Qubits missing
Quantum grammar checker says s/cable/capable/
The US National Academies of Sciences, Engineering, and Medicine issued a report this week on the state of quantum computing. It's a subject of some concern given speculation that such devices could render current encryption schemes worthless. The good news is your secrets should be safe from quantum prying for at least n …
I guess that depends on what you mean by fusion power. My understanding is that fusion takes place only in the core of the Sun, and that it takes photons thousands of years to get from there to the surface. It's of course not the same photon ending that journey and emerging into space that started it (assuming photons have an identity in the first place).
* strong AI
The original concept of AI i.e a robot brain as or of greater intelligence than person is still a long way off. However in certain limited applications we already have AI. The google search engine for example
* flying cars
Flying cars is easy. Getting a flying car that a non-pilot can fly and not crash into the myriad other flying cars is hard
* driverless cars
They seem to be a long way down that route, with some applications already ongoing. Whether people will accept them on the road is another questions, but expect military applications in the next few years
* cure to cancer
To all cancers, no. To some cancers, yes
* fusion power generation
Yes, unfortunately a long way off. However the problems are engineering and material science not physics
It all goes to show that technology progress is rarely quantum leaps, but slow hard slogs often relying on many different areas.
Just to add that you also have to balance "risk and cost of not achieving anything" against "potential benefits and possible spin offs"
So
strong AI, cure to cancer, fusion are strong categories in the latter column, while driverless and flying cars are more in the former camp (although crack AI, and you have driverless cards anyway).
Quantum computing is in the latter camp. Yes we may not achieve the final goal, but in doing so we gain a greater understanding of the quantum world. However if it is managed on a commercial scale, the benefits extend way beyond encryption. remember that most of the progress in the last 30 years has been due to increase in computing power, a curve which using present technologies may well have to stop
although crack AI, and you have driverless cards anyway
That depends on what you actually mean by "AI", which changes decade by decade.
Firstly it meant tree-searching algorithms (a machine that can play chess).
Then it meant fuzzy logic and expert systems (a machine that can diagnose disease).
Then it meant pattern matching and neural networks (a machine that can recognise faces).
None of these is anything like the public perception of AI, which is more along the lines of "I, Robot" or "Ex Machina": a fully self-aware, "living" machine.
If we get the latter, then it will be able to drive your car. Whether it chooses to or not, is another matter.
The original concept of AI i.e a robot brain as or of greater intelligence than person is still a long way off. However in certain limited applications we already have AI. The google search engine for example
about 15 years ago I did a degree level course about AI. There were only really two things I remember about it:
* Strong AI is about ten years away
* Strong AI will never happen, because whenever you get close to it, people will dismiss it as algorithms and brute force, not intelligence.
Anything much short of Star Trek's Data isn't Strong AI (IMO). Google self driving cars are very clever, but they aren't AI. It'll weave its way through time square, but stick it in the middle of a cattle ranch or on the moon and it won't know what the hell to do, even though it's still basically the same job.
"Google self driving cars are very clever, but they aren't AI. It'll weave its way through time square, but stick it in the middle of a cattle ranch or on the moon and it won't know what the hell to do, even though it's still basically the same job."
Stick a city slicker in the middle of nowhere, somewhere totally novel where none of their experience can really apply, and watch them struggle, too. It's the schadenfreude from the struggle that makes survivalist media so popular.
* However in certain limited applications we already have AI. The google search engine for example
The classic mistake of re-branding data-processing as AI.
* Getting a flying car that a non-pilot can fly and not crash into the myriad other flying cars is hard.
And land. Let's not forget landing. Any idiot can take off and fly and largely be successful at avoiding other things in the air. But getting safely back on the ground again? That's the REALLY hard part.
Almost as hard as building a personal vehicle with sufficient energy store and efficient propulsion to have a range that makes it useful for anything beyond saying "See! We built one!".
* Expect military applications [for driverless cars] in the next few years
But don't hold your breath for the broad and tumultuous changes to infrastructure and legal liability laws required to make widespread, meaningful personal use feasible.
* [cure] To all cancers, no. To some cancers, yes
Yes, the problem here is the categorisation of many disparate conditions with a common underlying cause into one "curable" thing. One might just as well posit a "cure for viruses". A cure for cancer is a null proposition.
* the problems [with developing fusion power] are engineering and material science not physics
That's true of any new technology. There was nothing stopping Neanderthal's from developing fusion power. The physics was the same, the lack of engineering and material science was the only problem. So sure, we can have anything (subject to the laws of physics being observed) we just need something unspecified and currently unknown and possibly unknowable to make it possible. You might extrapolate from this that it is only a question of time, possibly. But to put a timeBOX on that discovery, is to attempt to draw a long bow where the bow string is not fixed at one end - try it.
:)
I'd rather that a genuine "AI" didn't drive my car thanks. On the other hand, a computer system with outstanding sensors, control and prediction systems... yes.
The problem with the cure for cancer is that most people don't understand that cancer is not a disease as such. It's "just" the body's own cells mutating/malfunctioning, which they do all the time, and getting themselves into a state where they are not shut down automatically (the fate for most mutated cells) and reproduce in an uncontrolled way. If cells never mutated then we would not exist, it's a fine balance. So targettting our own cells which are malfunctioning when our own body's defences against malfunctioning cells have missed them is a tall order.
