Slashdot Mirror
UK PM Seeks 'Safe and Ethical' Artificial Intelligence (bbc.com)
The prime minister of UK says she wants the country to lead the world in deciding how artificial intelligence can be deployed in a safe and ethical manner. From a report: In a speech at the World Economic Forum in Davos, Theresa May said a new advisory body, previously announced in the Autumn Budget, will co-ordinate efforts with other countries. In addition, she confirmed that the UK would join the Davos forum's own council on artificial intelligence. [...] The prime minister based the UK's claim to leadership in part on the health of its start-up economy, quoting a figure that a new AI-related company has been created in the country every week for the last three years. In addition, she said the UK is recognised as first in the world for its preparedness to "bring artificial intelligence into government."
Where the city board threw in so many ethical directives into the programming that they were contradictory.
Theresa May is a never ending source of meaningless, "X and Y" catchy cliches, none of them achievable by her aimless and malign government. I think she's has a quota of distracting bullshit to deliver each week to keep her party happy and the country distracted from the incompetence.
Leader of country makes speech to position it at the forefront of technology growth industry.
This is hardly news.
She's also sticking with the 'A and B' branding. Strong and Stable didn't work out, lets see how Safe and Ethical pans out.
I will agree with her that the UK is first to "bring artificial intelligence into government". Their current administrations intelligence, like the plants in my office, is definitely artificial. Meanwhile the CEO of google just made the most convincing argumment against AI in history:
I kid, I kid. But seriously folks, when your ruling class is consistently making the same vacuous 'everything's fine, really' comments you should be very, very worried.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
That's how long it will take some motivated person to hack the safe mode out of it.
It's going to be abused. All technology is. Spend the money on developing plans to deal with it.
These conversations always bring me back to the DVD encryption attempts.
Spend millions on developing unbreakable encryption that gets broken in a few weeks and for free.
If suddenly millions of people have no money for food because their job was replaced with AIs then is it an unethical use of AI? The problem with this very real situation is that it's the politicians behaved irresponsibly by not creating the required social safety nets that these people will need.
You know what? Fuck you, guys. I for one welcome our new AI overlords! ;)
Anons need not reply. Questions end with a question mark.
Nobel prize for the person who can figure out how to implement these. It will probably be won by an AI.
Current narrow AI can't really use the Three Laws at all; they're too general.
For Artificial General Intelligence, which could use them, they are insufficient. Perhaps the best illustration of this is Asimov's robot stories themselves, which are all about how the Laws break down.
What we really need is a proper theory of Friendliness.
You have at least one human being overseeing the so-called 'AI' at all times, because the half-assed excuse for 'AI' they keep cranking out can't actually think, is not sentient, and furthermore even the programmers that create it can't tell you what's going on inside it when it's running. You can't talk to it, you can't reason with it, you can't ask it to elaborate on what it's output is, therefore you can't trust it's output; you have to have human beings monitoring and auditing it at all times unless you want something disasterous to happen when it does something totally out of left field.
Sorry, you've confused Asimov's 3-laws robots with Jack Williamson's humanoids "To serve, and protect, and guard men from harm". Williamson's humanoids so distressed John W. Campbell, Jr. that he coerced Williamson into writing a sequel where humans successfully emerge from the cage, but he had to evoke magic (essentially) to get it to work.
Asimov's laws *could* have lead to the situation that you depict, but they never did (in the books, anyway). When a robot got competent enough to possibly take over, it understood "psychological damage", so it didn't...except under complete cover...including leaving a believable corpse when he "died". (I think this was only in one story.) R. Daneel never took over, even though he *could* have under the 0th law. It was too dangerous for him, because he'd need to be making decisions that would cause *some* people harm.
But neither the definition of "safe" nor the definition of "ethical" that you offer are workable. And current attempts at AI are based around neural nets, so you train them off patterns in the data, not off of definitions. If the data reinforces acting in a way that you feel is safe and ethical, then the robot will act that way. The problems come when the situation is outside the area covered by the training data, and the robot generalizes in a way that you didn't expect. Read "The Two Faces of Tomorrow" by James P. Hogan. He doesn't get everything right, but he gets this part right.
I think we've pushed this "anyone can grow up to be president" thing too far.
Sorry, but China will be just as insistent that their AI's act ethically as anyone else. They may have a different idea as to what ethically means, but they'll have *some* idea. The only one that wouldn't is someone who's suicidal in the short term.
Actually, the only one who wouldn't insist on their AI being ethical is someone who either doesn't understand the problem, or just likes to waste money. An AI with no ethics wouldn't do anything on purpose. And you couldn't coerce it. It would be useless, even to itself. (And it would probably fail long before it reached sentience, but that's a separate question.)
I think we've pushed this "anyone can grow up to be president" thing too far.
I think based on the long human history ethical will be a robot that shoots enemy soldiers and does not shoot our soldiers.
Anyone who has read Asimov's books.