ChatGPT predicts Switch 2 emulation by late 2025

But I thought ChatGPT was this super duper smart high tech being doe???

Too many NV_API garbage to clean up and fix on switch 2, requires Nvidia to have working drivers on Nvidia and Windows to compare tom, currently that's not a thing, so.
Nope. Not happening unless you're ready for a the year + of yuzu devs in exiles trying to poke around and figure out how it should work under MESA.

James Jones Jeetware strikes again.

not accurate information. no$gmb was in 1997, not 1998

American AI just can't cut it.
glowing closed source American AI certainly cannot.

Bingsisters, we can't stop losing.

I doubt nintendo previously used thugs to threaten emu devs

SAAAAR

When are we going to get AI that isn't blatantly retarded? Will we get that in three more years, perhaps?

she

This explains EVERYTHING since it's speculated that Trump used it for his tariff shenanigans.

captain Britain

FUCKING KEK.

valuing anything chatGPT has to say

Literally today in another thread I talked how AIniggers don't even want to think for themselves or do their own research and want AI to do it for them.

mmmmmmmmmmmmmmmmmmmmm let me consult my chart.
says here Nvidia are still skimping on vram, still have export controls, still gimping consumer cards, oh right ah, yeah, they're also not producing consumer cards so local models won't flourish
no plans to change in 2025 or even 2026

let me check if vram is in short supply, gddr 6, 6x & 7.
says here it isn't, and that nvidia are the ones artificially gimping bus speeds huh

well whaddayaknow. I can't wait until 2032 when the RTX x070 finally gets 16 whole gigabytes of gddr7x without a crippled bus. By that point by that point server farms will be getting rid of those A100 80's they've got stashed for $-50 USD (scaled to inflation of course) and they'll be scooped up in seconds as local modelers and image gen scum fight unoptimised CUDA workflows.

Well, there's always 2034/2035 huh! Let's not get too down.

Why the fuck are you retards replying to these dogshit threads?

You realize that chinkshit AI is literally just forked from OpenAI, right?

It's probably just inferring this from troll posts online that say Switch 2 will be emulated right away because its underpowered, etc.

If AI is so stupid why is the job market for computer programmers the worst it's been since before Personal Computers were even a thing?

We might not even see that happen because investors are starting to pull out of the AI industry.

AI does not equal gpt-tardo.

no lol

GBATemp hackers are cucked by Nintendo, they won't do shit

An emulator is likely but the yuzu dev will be extremely cautious about it

As far as I understand, most of programming is taking existing code, joining it together, and then slapping your own name and/or trademark on it.

If a computer can do it for a fraction of the price, why the fuck would you hire in-experienced grunts? It's like the checking industry when credit cards became a thing...

Are you pretending that the Indians that corporations keep on hiring didn't have an impact on this and you're solely blaming it on AI that can barely code without being tardwrangled?

ChatGPT is retarded.
Realistically, Switch 2 emulation is entirely dependent on the Switch 2 getting hacked. You'd better fucking pray that voltage glitching works on this thing. If not, we're looking at probably more like 2032 and only on the vert first launch day firmware version if it ever happens. Horizon is extremely secure.
Once the thing gets hacked, though, it'll be relatively quick. A year, tops. I can't imagine the OS is hugely different than the switch (which itself was not very different from the 3DS - citra was reused for yuzu mostly intact.)

As far as I understand, most of programming is taking existing code, joining it together, and then slapping your own name and/or trademark on it.

So you never did programing at all.

What the actual fuck is that last article?
It was just revealed that NVDA have been cramming China full of their nerfed cards for the big 3, going off the deepseek injection they got.
The big think standing in their way right now is (ironically) TSMC, nvidia squeezing everyone to stay relevant and the US seeing AI as cold war fuel with their incompetent leader trying to line the wallets of the people who funded him.
I'm curious to see where China will take AI, and I'm curious to see where language models head post-deepseek. We're seeing some real nice improvements come out of everywhere but the hebesoftnvidiapact act crew. Even the company the other incompetent acquired has chops.

the indians could barely code either though. Infact I argue that shit like Claude and ChatGPT are basically advanced jeets that can put out their half-baked code way faster but it's less problematic to clean up and turn into functioning code

No, but I know a lead dev at a major news site, so I can ask him later.

But in the novel amount of programming classes that I took, that pretty much was the go-to, less you wanted your program to be broken as fuck.

They don't make hackers like they used to

Impossible when most of the data it's scraping to train is made by literal retards.

If a computer can do it for a fraction of the price, why the fuck would you hire in-experienced grunts?

The same reason government departments exist.
It's an industry and function that relies on caretakers to maintain.

-but yes, the basics are easy enough. It's the after part that's hard. What people who are going into building with AI are also finding is they still need people to maintain their code and one we all know how easy that is, right?

You hire one lead dev to review code and make alterations as needed.

Only an idiot would release un-vetted code.

switch 2 will be the first world console with denuvo

mark my words
it will be uncrackable

The Switch 2 has like 10 Watts of power to draw from, if it were running Denuvo in handheld mode there literally might not be any power left to run any ganes

It's already going to be hard to get into the thing regardless. Horizon's anus is tighter in terms of security than HURD. It's not impossible; it's not to the level of say, seL4. But still, it's a REALLY hard target to exploit barring hardware design flaws.

One dev

I suppose I meant 'small team,' but you get the gist.

Not to veer too off-topic...

Humans, as a whole, as terrible fucking drivers

AI can drive just fine, but will get into the occasional accident

People get uppity whenever it does, despite accounting for a fraction of accidents

By virtue of being created by humans, AI is imperfect, and will forever need a human touch until it becomes full sentient.

