keenmaster486 wrote on 2026-07-08, 18:07:

Don't shoot the messenger. All I'm saying is I have never heard anyone criticize a Japanese person's English by calling it "Engrish", even though I agree that the term is not exactly accurate when applied elsewhere. But terms often run away from their face value meanings, or were improperly applied in the first place,

Agreed with the above, the people that invented said phrase have only ever used it for written language, never for the spoken word.

Just as foreign cultures gain enjoyment from placing English words on objects like shirts, we in the English speaking regions also gain enjoyment from seeing said objects however cringeworthy.

And the Japanese specific connotation died 20 years ago. If you were to ask Gen Alpha, it’s usually in reference to properly spelled English words and slang but combined in random confusing ways. They are too young to remember when it actually was 80’s Japanese shirts, games and signs .

Originally it covered all Asian and eastern sources but now includes further abroad with Africa and India entering the mix of folks feeling the need to place English words on their stuff. Some Muslim materials meet the low bar of entry as well.

We are flattered the Africans are now placing English language words on objects and nothing serious happens when I mix up and heat some African sourced “Roast Cock Soup” mix for lunch.

people please - we are past that language discussion, let's not perpetuate it and try to stay on topic 😀

twiz11 wrote on 2026-07-11, 01:04:

but theres things ai can do that humans cant. I feel inadequate and need to use them to keep up. They can spout things ive never thought of. an indivual cant fight the machines. heh Fight the future and welcome to the machines

huh, what things?
All I've really seen LLMs spit out is just really generic stories or images, or poorly designed and unmaintainable code; the only thing I've seen them do that is actually noteworthy is somewhat passable on-device machine translation.

& Especially in niche subjects, they also often spit out incredibly incorrect or outdated information because those things aren't prevalent in the training data.
(For instance, to use a personal example, I do Spore modding sometimes and I've seen people trying to ask Gemini about Spore modding-related questions... In one case, it made up completely new and nonexistent classes in the game's C++ code, like "cOcean", when there is no in-depth ocean system that even warrants an ocean class, and the modapi's documentation also has no ocean class either.)

LLMs can't quite get old computers either so there's always a regurgitation about public opinion (like Voodoos being the fastest, which they are actually factually not) or misunderstanding how the Sound Blaster works that gives their "retro goodness" away. Sometimes even ancient Dreamcast fan exaggerations/claims of PowerVR2 causes a feedback loop of what PowerVR is, as i've seen the PCX2 referred to as a "high-end card" lately.

leileilol wrote on 2026-07-11, 21:13:

LLMs can't quite get old computers either so there's always a regurgitation about public opinion (like Voodoos being the fastest, which they are actually factually not) or misunderstanding how the Sound Blaster works that gives their "retro goodness" away. Sometimes even ancient Dreamcast fan exaggerations/claims of PowerVR2 causes a feedback loop of what PowerVR is, as i've seen the PCX2 referred to as a "high-end card" lately.

Well an LLM can only go on the information it's been fed in training, if the information is flawed then the output will be flawed too. It isn't just in an LLM this happens, I've been lead up blind alleys and fed wing information on here myself. You take an LLM as a starting point as with any research into something.
I've found even copilot can with the correct prompt shaping churn out some decent info

leileilol wrote on 2026-07-11, 21:13:

LLMs can't quite get old computers either so there's always a regurgitation about public opinion (like Voodoos being the fastest, which they are actually factually not) or misunderstanding how the Sound Blaster works that gives their "retro goodness" away. Sometimes even ancient Dreamcast fan exaggerations/claims of PowerVR2 causes a feedback loop of what PowerVR is, as i've seen the PCX2 referred to as a "high-end card" lately.

LLMs are completely unaware of many high end but sometimes moderately common devices from 1995-2005 (not even very old) because the amount of web data and personal sites from that era that was easily researched even 10 years ago is completely erased from the modern web now.

When I wanted a copy of a driver or information for Seiko and Shinko printers (that I own) I was used to doing a web search as opposed to locally storing data but can’t do that anymore, the amount of lost information especially the last few years on many inane topics is massive.

We are living through the start of the period of great erasure where physical evidence is unlikely to exist because magazines, disks and other hard physical objects that would at least describe a thing are rarely created and getting rarer.
With only digital descriptions the great sweeper of bankruptcy will continually erase history

we are living in the era where big companies are trying so much to take any hardware away from people. soon you will only have a display , the rest will be subscription... you will have a nvidia geforce now subscription with "hours left" for your games , subscription for your music , already in full force , subscription for your router , basically , will own nothing. and u will be at the mercy of big cvompanies. and oif they want they can shut you down , impose new rules , sell your data to the goverment , etc...

they are so desperate to take physical processing power from the people... tech , ai , computers are slowly transforming these big companies and governments into a dystopian society. the want to watch you , control you , 100%.

They don't want to "control" you, they want to maximize their regular streams of income. Producing a great new game or a great new graphics card every few years is difficult to achieve. Making one great thing once and then leasing it to you forever is easy.

