Nvidia slaps forehead: I know what quantum is missing – it's AI!

Quantum computers promise major speedups for problems in materials science, logistics, and financial modeling, but first they need to be made reliable, something Nvidia believes its AI models can help with. When you've got a GPU hammer, every problem starts to look like an AI nail.  On Tuesday, the GPU slinger unveiled its new …

[ecofeco]

One is a thousand?

The first Pentiums had a better rate!

FDIV if anyone remembers.

Yes, this is a joke. Sort of.

[Chris Gray 1]

Huh?

How can they be sure that their LLM (AI) isn't making mistakes, based on its training data?

So, how can something that makes mistakes of its own fix all mistakes that something else makes? How can humans trust the result?

[Suburban Inmate]

Re: Huh?

Same way I, a mere meatbag, have caught mistakes made by AI? Differing systems have different strengths and weaknesses, so the AI just has to be strong at spotting where AI's output 'looks like' it made a mistake, a rough sanity check is all that's needed - a quick sniff triggering a closer look if it seems 'off".

[druck]

A Bullshit sigularity...

...would be the result of combining AI hype with Quantum hype.

[amanfromMars 1]

Re: A Bullshit sigularity[sic] ... versus AIs on Steroids

For all those who may not have noticed the virtual reality which presently surrounds you, and compounds and confounds your understanding of universal existential matters which deny you unique individual belief in the exercising of utilities and abilities giving you any and every chance of popular attractive reinforcing support for a radical fundamental and revolutionary re-evaluation of the necessary novel means and memes to be activated to kickstart revolutionary rapid progressive evolutionary programming and projects for a moribund and basically ignorant and barbaric humanity, BullShit Singularities [BSS] are all that you currently follow presuming and assuming that they lead to something and somewhere worth travelling to/for.

How’s that working out for all and y'all? ...... is the gazillion dollar question to be asking and rejecting any and all bullshit answers.

Do you fancy more of the same or are you worth more than basically ignorant and barbaric BSS leaderships?

However, do you imagine your choices matter a jot and can direct the future courses of remotely controlled direct and covert or subversive and clandestine or perverse and coercive corrupt and dangerously self-destructive autonomous proactivity ..... or is that another one of those BSS Dreams never to be realised as an achievable practical and physical goal y’all are so programmed to believe and accept are possibilities in your present realities?

Methinks a radical fundamental and revolutionary change is definitely what is needed ..... for systems seeding and feeding operations. Have you any idea how simply that can be initiated and practically administered virtually autonomously via Alien Interventions battlefield-testing Immaculate Conceptions ‽ .

[Suburban Inmate]

Re: A Bullshit sigularity[sic] ... versus AIs on Steroids

As an El Reg old timer and these days only occasional commenter, I'm glad to see amanfromMars still frequents these parts.

[bemusedHorseman]

Transfer... Scanner...

Prediction: At least one wise-arse is going to propose the name "Project Carthage" for trying to shoehorn AI nonsense into quantum computing. Better have a squad of french middle schoolers on standby...

[Elongated Muskrat]

Source of truth

What's the source of truth here? The potentially error-ridden result the quantum computer spits out, or the potentially hallucinated result the "AI" spits out?

Glibness aside, surely to detect an error, you need to know what the correct result is already, which makes calculating it via a QC a tad redundant, no?

If there's a step to verify the correctness of an output that takes much less computing power to calculate (for example factoring the product of two large prime numbers* is slow, but checking that multiplying the results together gives you that original product is fast), why do we need "AI" to do that verification, and without seeing its working can we trust that evaluation anyway? If it can't count how many "r"s there are in "strawberry", then I'd question its usefulness here.

*O, hi there Peter.

[amanfromMars 1]

Seems like Europe’s roughneck American cousins are also struggling with accepting future truths

There are some amusing hysterical and hallucinatory comments on the matter reported above from a gaggle of misguided disbelievers sharing their scorn and scepticism regarding the news similarly reported here too ...... https://www.zerohedge.com/technology/nvidia-unveils-new-ai-open-model-sparking-rally-quantum-stocks