Slashdot Mirror
Satya Nadella Explores How Humans and AI Can Work Together To Solve Society's Greatest Challenges (geekwire.com)
In an op-ed for Slate, Microsoft CEO Satya Nadella has shared his views on AI, and how humans could work together with this nascent technology to do great things. Nadella feels that humans and machines can work together to address society's greatest challenges, including diseases and poverty. But he admits that this will require "a bold and ambition approach that goes beyond anything that can be achieved through incremental improvements to current technology," he wrote. You can read the long essay here. GeekWire has summarized the principles and goals postulated by Nadella. From the article:AI must be designed to assist humanity.
AI must be transparent.
AI must maximize efficiencies without destroying the dignity of people.
AI must be designed for intelligent privacy.
AI needs algorithmic accountability so humans can undo unintended harm.
AI must guard against bias.
It's critical for humans to have empathy.
It's critical for humans to have education.
The need for human creativity won't change.
A human has to be ultimately accountable for the outcome of a computer-generated diagnosis or decision.
AI must be transparent.
AI must maximize efficiencies without destroying the dignity of people.
AI must be designed for intelligent privacy.
AI needs algorithmic accountability so humans can undo unintended harm.
AI must guard against bias.
It's critical for humans to have empathy.
It's critical for humans to have education.
The need for human creativity won't change.
A human has to be ultimately accountable for the outcome of a computer-generated diagnosis or decision.
Never mind all the 'AI' bullshit, mister, how about you concentrate on not annexing every damned computer on the planet into your fucking Windows 10 spyware bot-net instead?
Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
Artificial intelligence, like genuine intelligence, is complex. Because it's complex, it can't be transparent.
Contribute to civilization: ari.aynrand.org/donate
...how humans could work together with this nascent technology to do great things....
Stop Windows Update from performing an unwanted update to Windows 10 for my PCs.
.
If it can handle that task, it can take on any challenge.
Twenty years of Internet advertising an eBay still haven't figured out that if I've just bought a widget, I don't want another of the same widget.
But you do! eBay knows that you bought crap and it needs replacing already.
And it's fun and weird to see this coming from Microsoft as well considering their behavior when it comes to Windows 10.
If builders built buildings the way programmers wrote programs, then the first woodpecker would destroy civilization.
In fairness, his essay doesn't say this is what will happen. In his own words, he has reflected on what are the principles and goals, as an industry and a society, that we should discuss and debate.
But as you have implied, this type of thoughtful discussion on how technology should be used for the greater good of society is not how it works. Those with the most resources will develop the most advanced technologies, and those technologies will primarily benefit the creators. If you want to guess how artificial intelligence will affect society, determine how it can be used to make people money. That is how you predict how it will be used. Basing predictions on how it benefits society the most is childish dreaming.
-- All that is necessary for the triumph of evil is that good men do nothing. -- Edmund Burke
-AI must be designed to assist humanity.
I'm sure he thinks reporting everything I do to the NSA will help humanity. This is just the zeroth law warmed over and when the rubber hits the road it becomes utterly meaningless. Whoever owns the AI decides what will help humanity. Iran thinks making nukes will help humanity. The US thinks killing durkadurkas will help humanity. Japan thinks imposing strict social order will help humanity. Google thinks Google having all the world's information will help humanity.
-AI must be transparent.
To its maker? It already is, unless you're talking about neural nets with unpredictable output. In which case the problem of determinism has been rehashed a million times and is involved in the discussion of every advanced AI ever made. To everyone else? HA! Ask some of those day trading AI companies for their source code. Let me know how that works for you.
-AI must maximize efficiencies without destroying the dignity of people.
Also meaningless, and for the same reason. Some people thing America's homeless living in the gutter have dignity if they're so much as given food stamps. Some people think my dignity is preserved even if a company is raping my files for every bit of information they can get. Efficiency is as debatable.
-AI must be designed for intelligent privacy.
Finally something I agree with. Too bad Microsoft doesn't. But each person building an AI will decide what privacy means. So how can it be enforced? With legislation?
-AI needs algorithmic accountability so humans can undo unintended harm.
This is just a rewording of the problem of determinism.
-AI must guard against bias.
How will you enforce it?
-It's critical for humans to have empathy.
What does that have to do with AI?
-It's critical for humans to have education.
What does that have to do with AI?
-The need for human creativity won't change.
The people writing AI's that will make tomorrow's music, novels, and salesmen disagree.
-A human has to be ultimately accountable for the outcome of a computer-generated diagnosis or decision.
And we're back to nebulous definitions. There are guns on the Korean DMZ that shoot anything human near them. Who is responsible for those bullets? If an escaping family gets mowed down I can promise no politician will accept responsibility. As soon as it becomes politically inconvenient, just as with human decisions, AI decisions will become the responsibility of "policy" or "the board" or "a bug".
I refuse to believe Satya Nadella is an idiot. And that means the whole speech is politically aimed attempts at redefining words, double talk, and pandering.
If video games influenced behavior the Pac Man generation would be eating pills and running away from their problems.
Replace the word "AI" with "Government" and I'm in:
Governments must be designed to assist humanity.
Governments must be transparent.
