'I get it if you don't make money for 2 or 3 years, but Amazon's 21' The EU's digital office has warned this week that Hungary's proposed tax on ISPs can't become a precedent in Europe. Ryan Heath, spokesperson for digital czar Steelie Neelie Kroes, told El Reg: This must be seen in a wider context as part of a pattern to limit media and internet freedom. That is why it is so important to …
This post has been deleted by its author
Yeah, it's always fun to read what people believe what AI can do.
It's the new "flying car". We only begin to understand what we lack to create a general AI. Even the pseudo AI like deep learning, hierarchical temporal memory are extremely limited and incomplete, if correct.
I doubt I will see GAS (general artificial stupidity) kill any neural network in my lifetime, sadly.
Friends of mine managed to trick me (and not just me) into temporarily believing their messaging robot was an actual person. A seemingly drugriddled and fragmented person with a terrible disposition and any number of disorders, but all the more human-like for all of its flaws. At first glance this output of this contraption resembled the ramblings of the average suburban teenager, which was why i paid less attention than i could have until the inevitable "gotcha" ultimately broke the spell.
We might be tempted to defer what we think of as menial tasks to some form of computer algorhythm or neural decision making structure (or a combination) but definitely should be aware of the risk involved with trusting any such contraption beyond the scope of its design - in other words, as far as we can throw it.
I spent many years doing research in AI. I know just how hard it is to make computers perceive, plan and act in the real world.
I do not believe the sci-fi vision of truly intelligent machines taking over will be a problem for generations - if ever. It is too hard to build those things.
But, the real problems will come from dumb computer systems, compounded by the fact that we tend to view them as being smarter than they really are!
Fast stock trading has produced mini-market crashes already. Several times, over the decades nuclear missiles have begun their launch countdowns through misinterpreting radar data, or radio interference from taxis. And how many demands have been sent out by computer billing systems for payments of zero? :D
How dependent have we made ourselves on technologies, like GPS, the Internet, electronics and electricity distribution systems, that may not survive the next Carrington event?
More generally, we are smart enough to build new technologies, but not smart enough to fully comprehend the consequences of deploying them.
Our natural stupidity may be what eradicates us in the end...
I hope you do really really well for seeing the technical, mathmatical and real problems and goals that are involved, and manage to surpass them. Too many are pushing "snake oil" (had a reply from a company when looking to buy one of their "AI toys" with "why would you want it to do that?" instead of actual acceptance of it's limits).
Until we see where the goal line is, we can never reach it. :)
The Real Problem ….. I spent many years doing research in AI. I know just how hard it is to make computers perceive, plan and act in the real world. …. hbarr
Quite so, hbarr. The novel solution to all of those problems is to have smarter beings perceive, plan and act in the real world as if virtual machines and advanced operating systemed computers. Or is that thought impossible for humans?
If it is so widely thought, then will humans have no end of problems, methinks, for it is realised by more than just a few to be the answer to practically every difficulty in need of an intelligent alternative direction, and when needs must, will there be all manner of strange disruptive and destructive shenanigans for indolent incumbent systems admins to waste time and effort and resources combatting during increasingly obvious collapse and conflicted resolution.
Well, hacking into systems to subvert their normal functions is a well known trade. Gangs of criminal hackers controlled by high authorities are also posited - mostly in non-allied countries.
Some people (me for example) might harbour dark thoughts that newspaper magnates, bankers, politicians etc act similarly on pulic opinion, markets, public policies etc but that is just daydreaming...
As far as AI goes, how do you make an open-minded computer?
Some people (me for example) might harbour dark thoughts that newspaper magnates, bankers, politicians etc act similarly on pulic opinion, markets, public policies etc but that is just daydreaming... ... Uffish
That aint no daydream, Uffish, for that is exactly what they try so badly to control and effect. It is though an art which is devilishly revealing of all poor performances.
@hbarr
One big issue, as some above expressed, is how do you define AI?
As humans, composed of meat, our intelligence is inescapably meat-based. (I am not talking ab out bacon as brain-food here.) What it means to be intelligent is therefore naturally tied up in the physical processes that give rise to it. How then do we even go about defining intelligence where such processes do not exist?
Most would agree that intelligence is not simply knowing that 2 + 2 = 4 but in some way understanding it. What does it mean to say that a computer understands something?
One would expect that there would be limits to such an intelligence, despite the fact that its computational power may significantly exceed our own. Could an AI actually think and theorise about the world? Take the theory of General Relativity, which by all accounts was a 'thought' of the most amazing order and a leap of quite remarkable genius and was not based on the input of some data that pointed to it.
Or what about any of the other theories initially proposed before there was any ability to verify or even test them? By that I mean those hypotheses that came about, in some measure simply by thinking about the problem at hand.
There is no way you can get to $150B market vaue without earning money.
They have just been clever at dodging taxes around the world and move money into deparments with operating losses (aka heavy investments). so 150B/21Y ≃ 7B/Y. Not bad at all.
As long as the stockholders do not require dividends they can still keep the stock price high.
All multinationals have lots of advisors working on how to avoid taxes and give as little as possible back to the community. IBM probably have more financal acrobats than engineers.
"There is no way you can get to $150B market value without earning money."
And the truth is, contrast Microsoft in the Ballmer era with Amazon.
Microsoft used a few cash cows to generate profit, with a few tweaks to keep 'new' versions selling, while Amazon constantly churns its cash into growth and new revenue sources.
Microsoft had Windows, Outlook, and Word, Amazon sold books.
Today Amazon sells damn near everything, and has a cloud service that the competition is racing to keep up with, and growing rapidly.
Microsoft still has Windows, Outlook, and Word, and losing ground rapidly.
True, they've started other projects recently, but all are 'Me too!' projects in a 'throw'em at the wall and see what sticks, we've got billions' attempts to chase taillights of everyone else making a profit.
I'm no fan of Amazon, but they're busy building company value, while Microsoft is busy chasing profits with an ever increasingly irrelevant portfolio.
"Really, how much else is there to see in the sky/stratosphere?"
Lots of stuff, actually. Clouds, carbon dioxide, CFCs, cow farts, car exhaust, climate change... oh wait, those all start with the letter "c." But really, lots of climate research focuses on levels of various gasses and particles in the atmosphere--oxygen, water, ozone, CO2, nitrogen and methane are always popular, but minor ones (CFCs, volcanic ash) can also have an outsize effect on the environment too. Most of these are currently studied by putting sensors on balloons and satellites; which, arguably, makes the stratosphere easier to study than the deep ocean.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
