Ask HN: What is the biggest problem LLMs solved in your life/work?

After reading about the recommendation system breakthrough[1], I'm more curious about just how much we're leaving on the table with classical algorithms. If you raised the amount of money being funneled into quantum computing and spent it purely funding classical algorithm research, would you be better off?

[1]: https://www.quantamagazine.org/teenager-finds-classical-alte...

cs702 · 3h ago

> Today, IBM (NYSE: IBM) and AMD (NASDAQ: AMD) announced plans to develop next-generation computing architectures based on the combination of quantum computers and high-performance computing, known as quantum-centric supercomputing. AMD and IBM are collaborating to develop scalable, open-source platforms that could redefine the future of computing, leveraging IBM's leadership in developing the world's most performant quantum computers and software, and AMD's leadership in high-performance computing and AI accelerators.

Translation from corporate-speak: "Maybe we can chip away at Nvidia's dominance by working together and promising something Nvidia can't offer?"

As I understand things, firing up a bunch of GPUs is still more cost-effective than any quantum computer available right now.

Nonetheless, I wish IBM and AMD lots of success. It would be nice if Nvidia gets real competition!

cjs_ac · 3h ago

> YORKTOWN HEIGHTS, N.Y. and AUSTIN, Texas, Aug. 26, 2025 /PRNewswire/ -- Today, IBM (NYSE: IBM) and AMD (NASDAQ: AMD) announced plans to develop next-generation computing architectures based on the combination of quantum computers and high-performance computing, known as quantum-centric supercomputing. AMD and IBM are collaborating to develop scalable, open-source platforms that could redefine the future of computing, leveraging IBM's leadership in developing the world's most performant quantum computers and software, and AMD's leadership in high-performance computing and AI accelerators.

Has anyone found a real-world problem that's best solved by a quantum computer that isn't cryptography? I exclude cryptography because if the only thing these machines are good for is breaking ciphers, then governments won't let anyone else buy one, will they?

dijit · 3h ago

Materials Science and Drug Discovery would suddenly become a lot easier, along with financial modelling (of our entire society possibly) and logistics/supply chains.

They would also be much better at training ML and doing pattern recognition.

Basically anything that requires a massively parallel computation on undeterminable states that are only clear in hindsight. They’re really important actually and its only an unfortunate side-effect that the same solution breaks all our cryptography.

(of course: the offensive wings of our defence ministries really enjoy that side-effect)

atq2119 · 3h ago

> Basically anything that requires a massively parallel computation on undeterminable states that are only clear in hindsight.

If only. This description makes it sound as if quantum computers could help efficiently solve all problems in NP, which is not believed to be true.

Those "undeterminable" states need some non-trivial algebraic structure so that destructive interference of states can do its magic in a quantum computer. Finding such a structure is incredibly difficult, if it exists at all.

kevinventullo · 3h ago

Basically anything that requires a massively parallel computation on undeterminable states that are only clear in hindsight.

From https://scottaaronson.blog/ :

“If you take nothing else from this blog: quantum computers won't solve hard problems instantly by just trying all solutions in parallel.”

mcmcmc · 3h ago

Better financial modeling? Oh boy, who’s ready for quantum dynamic pricing to really squeeze your wallet to the max

heisgone · 3h ago

Is IBM still a serious player? It seems to me they are living out of legacy software. Is there quantum computer stuff promising?

wmf · 3h ago

IBM really is a leader in quantum computing but it's not yet clear whether that field will ever have real-world value.

gentooflux · 3h ago

And if it does, that value will likely fall somewhere between 0 and 1.

bee_rider · 2h ago

I think it’ll be exactly one, or exactly zero. That’s the quantum part, right?

staringback · 3h ago

What do you mean? IBM is the world leader of AI development thanks to their cutting edge IBM Watson

stirfish · 46m ago

I can't tell if you're being serious. Watson is... fine. You can try out the granite models for yourself.

orionblastar · 4h ago

This reminds me of when IBM, Motorola, and Apple did PowerPC.

beeflet · 3h ago

IBM is still doing POWER. I have one running on my desk

ranger_danger · 2h ago

Is that the world's largest heatsink? Either way it's beautiful.

Towaway69 · 4h ago

Are they combining forces in an attempt to overtake nvidia in the AI space?

guerrilla · 3h ago

I feel like this is just another publicity stunt.

xfactorial · 2h ago

For what i know, IBM R&D is truly a marvel in terms of Innovation, but one thing is “we managed to Create a process to make xyz” and another is, indeed, putting it together at scale.

5 nanometer was something they worked on, but it was TMCS the one who actually made it happen.

Perhaps this is a good chance to put to work some of that research using AMD Manufacturing.

Let’s see how it goes.

pryelluw · 3h ago

From IBM?! No …

solardev · 3h ago

Why does the future of computing sound like we're back in 1999?

Who's next... Intel and SGI? Rockchip and Cyrix? Nvidia must be positively trembling...

Mistletoe · 3h ago

Wait until you see my LLM crunched on my Voodoo card.

solardev · 3h ago

I mean, technically Nvidia did buy 3dfx... so you're not far off :)