Governments must maximize efficiencies without destroying the dignity of people.
Governments must be designed for intelligent privacy.
Governments needs algorithmic accountability so humans can undo unintended harm.
Governments must guard against bias.
It's critical for humans to have empathy.
It's critical for humans to have education.
The need for human creativity won't change.
A human has to be ultimately accountable for the outcome of a government-generated diagnosis or decision.
It's funny to hear about how dependable AI will be coming from Microsoft, a company whose software has hundreds of megabytes of patches per month, whose software is responsible for millions and probably billions of dollars worth of financial losses to businesses and consumers every year.
Once Microsoft unleashes its AI upon the world, it will no doubt cause the entire planet to be reduced to green goo.
Good point. If we consider Windows 10 to be a kind of "robot," we can consider how it does in relation to Asimov's Three Laws of Robotics in the recent case where my elderly mother accidentally approved its installation as an "upgrade" of her Windows 7 system, which culminated in device-driver incompatibility warnings which she interpreted as making the computer unusable. (Elderly folks and non-techies get confused by things like that.) To wit:
"1) A robot may not injure a human being or, through inaction, allow a human being to come to harm." Its installation approval process did not adequately protect against accidental approval by the elderly human, thereby causing her to come to harm.
"2) A robot must obey the orders given it by human beings except where such orders would conflict with the First Law." Although the installation technically "obeyed orders given it by human beings," in this case by proceeding with an installation that she accidentally approved, that caused her harm per the First Law.
"3) A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws." Windows 10 would get high marks on this one in the "protect its own existence" category due to the fact that it can't be uninstalled once it's installed, except for having violated the First and Second Laws along the way.
Overall conclusion: "Bad robot!" (whacks Windows 10 with a newspaper.)
But he admits that this will require "a bold and ambition approach that goes beyond anything that can be achieved through incremental improvements to current technology," he wrote.
But the article says:
Doing so, however, requires a bold and ambitious approach
It's interesting that you needed to change ambitious to ambition.
Why?
We play the game with the bravery of being out of range
no undo harm (that leaves out alleged MS software)
What are you talking about?
Ctrl-Z saves me every time!
We play the game with the bravery of being out of range
As human civilization gets increasingly complex and reliant on computers to manage and maintain the things that allow us to exist in this First World, there will come a time when human civilization will have to us AI just to "maintain course"
We play the game with the bravery of being out of range
... I would like to see a machine genuinely cry at my wake because it knows it will miss our co-op game playing.
Ummm... you'll be DEAD dude! You ain't gonna see shit at your own wake!
'The Economy' is a giant Ponzi scheme whose most pitiable suckers are the youngest among us and the yet-unborn.
... Big businesses, Government? If we are not willingly giving control of our lives to those entities, why would we do so to an AI created by one?
Look around you. Facebook. Big Pharma. Propagandistic TV shows. Planned obsolescence. Journalists in the pockets of those they are 'investigating'. A critical mass of our fellow citizens has already drunk the Kool-Aid and signed on for substantial control over their existences. What makes you think they'll kick up a fuss over AI controlling their lives, so long as said AI keeps them comfortably numb and maintains the supply of bread and circuses?
'The Economy' is a giant Ponzi scheme whose most pitiable suckers are the youngest among us and the yet-unborn.
He isn't talking about strong AI, which not many AI researchers are actually working on and where no significant progress has been made in the past decades. The AI being discussed is narrow AI (or weak AI), where there are real world applications right now which could be very disruptive to our society.
It is fairly safe to assume any discussion of AI does not mean strong AI unless specifically stated.
-- All that is necessary for the triumph of evil is that good men do nothing. -- Edmund Burke
I see no need to reconcile it. Either a system is low-dimensional and not intelligent, or it's a generalized system with intelligence and can do pretty much anything.
If such a system were to gain a fully generalized thinking capability, it would not be LDNLS, because it would not be not low-dimensional.
I'm not saying that it could, just describing the bright line between the two ideas.
There's one other thing; it may be that there are non-neural routes to a general conscious intelligence. We know that nature has solved the neural intelligence problem at least two ways: the way we do it, and the way smart birds do it (higher neural densities, different brain structures.) The implication, as I take it, is that there's definitely more than one solution possible, and I (hand-wavingly) am willing to generalize that to "perhaps there's a non-neural way to host an intelligence." Algorithmic solutions are the most obvious suspect at this point, but they've been unable to get there so far.
I reject out of hand the idea that something that is not conscious is intelligent. That strikes me as purest marketing hype. The thermostat is not intelligent. No matter what the marketing claims. :)
I've fallen off your lawn, and I can't get up.
Umm... no. It would be in everybody's interest that these terms get defined precisely. We could start using acronyms [something that the corporate world is very fond of], so it would be WAI or SAI for you.
Sometimes the English language doesn't adhere to what is in everybody's interest. My description is how AI is defined and used both colloquially and in research. Do a quick Amazon search of the most popular AI textbooks (written by many of the lead researchers in the field) and see if they only discuss conscious machines or if they use a more broad definition of AI. That is far more definitive than two people arguing in an Internet forum.
-- All that is necessary for the triumph of evil is that good men do nothing. -- Edmund Burke