Slashdot Mirror
AI Experts Say Some Advances Should Be Kept Secret (technologyreview.com)
AI could reboot industries and make the economy more productive; it's already infusing many of the products we use daily. But a new report [PDF] by more than 20 researchers from the Universities of Oxford and Cambridge, OpenAI, and the Electronic Frontier Foundation warns that the same technology creates new opportunities for criminals, political operatives, and oppressive governments -- so much so that some AI research may need to be kept secret. From a report: [...] The study is less sure of how to counter such threats. It recommends more research and debate on the risks of AI and suggests that AI researchers need a strong code of ethics. But it also says they should explore ways of restricting potentially dangerous information, in the way that research into other "dual use" technologies with weapons potential is sometimes controlled.
As if criminals, political operatives, and oppressive governments won't get hold of the required information regardless. Just publish everything and give the rest of us a fighting chance to figure out what's going on and defend ourselves.
The biggest danger is secrecy, not technology. We should never grant the state any advantage. If we don't fight back, we are doomed... DOOMED!
Isn't that what they said what about OSS? To be honest, aren't the bad guys just going to use last generation's AI to crack the current generation and then make it available on black market? Look at how long it tool to crack DVD and Bluray keys? It was meant to be the most advanced of it's time.
You say things that offend me and I can deal with it. Can you?
Yeah! Down with the antiquated notions of information seeking to be free, and let us all welcome the concept of security through obscurity.
Right! And let's reimpose limits on exporting strong encryption, while we are it.
In Soviet Washington the swamp drains you.
How do you prevent people, good or bad, from evolving the technology or science? They only fool-proof way of keeping something secret is to not find out from the beginning. This sounds like an effort to stop the wind from blowing. A lot of people seems to be afraid of AI, but I fear the stupidity of people more than I do the intelligence of machines. Maybe the real threat is that we seem to accelerate our own stupidity.
When the US was restricting export of public key cryptography, geeks used to print the equation on t-shirts. The only technology that's even been kind of successfully restricted is nuclear, and that's mostly worked by restricting physical equipment rather than knowledge.
How are you going to keep it secret? By outlaw the knowledge. If the knowledge is outlawed, only the outlaws will have the knowledge.
It reminds me how some people share their passwords or PIN codes (with spouses or kids or whomever). If you tell somebody your secret, it isn't a secret anymore. The standard when somebody wants to know my secret is:
"Do you know how to keep a secret?"
"Yes."
"So do I. (Silence after that till they get it. Staring at them helps.)"
This discussion is identical to closed versus open source. (Not similar, identical.)
Don't fight for your country, if your country does not fight for you.
Like the fact that there's no AI whatsoever. There are limited-purpose algorithms for very narrow tasks which work a lot like the calculator: i.e. they far surpass what the average human being is capable of (most people cannot compute in their heads), yet they cannot reason (which is why image recognition systems can be easily fooled), think (which is why proper translation is a pipe dream) or invent anything (which is why they cannot come up with new ideas). The hype about AI is so strong, people actually fear will be enslaved by robot overlords soon yet we are not close to general AI than we were 50 years ago, we just have much better hardware to use to train those algorithms.
Even image recognition AI which is touted as a breakthrough is largely incomplete because animals intelligence doesn't require petabytes of data and petaflops of compute power to recognize objects in all their embodiments maybe because we've deciphered only the outmost layer of the nervous system.
In short there's no AI to speak of. Up to this day we've just been automating the most primitive tasks which don't require intelligence per se. They require statistics, lots of data and lots of compute power.
Oh, and these algorithms are almost completely opaque, so we cannot understand them, properly tune them or even expect proper answers from them all the time.
So a research group thinks there should be more research into AI. But they think the topics to be researched should be kept secret.
I wonder if their AI driven grant application writer came up with that one.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
This seems pretty common sensy and it's also what we've determined to always be best in the computer world. So far.
But this approach seems flawed if we think about, say, nuclear weapons engineering. The more everyone (including you) knows about how to make a nuclear weapon detonate correctly, the more dangerous other people become but you don't really get to apply any of that to your defense. It's not like your bomb shelter will get better because of you finally figured out how to get the imploder timed right. It's not like your political efforts to limit nuclear proliferation benefit from proliferation of the engineering knowledge. It's not like your coping-with-horror-by-using-fatalistic-nihilism-and-humor will benefit from th-- wait, ok, so it does happen to help that one defense, but that's an unusual case.
