US military chucks $2bn at AI, Google touts machine-learning data search, and more Here's a quick roundup of AI announcements this week beyond what we've already covered. US Defense research unit funding billions for AI Next campaign: DARPA is dumping a whopping $2 billion to kickstart the “third wave” of AI focused on making systems more adaptable. It’s all part of the US military's ‘AI Next’ campaign, …
If it is "ultimately impossible" to train a computer for every eventuality, then it is also ultimately impossible to test whether an AI equipped computer will respond correctly to any given scenario.
Given that AI systems will eventually get a degree of autonomy, we can rightfully be a little fearful of the consequences.
A standard system, will hold up its hands when faced with a new scenario developers hadn't trained it for, whereas the AI system will just plow on, making decisions about a scenario that developers hadn't foreseen.
I agree, mostly, and the fear is that "AI" will be used past its ability, to be sure, in ways that can be lethal. I was doing NNs and ML in the previous craze in the '90s or thereabouts and am well aware of the limitations and how little actual progress there has been - just more data, more cycles...And doing some things (like using more layers with reduced precision as a substitute for proper design and data prep) we knew were wrong back then - and they're still wrong. It just takes less skill to produce "something I can show the boss/investors" to do it the wrong way.
A properly designed network/classifier, though, even back then, could produce as a byproduct or even as a parallel process, a confidence in whatever it was deciding - and even just produce a list of the top few most likely for another entity (process, or human) to handle...
So your comment about "throw up it's hands" at least used to be true for ML systems too - if the designers weren't hiding the flaws to keep those dead presidents coming. They frequently said, or could easily be made to say "I dunno, you take a look".
Those tests worked really well.
Show the machine vision lots of pictures of Russian tanks and lots of pictures of US tanks.
It identified both sets almost perfectly even in partial images at different orientations and in all tests in the USA. Then they tried it in exercises in europe, in bad weather.
It turns out the "AI" (ie simply 80s era NN) had learnt that shiny well lit pictures of tanks on sunny parade grounds were American, grainy long range covert images of tanks in mud were Russian.
‘I speak as one part of his "useless class" who can't even beat a normal chess program on my crappy computer.’
Can this ‘normal chess program’ write a program that can beat it in chess, I know I can. The reason the chess-program wins is it substututes speed over intelligence. If you could slow down or stop the clock between moves then you could most definitely win.
Apropos artificial intelligence, this whole AI thing is just so much hyped snakeoil, we're decades away from any real-world practical solution. In this case I guess it's a good a pretext as any to fling $2bn at the US military. A technological solution to a human problem, the problem being how to the ability to kill more of the enemy than he can kill yours. The snakeoil being the promise of fighting a war without taking human loses. Stalingrad and the Vietnam War would have demonstrated the falsehood of thinking technology could win against a tenacious and dedicated low-tech enemy.
It is not really a secret that they keep fabricating incidents just to go to war. The USA is in constant war. There has to be an enemy – always.
You have to ask yourself what it is all for? The answer to this is rather simple if you take look at the numbers. It's for keeping a few people having everything, while the rest of the world feasts on the scraps – just like in the Middle Ages and before that...
You have to explain why your main freinds in the Arab world are funding ISIS while your main enemies in the cold war are fighting it.
Why the terrorist attacks that justified you attacking your former allies were trained and equipped by your own side to fight the previous enemy
Why you are pro-democracy but have overthrown or declared war on every democracy in the region while supporting every dictator
That sounds like one hell of a task for IBM's Watson
"Yuval Noah Harari": There is nothing inevitable about democracy. For all the success that democracies have had over the past century or more, they are blips in history.
Well, we are already at the point where democracy has degenerated to the point where bullshit is served up cold (including "Anonymous Op-Eds" in the NYT pining for Bush III ... the mind boggles) and the public has to eat it and vote for it, or else. And the people in power will import a new populace and shout from all outlets that it's a good idea if that is what it takes. Worthy White US Politicians start to notice that they are being killed off by Very Left Hispanics in the polls as the make-up of the population shifts to something more like Venezuela. I cackle.
Talking of Total Control through Technology, after weeks and weeks of USUK and France slavering for an attack on Syria and relentless telegraphing that a false flag chemical weapons attack is coming, Youtube has just killed off Syrian news channels pour cause de "Violations of the Terms of Service" ("Net Neutrality" is no match for the Neocon Imperative of Total Social Justice). I have lived long enough to see 1984.
“Banks such as China Union Pay, AI Bank and the Agricultural Bank of China are using MLK-DNN Baidu’s Cloud for fraud detection. iQiyi, commonly referred to China’s answer to Netflix, will use OpenVINO to flag up videos for dodgy content.”
Does this mean that a company from the freedom loving west is teaming up with a communist dictatorship to spy on its own citicizens, for money?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
