We sent a Reg vulture to RSA to learn about the future of AI and security. And it's no use. It's bots all the way down AI algorithms will in the future form and direct swarms of physical and virtual bots that will live among us... according to this chap speaking at 2019's RSA conference in San Francisco on Tuesday. Thomas Caldwell, founder and president of the League of AI, a consultancy focused on security and robots, talked about the rapid …
The agents would also, presumably, learn what's normal on the network
this is where it falls down, like "a young person who had too much drink" (BOFH, paraphrased, and thus, the icon). I do have some colleagues pushing in that direction, and then when I really sit them down to ask what is normal in our data and what not then all of a sudden it no longer seems like such a good idea. The problem with our data is that some stuff that looks out of place is actually correct (there are things that take bloody ages because they are complex, not because some component failed), and some hints of problems are too subtle to flag them as errors. So any ML system will, with a bad training set like this, never be able to do what we want. What it will be able to identify is what we can see with more oldish fashioned stats as well, and we can actually see and understand why something gets flagged.
Drones everywhere, so police can cite you for even minor crimes like jaywalking across a quiet residential street. And no doubt the "gun recognition" ability will tilted in favor of never having a false negative - which implies a lot of false positives and cops absolved from shooting innocent people because "the drone told me he had a gun!"
FAO ..Thomas Caldwell, founder and president of the League of AI ........ AIdVenturing with CHAOS
A few missing links for you there, TC.
And with particular and peculiar regard to Your imagination is the limit, really. Just use your imagination. Because it's, right now, admittedly, mostly imagination.
Actually, right now, although not immediately recognised by all, is everything Applied Imagination, Katyanna/El Regers ‽ .
I mean, if even 1% of all this AI BS was true, what on earth would convince anyone it was advisable? .... Sir Runcible Spoon
How very strange, Sir RS, that anyone would think AI would care to convince or advise anyone about what it be doing and what has been done in concert with IT and Grand GeoPolitical Programmers/Greater IntelAIgent Gamers.
You might like to consider it has been considered and deemed to be too misunderstandable and complicated for all but a chosen few to believe and accept and further aid as the current honest reality for Future Forking Mass Multi Media Program Presentation.
The secret to both relieve and dispel unfounded anxieties is to follow and even question the views presented in order to witness and experience the facts even as one may be programmed to reject them as fictions.
Such simple deeply unquestioning rejections and reclassifications are an ancient perverted and well enough tested petrified stagnant status quo response and idiotic knee jerk reaction to that which assails and destroys the imbalance which fake news and false witness acquire and display.
I trust that make perfect sense to all wherever they be, just passing through here. It is simple common knowledge to a right choice few who may number considerably more than ever before thought possible.
Are there many in the full flush of arrogant hubris and ignorant discourse realising that extremely dangerous to them and corrupted status quo operations/perverted SCADASystems?
So it's normal ML, which is repeating the past. Which is fine for certain situations, but not so hot for many of the fuzzy ones.
"They could be decked out with cameras that identify suspicious items including guns and knives, or include microphones to listen out for gunshots and tell-tale screams."
Don't we have that already with CCTV? Don't we already have the ability to use AI on a central node (rather than an edge device) on these things? So why isn't it being used? Oh wait, it's a lot bloody harder to use sensor data in an uncontrolled environment. Or is the (oft made) claim that all we need are more and better sensors, and better data, so we just need improvements in CCTV :)
"or work alongside police officers."
Er no. Sorry. If there is one job that's not going to be replaced with AI it's beat cops. Many de-escalation techniques involve building a rapport and identification with the suspect, and I'm not sure that a robot is going to have the emotional intelligence or creative thinking to deal with this*. Having non-LEOs working with the cops to deal with situations can also work well, but I'd take a dog over a robot most days.
"Researchers, we were told, are interested in creating roving network-inspecting bots that can study past cyber-attacks, identify patterns in the intruders' data accesses and methods, and use that knowledge to detect future network compromises. "
OK, I'm confused. Is this a fancy network log analyser? Or is it going to have the ability to act on it's suspicions? Because having a tool to flag stuff up to a meatbag is nice, but it's not bloody AI.
* or most members of the public for that matter.
This guy apparently tries to use the arguments about standard, general AI to support the specialist tool of Swarms. Sounds ridiculous (and somewhat desperate).
We have two completely different technologies here, the "AI" one, and the "Swarm Intelligence", which is a subset of "AI". I'm pretty sure there are valid uses for Swarms, but they are specialist ones, mostly in the "Search and..." domains ("search and destroy" being the most obvious).
If you think about it, school surveillance ("Won't somebody think of the children"...) or malware control do not need Swarms, one could even say they'd be better off without it. I'm defining "Swarms" as a group of independent agents with a deeply collaborating intelligence, and fail to see any operational advantage of having that in my computer compared to a classical set of security tools (using AI or not).
As about schools covering under clouds of AI-controlled killer drones ready to pounce on anybody holding something looking like a gun (or a photo of a gun, a banana, a turtle, whatever else), I think that even in trigger-happy USA this would raise some concerns (At the latest when half the students have been accidentally massacred by the swarm and the rest failed all grades due to the constant, ear-splitting buzzing of the killer swarm...).
Your article pic put me in mind of Peake's Titus Alone, published in 1959. The era of the Cold War thriller, and spies from Le Carré to Bond.
That's not so far-fetched. I was born not so very much later than that, and grew up with all that literature[1]. Consequently I never thought I could rely on privacy: a hidden microphone or camera could be anywhere.
But I guess a bit of paranoia must be normal. To other generations, it might have been God or other such supernatural beings watching them.
[1] OK, I never read Fleming, and I didn't come to Le Carré until later. But moving from Biggles to the likes of Alistair MacLean and Desmond Bagley was plenty of action thriller to give a young lad the idea that someone could be watching. And then there was Orwell making it explicit and being taught at school ...
...like most similar pie-in-the-sky predictions, this one doesn't factor in that setting up the infrastructure necessary will be very expensive, and once the numbers start getting crunched and people start asking who'll foot the bill for it, the novelty wears off pretty quick.
A municipality that can barely afford to fix potholes on its main roads certainly won't be able to invest in an armada of drones to replace or augment the local police force. Corporations aren't going to just donate an investment that large to Smalltown out of the goodness of their hearts. Federal grants would no doubt just be whizzed away on unrelated projects. So where would the money for such a project come from?
> private security for very rich people
Very rich people value their tranquility, and that precludes a swarm of droning drones hovering over their property.
Besides they can afford to employ smartly dressed, well behaved, and most of all discrete humans. Swarms of killer drones are only fit for mad scientists and evil overlords...
Humans can do things dogs can't, like drive a car, answer the phone/door, go fetch your suit from the dry cleaning, and all that. Also they are far more discrete than dogs, in that they don't communicate through annoying barking.
As for double-crossing you, I'm not sure dogs are that perfect. While a little less prone to sleep with your spouse, they are much easier to bypass than humans, who usually don't fall for the poisoned steak trick.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