For the most part, nuclear weapon engineering proliferation is bad for everyone, in a way that completely contrasts with, say, knowing that fingerd has an exploitable buffer overflow bug.
Are there some conditions where software tech crosses over into being more like nuclear weapons and less like other software tech? More to the point: what are the general conditions where tech knowledge proliferation is bad rather than good, such that buffer overflows get categorized one way and nukes the other? The condition isn't really "software good, hardware bad," no way.
That some people think some software tech is crossing over or soon may, makes me wonder WTF they figured out how to do!
(BTW, for some reason I actually like that they used movie plot threats in the guise of latching onto Black Mirror trendiness. Let's face it, everyone: movie plot threats are fun to think, about and I don't care what The Almighty Bruce says!)
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
This was written in 1853 by Charles Tomlinson, and is only an excerpt of the the treatise, but it shows that people recognized that 'security' trough obscurity was not really security at all, way before the digital age. A commercial, and in some respects a social, doubt has been started within the last year or two, whether or not it is right to discuss so openly the security or insecurity of locks. Many well-meaning persons suppose that the discussion respecting the means for baffling the supposed safety of locks offers a premium for dishonesty, by showing others how to be dishonest. This is a fallacy. Rogues are very keen in their profession, and already know much more than we can teach them respecting their several kinds of roguery. Rogues knew a good deal about lockpicking long before locksmiths discussed it among themselves, as they have lately done. If a lock -- let it have been made in whatever country, or by whatever maker -- is not so inviolable as it has hitherto been deemed to be, surely it is in the interest of honest persons to know this fact, because the dishonest are tolerably certain to be the first to apply the knowledge practically; and the spread of knowledge is necessary to give fair play to those who might suffer by ignorance. It cannot be too earnestly urged, that an acquintance with real facts will, in the end, be better for all parties. Some time ago, when the reading public was alarmed at being told how London milk is adulterated, timid persons deprecated the exposure, on the plea that it would give instructions in the art of adulterating milk; a vain fear -- milkmen knew all about it before, whether they practiced it or not; and the exposure only taught purchasers the necessity of a little scrutiny and caution, leaving them to obey this necessity or not, as they pleased. ...The unscrupulous have the command of much of this kind of knowledge without our aid; and there is moral and commercial justice in placing on their guard those who might possibly suffer therefrom. We employ these stray expressions concerning adulteration, debasement, roguery, and so forth, simply as a mode of illustrating a principle -- the advantage of publicity. In respect to lock-making, there can scarcely be such a thing as dishonesty of intention: the inventor produces a lock which he honestly thinks will posess such and such qualities; and he declares his belief to the world. If others differ from him in opinion concerning those qualities, it is open to them to say so; and the discussion, truthfully conducted, must lead to public advantage: the discussion stimulates curiosity, and curiosity stimulates invention. Nothing but a partial and limited view of the question could lead to the opinion that harm can result: if there be harm, it will be much more than counterbalanced by good.
Shutting down free speech with violence isn't fighting fascism. It IS fascism!
Sure FPGA based stuff might be faster, but, in the end, its either intelligent, or its not. And in 100% of cases I have examined, its not
There is, in fact, no evidence at all of actually "intelligence" in any of the stuff sold as AI at present.
There is good evidence for the ability to solve certain specific problems (Playing GO or Chess), and a lot of evidence that multi-threaded attacks on complex problems can do things that people find difficult.
That is NOT intelligence.
I am still waiting for a definition of intelligence that would be generally acceptable to most people, including those with the resources to implement it.
Sent from my ASR33 using ASCII
The problem is... that with a few small hints as to how the technique works, a scientist can intuit the rest. It's not like someone gave Kim Jong the details on how to create a nuke... he started with what was publicly available, and stole what he could, and worked out the rest (or his scientists did)... it's no different with AI. Keeping a certain method or technology is only useful if that tech can't be reverse engineered, intuited, etc... I personally think the cat's out of the bag on this one.
Which has more power: the hammer, or the anvil?
This is so eerie... I have guns, and a 4x4... AND I have access to sourceforge and git hub... I'm so conflicted right now!!!!!
Which has more power: the hammer, or the anvil?