It's because companies are run by mediocre people who don't want to create great things, they just want to optimize the very narrow mission of "keep the company alive for as long as possible".

If you told an AI that its sole mission was to keep the company in the green, regardless of the products and services the company actually exists to create and provide, this is exactly what it would do.

keenmaster486 wrote on 2026-07-12, 20:25:

They don't want to "control" you, they want to maximize their regular streams of income. Producing a great new game or a great new graphics card every few years is difficult to achieve. Making one great thing once and then leasing it to you forever is easy.

It's because companies are run by mediocre people who don't want to create great things, they just want to optimize the very narrow mission of "keep the company alive for as long as possible".

If you told an AI that its sole mission was to keep the company in the green, regardless of the products and services the company actually exists to create and provide, this is exactly what it would do.

hence there should be laws forcing companies to do x or y...

but wait... governments are more corrupt than companies...

bitzu101 wrote on 2026-07-14, 18:22:

keenmaster486 wrote on 2026-07-12, 20:25:

They don't want to "control" you, they want to maximize their regular streams of income. Producing a great new game or a great new graphics card every few years is difficult to achieve. Making one great thing once and then leasing it to you forever is easy.

It's because companies are run by mediocre people who don't want to create great things, they just want to optimize the very narrow mission of "keep the company alive for as long as possible".

If you told an AI that its sole mission was to keep the company in the green, regardless of the products and services the company actually exists to create and provide, this is exactly what it would do.

hence there should be laws forcing companies to do x or y...

but wait... governments are more corrupt than companies...

gov't, corps, unions are bodies of people... legal fiction that we give rights to to humanize the herd. The indivivual gives up their sovergnity in exchange for protection.

AI has no soverignty like like indivuals which is why i understand

Talk about mediocre people... there is little difference between large corporations and the government in terms of the quality of the people in charge of them.

Everything great that has ever been done was accomplished by relatively small teams of highly competent individuals who were given a blank check to "just do the thing".

keenmaster486 wrote on 2026-07-14, 19:57:

Talk about mediocre people... there is little difference between large corporations and the government in terms of the quality of the people in charge of them.

Everything great that has ever been done was accomplished by relatively small teams of highly competent individuals who were given a blank check to "just do the thing".

offcourse it s no difference. there is no incentive.

pure capitalism is so bad... just as bad as pure socialism.

u need to be somewhere in the middle.... big corporations will only care about profits and big govt bodies are easily coprrputed and bribed.

the govt will sell you and me to the highest bidder... and nowadays , certain companies are so big and influential , it s criminal.

Pure capitalism and pure communism are ideals that aren't workable in reality (thankfully), but I know what you mean.

You'd think the U.S. system is just dandy for business but apparently it's not enough:
Breaking the chains of legislation and regulation is what the promise of this current bubble is all about.

In space you don't need to worry about environment and such pesky things.
The resources are handily there for the taking too, free of charge.
No need to deal with workers unions with an AI workforce.
And the market potential for AI? Just about every freaking company out there.

That's the vision or rather, the fever dream of this new unbridled capitalism.
But I very much doubt that any of this promise will be realized any time soon.

Don't get me wrong, the tech itself has some potential and it's getting better.
It's just way overhyped, the current investments are unlikely to be earned back.
I guess this is all Bubbles 101. You just can't ever know how long the music will keep playing.

mr.cat wrote on 2026-07-16, 08:17:

Pure capitalism and pure communism are ideals that aren't workable in reality (thankfully), but I know what you mean. […]
Show full quote

Pure capitalism and pure communism are ideals that aren't workable in reality (thankfully), but I know what you mean.

You'd think the U.S. system is just dandy for business but apparently it's not enough:
Breaking the chains of legislation and regulation is what the promise of this current bubble is all about.

In space you don't need to worry about environment and such pesky things.
The resources are handily there for the taking too, free of charge.
No need to deal with workers unions with an AI workforce.
And the market potential for AI? Just about every freaking company out there.

That's the vision or rather, the fever dream of this new unbridled capitalism.
But I very much doubt that any of this promise will be realized any time soon.

Don't get me wrong, the tech itself has some potential and it's getting better.
It's just way overhyped, the current investments are unlikely to be earned back.
I guess this is all Bubbles 101. You just can't ever know how long the music will keep playing.

There are valid reasons to say that AI has it s own space and is beneficial to humanity. that is correct 100%.

But because of the capitalistic system , in the beginning it s all a bubble. Just like the .com bubble was. then will settle down and be more down to earth...

Right now , it s a rinse of the people. Hardware is so freaking expensive it s stupid.

The US keeps saying that they are trying to keep an advantage over china , etc... but they failed to understand that at the end of the day , it s not the DESIGN that matters the most , but the physical factory making the actual product. and because these huge companies keep consolidating , etc , there are 3 main manufacturers. 2 in korea and 1 in usa. 3 main companies is crazy.

the bubble will burst soon , thinking next year or so... it s impossible to keep it up like this for longer...

