Slashdot Mirror
DARPA Wants To Build an AI To Find the Patterns Hidden in Global Chaos (techcrunch.com)
A new program at DARPA is aimed at creating a machine learning system that can sift through the innumerable events and pieces of media generated every day and identify any threads of connection or narrative in them. It's called KAIROS: Knowledge-directed Artificial Intelligence Reasoning Over Schemas. From a report: "Schema" in this case has a very specific meaning. It's the idea of a basic process humans use to understand the world around them by creating little stories of interlinked events. For instance when you buy something at a store, you know that you generally walk into the store, select an item, bring it to the cashier, who scans it, then you pay in some way, and then leave the store. This "buying something" process is a schema we all recognize, and could of course have schemas within it (selecting a product; payment process) or be part of another schema (gift giving; home cooking).
Although these are easily imagined inside our heads, they're surprisingly difficult to define formally in such a way that a computer system would be able to understand. They're familiar to us from long use and understanding, but they're not immediately obvious or rule-bound, like how an apple will fall downwards from a tree at a constant acceleration. And the more data there are, the more difficult it is to define. Buying something is comparatively simple, but how do you create a schema for recognizing a cold war, or a bear market? That's what DARPA wants to look into.
Although these are easily imagined inside our heads, they're surprisingly difficult to define formally in such a way that a computer system would be able to understand. They're familiar to us from long use and understanding, but they're not immediately obvious or rule-bound, like how an apple will fall downwards from a tree at a constant acceleration. And the more data there are, the more difficult it is to define. Buying something is comparatively simple, but how do you create a schema for recognizing a cold war, or a bear market? That's what DARPA wants to look into.
>> a machine learning system that can sift through the innumerable events and pieces of media generated every day and identify any threads of connection or narrative in them.
This sound like a marketing question. As in, "how well are the talking points from various agencies and political groups represented in the media." There are communications firms that perform this type of analysis today on the messages they try to get out into the public (e.g., "this statistic we created - that's just a little bit off the official one so we can track it - has been republished in 228 news stories in the past 6 months").
Not freakonomics but psychohistory.
So watching Fox News (or MSNBC, etc) is going to answer this perennial college sophomore question? Couldn't we just watch Three Days of The Condor instead?
Maybe if you are an AI trying to amass enough resources to build yourself a physical body, then download your mind into it and start a new world order using your superior intellect.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
Mr KAIROS doesn't have the same ring to it.
My ism, it's full of beliefs.
Just ask Harold from Person of Interest, he'll do it for just one dollar but there is a catch. ;)
Anons need not reply. Questions end with a question mark.
Fredrick Brooks was absolutely correct. There is no silver bullet in programming.
I think the problem described above "buy something" has an analog in the Object Oriented programming mind set, where the process of buying something can be defined in more and more detail... So you abstract "Buy something".... "In a store" or "online"... "Using a credit card"..... Just like we abstracted "Vehicle" which is "A Car" has "an engine" and the like.
Brooks was right, programming takes effort and AI isn't the answer.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
Patterns, sure. But is it worthwhile to know them?
If you already have the data, then processing it is not expensive. If you are already pre-processing it for other reasons, then it is even cheaper.
Using a security camera to detect someone buying groceries may not be useful, but using an automated drone camera to detect someone placing an IED is obviously valuable.
They're either buying into the same marketing and media hype for the half-assed excuse for AI everyone keeps trotting out, or they've got something nobody else has, meaning general AI. The latter is highly unlikely, if they did we wouldn't be hearing about it at all.
If they had general AI, they wouldn't need to build a super computer powered expert system just to tell whether someone is buying something. This is a complete waste of time. You have a better chance of reaching the moon by building longer and longer ladders than you do reaching general intelligence by hardcoding facts.
I guess if anyone has the budget for something like this, it's DARPA, but my mind would be blown if this were anywhere near useful in less than 50-70 years. Predicting markets alone would be an incredibly daunting task to try and handle with AI, let alone something as generic as "schemas".
damn you beat me to it
It is neither.
This is not exactly a new, they put out these proposal requests all the time. They select some problem of interest, or more often a whole bag of problems and post a request for proposals, then see what various groups think they can accomplish along those lines. The actual research is a lot less dramatic than pieces like this suggest, and really just represent DARPA giving seed grants based off some theme to a bunch of teams and seeing what they come up with.
They're either buying into the same marketing and media hype for the half-assed excuse for AI everyone keeps trotting out, or they've got something nobody else has, meaning general AI. The latter is highly unlikely, if they did we wouldn't be hearing about it at all.
Or the option you didn't mention -- they know that existing ML-based systems are really good at pattern-matching, and that many forms of pattern-matching don't require full general AI in order to be cost effective, and they're applying it (like the headline and article says) to another domain for pattern-matching.
They are. Think of realtime transcripts of all US talk radio. The voice prints of all US cell phone calls getting keyword detection.
Then have the computing power to not just use keywords but seek patterns in they way people are responding to news.