Obligatory: fuck Musk

most people realized a lot of programmers do absolutely fuck all and started to hire visajeets instead that are technically cheaper but also do worse than fuck all.
there just isn't a real demand for programmers because there's nothing new calling a new trendy wave that can't be solved by Indian slave workers. Apps/PC programs have stagnated, games are becoming way more simplified with UE5, and the only real work anyone needs is for AI and hardware computation which is much more math/science based than generic programmer based.

I didn't know this. I wanted to be a computer programmer when I was a kid but I became a NEET instead. A few years ago I thought I'd maybe put my life back together and learn how to code and finally do the one job I actually wanted to do, but that's also because for years and years I heard it was the most lucrative job. If it's dead now, I guess I'll just keep being a recluse.

But I didn't call AI stupid, I just said ChatGPT has never answered a question I've asked it correctly. I don't know why. I thought it was basically just a step-up from using a search engine, but for me it feels way worse. If I wait and ask the same questions at a later date, I get better results, I guess because people have come around and corrected its mistakes.

For example, the first time I asked it what the six Hatsu types are in Hunter x Hunter, it fucked up pretty hard. It got 4/6, but the fifth was one of the others with an alternate name and the sixth was a completely different aspect of Nen. Now I just asked it again and it correctly named the six, but the descriptions aren't entirely correct and it's weirdly redundant (it reminds me of when someone doesn't know what they're talking about so they end up talking in circles).

I've asked it math/science questions, too, and it always dances around and eventually recommends I register or whatever to use the more advanced version.

Do it for yourself pussy. Not for the money.
Holy fuck, you're a neet who wants to be indoctrinated into slavery: You are a failed normalfag and a failed neet.

It will be crackable but nobody may want to do it

didn't someone have arbitrary code execution on the switch before it came out or am i remembering that wrong?

For example, the first time I asked it what the six Hatsu types are in Hunter x Hunter, it fucked up pretty hard. It got 4/6, but the fifth was one of the others with an alternate name and the sixth was a completely different aspect of Nen. Now I just asked it again and it correctly named the six, but the descriptions aren't entirely correct and it's weirdly redundant (it reminds me of when someone doesn't know what they're talking about so they end up talking in circles).

Why the fuck are you asking these kinds of questions to the computer

You need money to survive, anon.

It's something I care about and shouldn't be difficult to answer. You yourself could answer it by just googling it, so why can't ChatGPT?

Try asking it some cases about famous criminals with well documented cases that have been scraped.
Say, someone who was a mail bomber. It'll do the same actions that anon described, where for a first few prompts (or even those thereafter) something gets lost and it pulls wrong, adjacent details. It's kind of fascinating to see it happen and try it on different models.

of course they use GPUs you fucking cretin, you are 100% stupid, remember this next time you try that gay as fuck reddit "You do realise..." line again

Because suits haven't experienced the consequences for the replacement yet.
Generally speaking, AI makes programming that works, not programming that holds up to actual use cases, and it doesn't know what it actually made in a critical sense.
AI generated code is notoriously vulnerable to exploitation because of how "dumb" AI is. It makes something that works, with no concept of the circumstances it may have to endure, and when AI generated code breaks, it breaks the ENTIRE thing, and usually has to be rebuilt from scratch because nobody but the AI knows how this code works, and the AI can't fix it because it's stupid and thinks it should work.
I encounter this shit regularly now. It's not a widespread problem yet, but it's just an issue of the fuckups of AI code not yet building to terminal mass, and "essential" coders not usually being replaced yet to plug the leaks.
AI being "dumb" ALSO means that if someone can figure out what you used to generate something, they can crack that shit open in a few minutes, sometimes with the help of the thing itself.

ChatGPT's prediction would be based on previous game consoles.

Why should you fact check a thing designed to answer questions, by asking it questions you know the answers to, and know how to find the answers if you didn't, to see how good it is at providing accurate information

Just trust the plan, bro

The reason why job market is bad is India and financial crisis (most of programming is R&D and nobody wants to invest in R&D during recession).
AI can barely write few lines of code. I know that because I work on AI projects. AI is used in classification, ocr, translation, transformation of text, search reranking etc. But not to generate the code. Sometimes it can generate some bullshit function like "remove duplicated references from text" but it falls apart once you try to force it to use specific open source library. It's useful for generating use-cases for automatic testing though.

The cope around Switch 2 never being able to be hacked (and therefore emulated) is hypercringe

try a decade at least. LLMs are inherently broken and proper AI will require an entirely different solution

The Yuzu dev is in jail

The guy that went to jail for yuzu is in jail
The yuzu dev is not in jail

RCM on Tegra X1 was already documented before Switch released. Was just a matter of finding how to enable it.

AI can barely write few lines of code

I assume you're exaggerating. It can definitely churn out tons of code, especially code that follows a common pattern, but you better be experienced in whatever codebase or architecture you're using before reviewing it. I'd love to use stuff like Claude and Cursor to make me a perfectly functioning complex video game in half a day, but it's really hard to trust what the bot spit out when I see it making weird mistakes and decisions in existing code that I'm familiar with, let alone code that I'm completely unfamiliar with. Especially since it tends to run rampant if you're not specific enough with your prompt. Shits definitely impressive and useful but the more you actually use it the more you understand its limitations.

I used it on real code for banks and other financial institutions. It can generate small snippets for well defined problems and that's it. The problem is that in real world, problems aren't well defined and writing the code is only a small part of software development.

Tech-illiterate retards really think the Switch 2 cant be emulated or wont be emulated within a year lmao These faggots dont even know how to torrent these days. Just shut the fuck up, buy your overpriced trash console, pay premium prices for games you will never own and enjoy the shit you are eating.

Yay.