> And land. Let's not forget landing. Any idiot can take off and fly and largely be successful at avoiding other things in the air. But getting safely back on the ground again? That's the REALLY hard part.
Landing at an airport is actually pretty easy - commercial planes have had fully automated landing systems for years. It's very much following rules.
Landing safely in someone's garden wouldn't be that hard if the flying car has drone-like flying characteristics. Choosing a sensible landing spot might be harder, but at worst nominated "safe places" could be marked up on a map. Not landing on top of another flying car or human is probably the most difficult part, but that's comparable to the job a self-driving car has to do.
"Landing safely in someone's garden wouldn't be that hard if the flying car has drone-like flying characteristics."
The thing is, probably the most-desired feature of a flying car is the ability to go (within fuel limits) between any two arbitrary points without having to deal with things like traffic and so on. So VTOL on a driveway or curbside is going to be an assumed goal.
Not if the government orders them to stay hush and denies its existence, on penalty of endangering national security. Recall that the F-117 was "black" for quite some time. What's to say a working quantum computer wouldn't be kept hush-hush as well?
We have yet to understand what drives quantum behavior and what its limits are. A few days ago a preprint paper was posted suggesting (credibly) that quantum phenomena might not be universal, this means that quantum phenomena might break down when certain thresholds are crossed.
If the hints provided are proven to be correct (and we have very good reasons to believe that they are) then we will never be able to build large quantum computers only very small ones.
People working on quantum computers today are missing the big picture, I really doubt they will someday produce anything that is mildly useful, perhaps in the very far future someone will be able to pull that off but it will be not be them, they are too clueless.
This seems a common thread in science.
Scientists are single minded, never see the big picture, and are only interested in funding.
this is normally put forward by people who have no specialist knowledge in a particular area, but latch onto the summary of a outlying research paper.
University research tends to be pretty open to new ideas. Proving that quantum computing is not possible would almost as big a break through as proving it is.
I'm pretty sure there are many quantum computing researchers who are aware of the paper and are taken it into account and are more qualified to understand its applicability and ramifications.
Saying that when commercial concerns get involved there is a a tendency to downgrade any information which stops the pork barrel funding, so you need to differentiate between basic research and commercial application
Quantum annealing appears to work to some degree, that indicates that quantum computing might be possible, on small scales at least it appears to do its thing.
But we need more, they'll have to go and do something that solves a wider range of problems while working at scale, that is all the proof that is needed. I wouldn't hold my breath while they're at it though.
Quantum annealing appears to work to some degree, that indicates that quantum computing might be possible
Why do you think quantum annealing says anything about the practicality of large-scale general QC?
Tunnel-effect transistors work great. They don't demonstrate anything about the practicality of general QC either.
I'm still not seeing anything most developers would recognize as a computer.
IE something you can actually program
Y'know, not by wiring it up, changing its function by what instructions it executes.
That seems f**k know how far away from reality as it's always been.
Your encrypted data is stolen today including the key exchange bits. Don't be smug.
If your data has a lifespan longer than 10 years (say, the names of all the spies and moles in <name your country>, or your GDPR protected data where your company is bankrupted by the brusselcrats when the data is revealed, or your carefully constructed pile-o-shell companies for tax evasion) you are exposed when that quantum computer pops into existence. Yes, I know, the inflexion isn't like that but you get the drift. And it could be never, or 10 years from now, or 2 years from now, or 2 years ago that a suitable QC exists to crack vulnerable encryption.
The data has to be resistant to quantum attack n years before a QC attack is feasible, where n is the time value of the data.
Better hope that QC are further than 10 years away, because it will take longer than that to modify the infrastructure to be quantum resistant... on the other hand, it is a brave new world for stealing valuable resources. The number of vulnerable points is truly astonishing, QC as the supernal zero-day.
The data has to be resistant to quantum attack n years before a QC attack is feasible, where n is the time value of the data.
That's a naive threat model. Data has to be resistant to attack by GQC machines until that attack's cost drops below the value of the data - just as with any other attack vector.
Even if the NSA has a unicorn-powered large-scale GQC machine now (vanishingly improbable), it is orders of magnitude less likely that using it to using it to crack a large number of keys is possible, much less cheap enough to be worth doing. Given the vast number of qubits required for QEC for decent-sized problems, even a big-enough-to-be-useful GQC machine will almost certainly be applied to only a handful of extremely valuable problems.
I think research into post-quantum crypto is swell. It's nice that RLWE and similar algorithm families are becoming feasible for everyday use. But the data-lifespan arguments for PQC are mostly based on some highly unlikely assumptions.
"But the data-lifespan arguments for PQC are mostly based on some highly unlikely assumptions."
I think that's the biggest problem, too. I don't think people realize that data can have a longer lifespan than they think, especially if they're "extended" by being chained with related bits of data.
D-Wave, which sells quantum computing hardware
D-Wave sell adiabatic quantum computing hardware (assuming they're not just selling snake oil). That's very little like general quantum computing hardware.
Adiabatic QC is fine if you can represent your problem as annealing a spin glass. (Fine, but maybe not useful; last I checked, it wasn't clear that there were many real-world problems for which AQC provided any real-world advantage.) It doesn't solve problems in BQP in better-than-deterministic-computing time.
Asking D-Wave to comment on this report is like asking a manufacturer of roller skates to comment on the automobile industry.