crippling your labor force in the long term. Stop using it.
SK’s population problems too, and its worse than japans.
The services are expected to connect to government systems rather than function solely as standalone chatbots. Residents will be able to use them to book doctor appointments, search for apartments, and get tax guidance. Small businesses will be able to calculate taxes and check whether they qualify for government support. Parents will receive recommendations for educational content.
I doubt if any of this is a good idea. Put my trust in an LLM, and specifically for the more important aspects of my life? Not for me. But maybe people will. I have learned I’m bad at predicting the masses.
The government plans to distribute up to 512 Nvidia B200 chips among the three consortia and help cover operating costs.
That seems like a very low number for a population of 50’000’000 of which a fourth is currently paying for AI access. I mean it’s less than what would be in sixteen NVL36x2 racks. That’s like one aisle in a data center. And they want to split it over three groups… okay?
I am beginning to wonder if this is a genuine attempt, or if they just want to pump the KOSPI, since Samsung and SK Hynix profit so much off of selling shovels (the shovels are memory in this case).
I am beginning to wonder if this is a genuine attempt, or if they just want to pump the KOSPI, since Samsung and SK Hynix profit so much off of selling shovels (the shovels are memory in this case).
The entire Korean stock market is just 3 companies in a trench coat as seen when the bottom recently fell out.
Same thing happened recently in Japan too with the US diving to try and save them by selling bonds and buying yen to no avail.
But there’s no bubble, right? I’m sure that won’t happen to the US, the stock market is way more diversified and it’s not 5 companies all trading the same money back and forth, right? Right?
Can we abolish the stock markets yet?
Maybe that’s the plan for giving unlimited token access: responses capped at 1 token per second
You’d be surprised how far it goes when you aren’t doing training. I’m serving an entire university on 1/2 of a 6000B
Good god, those pollution centers will be on max throughput…
I’m reminded of something I saw last year.
Do they want everyone to become stupid?
This is an interesting approach. It depends who owns and provides the compute. If it’s the state or one of Korea’s quasi-private firms Samsung/LG/Hyundai that are effectively a part of the state, then this could have some positive outcomes in encouraging use cases that don’t seek profit maximization for the compute providers. In such a reality people could actually apply LLMs to productivity-increasing use cases in all sorts of sectors.
I do think it is healthy for a country to have their own AI for their people. I think they are way better off keeping their own data in their own country.
And I do love some LLM - but I don’t think the general population actually better with it…
“That’s a great question, Do-yun, and of course I can look after that for you! In the meantime, why don’t you and the Mrs keep on with that making babies task the government set?”
At least it will be shorter lived there.
I’m not sure I understand what you mean. 25% seems like a pretty high uptake, I don’t understand the attraction but that’s a lot of people. Will that higher number mean that they will end up not seeing any worth in paying for it sooner than folks in USA?
You got it. LLMs cost lots of money, waste lots of power, unemploy swaths of people, causes psychosis, and produce terrible outputs with liability tacked on. The more of it they use the more they’ll realize how useless it is and move on. You see it happening all the time on smaller scales, for an example: Taco Bell goes all in on AI order kiosks and drive through, people order 9999 waters, CEO realizes the AI isn’t cheaper at all and cuts the whole program. That actually happened.
Taco Bell goes all in on AI order kiosks and drive through, people order 9999 waters
This also speaks a lot to the state of humanity. I may well be immature or uncreative, but if presented with an AI ordering interface, I would never think about this! I do deal with the public a great deal through the course of my job, and am frequently disappointed at the level of (my) perceived narcissism. Folks are butts.
Only an absolute concaveman would look at the failures of AI and assume humanity is the issue.
“you’re holding it wrong”
Only an absolute concaveman
fair haha. I am older and avoid AI at all costs. Sounds like I need to learn a bigger picture! Also curious what you mean by concaveman… Same as caveman?
i presume it means their economy will crash sooner than USA’s.
Is that a thing that is being considered? I have to admit I do not know a lot about South Korea’s economy, but based on automotive and (I think) tech sectors, they sell things world-wide. At this point, it seems like USA has an economy crash risk. IMO
they have the same problems we see in other advanced economies:
- aging population
- lack of housing
- not enough births
- household financial pressure
- using trading/gambling as an additional income stream
They do have well known global industries but workers generally-speaking aren’t paid enough / have limited free time outside of their work. To add more pressure I believe the government increased the pension which is typically paid by the working population who are already financially stretched as it is.
They’re on track for destabilisation in the next 20-30 years.
Plus most of their high tech have been stolen by China in recent years, such as competition from CXMT.
Wow, these are challenging things. It sounds like they are in a tight spot, even if I do not understand where AI subscription fits in. 20-30 years in essentially tomorrow. It must be on their minds, and maybe explains using AI for what, I don’t know, but something.
I’m no expert so take what I say with a grain of salt.
LLMs are prohibitively expensive to run.
If I wanted to make a basic rollerball pen, the cost would be high because I would have to gather the resources just to make a single unit. Mass manufacturing drops the price considerably to make it affordable for me to buy a pen.
But tokens aren’t the same as a factory line. If it’s $1 for me to use an LLM then it’s $100 for 100 people. It scales poorly.
LLM availability over cloud infrastructure requires resources that are expensive to build and depreciate quickly to the point that ROI is untenable (ie. Nvidia’s GPUs).
Now imagine a stock market that hinges around a handful of companies that are involved in that infrastructure component (eg. SK hynix). If many retail investors have dumped their savings into those stocks which can’t provide an expected return then the value of people’s savings will inevitably fluctuate on the whims of the AI industry.
You can imagine what will happen the next time a Chinese company releases a free and efficient LLM that can run on an 8GB RAM personal laptop. The panic selling will reverberate through the markets and a lot of people’s savings will evaporate overnight.
But tokens aren’t the same as a factory line. If it’s $1 for me to use an LLM then it’s $100 for 100 people. It scales poorly.
They still scale to some degree. There’s a certain amount of VRAM you need to even load a model (frontier models likely over a terabyte now), but once you’re there, adding extra users is less cost since you only need to accommodate their KV caches on top of the model of itself. You’ll need to add more GPUs once you’ve saturated what you have, but it’s nothing compared to the ~5 B300s for the first user (they actually come in servers containing 8 GPUs that cost about 500k EUR each, so realistically you’d buy one DGX B300 server to start with). Once you’ve loaded the model itself which has a fixed size, all the extra capacity you add is free to use for KV cache (which means more users).
depreciate quickly to the point that ROI is untenable (ie. Nvidia’s GPUs)
It’s true, but the reason they depreciate so quickly is that there’s a new generation out every 2-3 years that does more using less energy. You could actually run old GPUs, it’s just cheaper to buy new ones if you’re a hyperscaler.
I have to admit that I do not understand most of what you said :(
But, I can see that (I think) you are saying that many people in South Korea invested in AI, and that is problematic. I do know that, as you said, LLMs are resource intensive and do not really work financially. If many people there invested in AI companies, I could see them also subscribing to them. I think.
My avoidance of AI has lead to me not knowing enough about it. I don’t know what you mean by token… This is a common theme for me today! I won’t be able to avoid this if I don’t educate myself.
Why though 💀
While true spawn agent “spawn more agents”
Boom. Only takes one angry mf.
Literally cant go tits up.
As an armchair internet commenter, South Korea sounds like the #1 Cyberpunk dystopia prototype