The trouble is that ai/datacenter build in the us is being used primarily to track all movements and behaviors of citizens.

Any system that is primarily turned inward tends to dehumanize and punish innocent people.

We don’t live in a Star Trek universe we’re everything is used for a just verdant kind benevolent purpose

Nah, they already have things to track citizens with, AI datacentres are way too energy inefficient for that.
even just TVs are more likely to be used to spy on you, which- okay, smart TVs probably do, but still.

I can't really think of any other use case for even just one of the huge datacentres that'd be profitable enough to actually keep the lights on in there once the bubble's burst. Maybe a handful of the server racks end up in universities where they can actually be used for research that requires immense parallelism, and maybe the buildings get repurposed to host regular servers (which might be used to process data from stuff used to spy on people), but the AI datacentres themselves are being built with no plan outside of "yeah but the dotcom bubble paid off!!!".

VanillaFairy wrote on 2026-07-23, 00:21:

Nah, they already have things to track citizens with, AI datacentres are way too energy inefficient for that.
even just TVs are more likely to be used to spy on you, which- okay, smart TVs probably do, but still.

I can't really think of any other use case for even just one of the huge datacentres that'd be profitable enough to actually keep the lights on in there once the bubble's burst. Maybe a handful of the server racks end up in universities where they can actually be used for research that requires immense parallelism, and maybe the buildings get repurposed to host regular servers (which might be used to process data from stuff used to spy on people), but the AI datacentres themselves are being built with no plan outside of "yeah but the dotcom bubble paid off!!!".

Think of what the East Germans did for surveillance of its citizens, manually. A government in which the people do not hold the power to control the volume of its currency, will not hesitate to spare any expense to control its citizens. When a government puts itself over its people they all are dangerous, not matter what idealogy, though that is the central premise for some.

VanillaFairy wrote on 2026-07-11, 13:53:

All I've really seen LLMs spit out is just really generic stories or images, or poorly designed and unmaintainable code; the only thing I've seen them do that is actually noteworthy is somewhat passable on-device machine translation.

& Especially in niche subjects, they also often spit out incredibly incorrect or outdated information because those things aren't prevalent in the training data.
(For instance, to use a personal example, I do Spore modding sometimes and I've seen people trying to ask Gemini about Spore modding-related questions... In one case, it made up completely new and nonexistent classes in the game's C++ code, like "cOcean", when there is no in-depth ocean system that even warrants an ocean class, and the modapi's documentation also has no ocean class either.)

That has been my experience as well. It either writes something ridiculously long that says nothing or makes up stuff that doesn't exist. The GPT updates haven't really affected the quality of the output, but the writing style changed from "high school student padding out a homework assignment" to being an arrogant prick (for example, auto-generated status requests in Jira). I haven't tried any kind of machine translation except for the translator in Firefox, which is useless.

VanillaFairy wrote on 2026-07-23, 00:21:

AI datacentres are way too energy inefficient for that.

Nobody cares about efficiency anymore. Just add that "AI" buzzword to everything. Besides, everything else made in the past four years has been getting less efficient.

VanillaFairy wrote on 2026-07-23, 00:21:

even just TVs are more likely to be used to spy on you, which- okay, smart TVs probably do, but still.

They do. It has been proven.

Shagittarius wrote on 2026-07-23, 03:40:

Think of what the East Germans did for surveillance of its citizens, manually. A government in which the people do not hold the power to control the volume of its currency, will not hesitate to spare any expense to control its citizens. When a government puts itself over its people they all are dangerous, not matter what idealogy, though that is the central premise for some.

lti wrote on 2026-07-23, 04:06:

Nobody cares about efficiency anymore. Just add that "AI" buzzword to everything. Besides, everything else made in the past four years has been getting less efficient.

I meant moreso that a datacentre with regular server parts would save the government a whole lot of money on spying compared to the "AI"datacentres. the GPUs just aren't really needed for spying, there's ways to spy on people that don't cost the same level of billions...
Actually, speed cameras should suffice for tracking peoples' movements for car journeys at least; just gotta send the data of each car to get stored somewhere instead of just using them to fine people who break speed limits.

VanillaFairy wrote on 2026-07-23, 00:21:

Nah, they already have things to track citizens with, AI datacentres are way too energy inefficient for that.
even just TVs are more likely to be used to spy on you, which- okay, smart TVs probably do, but still.

I can't really think of any other use case for even just one of the huge datacentres that'd be profitable enough to actually keep the lights on in there once the bubble's burst. Maybe a handful of the server racks end up in universities where they can actually be used for research that requires immense parallelism, and maybe the buildings get repurposed to host regular servers (which might be used to process data from stuff used to spy on people), but the AI datacentres themselves are being built with no plan outside of "yeah but the dotcom bubble paid off!!!".

one day? most of the smart tv s spy on you. they record and send everything they can about you.

monitors do it too...

https://www.youtube.com/watch?v=Q9uefFYe6bM