Will the first AI Superintelligence be installed on a classical computer, a quantum computer, or some other alternative?
13
140
380
2030
3%
Quantum Computer
75%
Classical Computer
20%
Other
1.8%
AI Superintelligence is very unlikely to be built, or is impossible.

A lot of speculation abounds as to when AI Superintelligence will emerge, or if it can do so to begin with. But very little is discussed on what kind of computer hardware an AI Superintelligence will be be installed on. So, I will pose the question here.

Both AI and quantum computing have seen rapid progress in recent decades, but so far the bulk of AI development has occurred on classical computing architecture. Some have predicted that AGI, and also AI Superintelligence soon thereafter, will be developed or emerge from existing systems in as little as 5 - 20 years. On the other hand, it is possible that we might see the first commercially viable quantum computing systems in less than 10 years, with Google already having claimed quantum supremacy on multiple occasions.

As such, it is also a question of which might be achieved first given current trends. Poll will remain open until 2030, or when quantum computers start to make significant penetration into both households and industry, or if AI Superintelligence is achieved on classical computing hardware.

Update 1 - By "Classical" I mean any computer with Von-Neumann architecture or GPU's, which serves as the primary foundation of the vast majority of computer technology today. The "Other" category can be assumed to cover over any alternatives to that (DNA chips, hybrid systems, etc.)

Get Ṁ600 play money
Sort by:
bought Ṁ3 of Quantum Computer YES

Where's the option for BOTH / Hybrid System?

@TomPotter That would be in the “Other” category. Perhaps I should make it a separate line item?

bought Ṁ10 of Classical Computer YES

@Meta_C I don't know what other options there are besides quantum, classical, both, and neither (not built).

@TomPotter ..I guess there's a chance we figure out some entirely new method of computing, but depending on definitions, that could be nearly impossible... would the human brain be considered here to be a form of classical computer (does classical computation), for example?

I mean, I basically interpret "classical computer" to mean "non-quantum computer".

@TomPotter I’m defining “classical” computer as any computer with Von-Neumann architecture in this case. The “Other” category is intended to cover things like hybrids, DNA computers, neuromorphic computing, etc.

But, as I can’t change the options anymore, I’ll clarify in the description…

bought Ṁ10 of Other YES

@Meta_C Ok, interesting. That's a much larger set, then; that puts much more weight on Other then, for me.

It might still be good though to have "hybrid" be its own category, to differentiate from those other options, since a lot of Other otherwise would just be hybrid, and it's interesting to me to see the amount of probability for actual other architectures instead of just "classical with some quantum thrown in".

Is it possible to add new line-items, just not change them? Or is it simply no updates allowed?

@TomPotter Not anymore unfortunately, unless I delete and recreate this poll. The ability to do that has to be set before posting, and I wasn’t counting on having to change the options at the time.

But I can always update the description as necessary 🙂

@Meta_C Fine with me to remake! But we can work with what we have too..

There's also David's statement:

"Quantum computers" are not general purpose computers. There is not a 34% chance of that happening, or any chance of that happening.

Which is probably true if "Quantum Computer" means literally no work at all is done using classical computers.

So, we could simply have "Quantum Computer" allow some classical computation, and just mean that a significant portion of the intelligence computation is done using quantum methods. That said, the title then becomes maybe a bit false-binary in comparison to the description. (Is it editable too?)

There's also the question of degree. I'd be surprised if there wasn't at least a bit of quantum thrown in there for one or two algos that might benefit a lot from it, even if most of the computation is done classically.

@TomPotter I can still edit the title it seems, let’s see about making a title which efficiently encapsulates all that…

@TomPotter And done! There is a character limit to the title, so I’ve adapted it the best I could.

@TomPotter Also, there is research on the possibility of general purpose quantum computers, see here for example. https://spac.umd.edu/news/story/new-quantum-computer-module-sets-stage-for-generalpurpose-quantum-computers

@Meta_C It seems to me like there's still an issue associated with the degrees of quantum/classical. I don't know how to parse it out, because it seems like a somewhat smooth gradient from one to the other, as far as how much of the calculation is done on each. I.E. I would expect any quantum solution would have a good chunk of classical for "supporting systems," and it may become common for classical systems to offload certain operations to quantum subsystems.

Just saw your mention of general purpose quantum -- huh, interesting.

@TomPotter Indeed. I’ll have to give some more thought as to how to better parse the other options in an efficient way (since there are potentially a vast number of possible options or different gradients one could go down). For now I’ll just wait and see how others respond to this poll, and iterate from there…

"Quantum computers" are not general purpose computers. There is not a 34% chance of that happening, or any chance of that happening.

More related questions