Thats old tech and its days too late in the news cycle. Days after talk radio has made a local story national news again.
How is the US going to be swayed by NATO/UK/US gov propaganda again?
Ready for another Syria/Libya/Vietnam/Iraq war?
Can local news in any one part of the USA online be prevented from spreading all over the USA in a news cycle?
Say the police and national US media are sure one "person" did something. Reports about political motives are made by top national experts and trusted reporters.
Then it reverts back to common local gang crime? National new reporters and their top experts are wrong again.
Can that national reporting be walked back and removed from the internet in 24 hours? Be removed from common realtime archives and video sites?
The federal ability to hide any emerging news story around the USA before reporters are reported on as not able to follow a simple local news story.
The DARPA magic is to make news never be wrong again and to hide any local news that could show the real events.
To keep local news from spreading all over the USA. To ensure US national news reports/experts never have to go back and correct what was reported on in a city/state.
The chaos to the US gov/mil is the spreading of local news nationally. The way national news attempts to put is expert political spin on any issues.
Then reality emerges due to the "internet". With its easy to understand memes and images, archived reports and timelines.
DARPA wants to reset the way patterns of local/national news get spread in realtime. To finally tame the internet and ensure NATO/UK/US nation wide propaganda works as it did in the 1950s - 1990s.
Bring the internet to heel so the internet can be about the next going to war again narrative.
The deep state wants its war news cycle back.
Domestic spying is now "Benign Information Gathering"
I was thinking more 'Pi'...
"Listen to me. The Ancient
Japanese considered the Go
board to be a microcosm of
the universe. Although when
it is empty it appears to be
simple and ordered, in fact,
the possibilities of game play
are endless. They say that no
two Go games have ever been
alike. Just like snowflakes.
So, the Go board actually
represents an extremely complex
and chaotic universe. That is
the truth of our world, Max.
It can't be easily summed up
with math. There is no simple
pattern."
"I bless every day that I continue to live, for every day is pure profit."
Roger C. Schank, Robert P. Abelson et al did this decades ago. What DARPA calls a "schema" was called a "script' by Schank. His most-referenced book was "Scripts, Plans, Goals, and Understanding: An Inquiry Into Human Knowledge Structures "
I am quite astonished to find the DARPA reference b/c I know that there are literally hundreds of DOD-related systems that use ideas from Schank and others to gather and correlate intelligence and make suggestions for action. These are not new systems, most have been in the field for at least a decade. So it's as if DARPA has been hiding use of AI for decades and is now beginning to release it for commercial use.
Dear lords....it would be much easier for an AI to gain the trust of a sub intelligent human who is totally loyal and work through him/her.
I reserve the write to mangle english.
I regularly visit your site and find a lot of interesting information. Not only good posts but also great comments. Thank you and look forward to your page growing stronger spanish to english
You have a better chance of reaching the moon by building longer and longer ladders than you do reaching general intelligence by hardcoding facts.
I know; you're preaching to the choir. The entire approach is wrong but the AI fanbois insist that it's like The Moon Is A Harsh Mistress and adding more hardware will magically make it 'wake up' and become sentient. The marketing and media hype-machine has done it's job too well on too many people.
What are you talking about?
AlphaZero represents a colossal silver bullet over the tradition of hand-crafted alpha-beta implementations.
Ken Thompson actually said of his hardware implementation of Belle that the only reason it won is because their software had fewer bugs. (Bugs in a minimax algorithm are often subtle and hard to notice, so long as it always returns a legal move, and almost always returns something better than a blithering blunder.)
Getting it to work with the specialized hardware forces the programmer to clear his or her mind to a higher level.
(Gimli: It's true you don't see many dwarf women. And in fact, they are so alike in voice and appearance, that they are often mistaken for dwarf men. Aragorn: It's the beards.)
Few, if any, of this category of historical bugs exists in AlphaZero. It's pretty easy to keep your eye on the accuracy of a high-speed matrix multiply (extremely potent reductive invariant simply begs to be checked). All of the engineering is focused on backbone algorithms. All of the fussy, detailed, error-prone heuristics are manna from heaven.
This is absolutely a silver bullet, by any reasonable standard. Unfortunately, it requires a C4 explosive GHz GPU instead of black powder MHz CPU, so this particular bullet was not available to computational antiquity.
Sheesh: a silver bullet plummets down from the heavens, and nearly strikes us dead in the nads, and we're so deeply invested in Stockholm syndrome (a sixty-year captivity of brittle, hand-coded heuristics) that don't even notice its inherent silverbulletness.
"Waaa! But I wanted a silver bullet that better helped me to continue to do it the wrong way!"
If you're that addicted to doing it the wrong way, admit it: you're not in the market for a sliver bullet, you're in the market for a magic mushroom.
OMG it spells Bilderberg!
Remember kids, if you're not paying for the service, YOU ARE THE PRODUCT THAT IS BEING SOLD.
the machine.
On a long enough timeline, the survival rate for everyone drops to zero.