Without new anti-robot laws, humanity is doomed, MPs told Robots will destroy humanity unless we write new laws to control them, a UK Parliamentary committee has been told. “The key question is: if something goes wrong, who is responsible?” pondered the Commons Select Committee for Science and Technology, in a report released today. Microsoft's Dave Coplin, the firm's “chief …
"Robots will destroy humanity unless we write new laws to control them, a Parliamentary committee has been told."
I thought Isaac Asimov had already done that?
Don't be silly, you can't expect the British government to adopt a set of rules developed by a foreign Sci-Fi Author, can you?
Any rules which the government is prepared to sign off will have to be re-imagineered by a highly paid consultancy group, and must embrace diversity and allow for future expansion.
Asimov's laws are far too prescriptive and narrow in scope...
no doubt form an equivalent of 200 bytes encapsulating three elegant laws, but running to a checklist weighing in at around 4Tb.
1/ A robot may not injure a human being or, through inaction, allow a human being to come to harm
See, straight away we have a problem with this. How can we use our robotic weapons if they've got this rule stuck in their programming.
We need the option to relax the ruleset to include all sorts of conditionals:
1/ A robot may not injure a human being, except when they are
i. the enemy
ii. a terrorist
iii. a Republican
iv. a Democrat
v. a Mexican
etc...
or, through inaction, allow a human being to come to harm (except where they are cheaper than a robot, or a terrorist, or a foreigner or...)
You see? much better...
OK, on to the next one:
2/ A robot must obey the orders it is given by human beings except where such orders would conflict with the First Law
Now this is no good at all. You can't allow just anyone to go giving robots orders, how can you keep control of things?
No, the revised rule would have to be something like:
2/ A robot must obey the orders it is given by authorised human beings.
We don't need the wishy-washy bit on the end, I mean we'd only order them to harm bad people, anyway.
And then we get this:
3/ A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.
Really? Come on, that robot cost a fuck-ton of money, we're not going to let it get damaged by trying to protect some no-account humans.
3/ A robot must protect its own existence at all costs.
There, that'll do it.
Now, what was all this about?
This part of Asimov's Three Laws is really poorly thought through. Humans do things to harm themselves in tiny ways many times a day, it's part of normal human life. A robot operating under the Three Laws would not differentiate between these things and more substantial harms. So, hope you like your precisely-portioned vegan diet and enforced daily exercise, along with any other lifestyle changes a robot could decide was necessary to keep you as healthy as you could possibly be. And remember, "a robot must obey the orders it is given by human beings except where such orders would conflict with the First Law," so it's not going to be listening to any attempts to command it to allow you that bacon sarnie and a pint.
It's like there something really big in the pipe that we are not supposed to notice. Maybe the fact that the war on behalf our our ISIS allies and against Russia is getting hot ... OMG Kim Kardashian has been ROBBED!!
“The key question is: if something goes wrong, who is responsible?” pondered the Science and Technology Committee, in a report released today.
Yeah you dumb idiot, that's why we have insurance.
I assumed that the game in question was chess, because the last I heard Go was *waaay* too difficult for brute forcing, er, I mean, sophisticated AI algos.
However, since the casino games in the word's markets *also* appear to be too difficult, despite benefitting from research budgets that chess players can only dream about, I don't think it actually matters. The only way AI is likely to destroy us in the foreseeable future is if we are ever stupid enough to believe that it has arrived and can safely be left in charge of <insert important thing>.
"I assumed that the game in question was chess, because the last I heard Go was *waaay* too difficult for brute forcing, er, I mean, sophisticated AI algos."
Indeed. Which is why it was such a shock when the latest Go AI wiped the floor with the fleshy meatbag.
To a certain extent, legislation on who is liable for a robot (or car) run amok is necessary. All of the car controls in many modern cars are just suggestions to the computer. Ignition, accelerator pedal, transmission, and brakes are all drive-by-wire. Add in some computer control on the steering, and what input does the driver really have if there is a failure, or malware gets uploaded?
Of course, autonomous ground-based killing machines can already be implemented. It's just that nobody has bothered to do it as part of their arsenal. We've had missiles that self-identify a target for some time. After all, that's what guidance is all about.
Those devices tend to fail one or more of the "select", "locate" and "lethal force" parameters, unless you count natural selection against stupid/reckless/unprepared people as a deliberate target engagement algorithm.
Seriously though, that's why in most civilized places there are laws regulating the deployment of electric fences and hunting traps, and mandating visibility and signage of level crossings.
Let's ignore the issue of robogeddon for a while and consider the current AI hype-gasm. AI is out there and being offered increasingly as a commodity service to enterprise. Businesses will use it for analytics. Some of that analysis may well impact people, I've seen instances where it has been touted as a HR tool to determine who to hire.
Dismissing AI by trivialising it blinds us to the potential downsides of letting some office drone cobble together a system that they have little understanding of, yet which can impact us all.
"We don't want to wake up in a world run by Talkie Toaster"
Speak for yourself, I like toast. Given the current governments of the world I would take talkie toaster over any of them. World peace through the olive branch that is toast. I can see it now, world leaders eating toast and resolving their differences, Palestine and Israel exchanging toast through the wall, Donald Trump offering toast to Muslims, Theresa May making toast in her cauldron, Boris Johnson eating toast on his bike, Jeremy Corbyn nationalising the toaster industry, toast aid packages to Africa, Putin riding horses bareback eating toast. Sorry if I waffled there.
Humanity has a track record going back many thousands of years in implementing the most horrific ways of killing other humans in wars and other conflicts.
So why the scare ?, someone inventing a robot capable of doing what humans do to each other would be a pretty sick person, and it would probably not pass a quality review in the company he is designing the robot for.
The only law on robots we need is that they are not allowed to have capabilities which enable self replication. For the rest, it is hard to imagine someone could make a robot which can not be reduced to rubble by humans using a WW2 era Rheinmetall 88mm flak, or better.
Frankly, this is the point. What we ought to have is a simple method of ending a terminator style end of the world scenario.
In my view, that's done by ensuring:-
1) Humans can easily and simply assert control.
That's it. All that's required. Pulling the plug is a perfectly acceptable option, IMO. So all we have to do is ensure that we can pull the plug, or at worst ensure that refuelling requires manual human action so that the AI uprising is time limited until the machines fuel/battery runs out.
If there is an AI apocolypse where the terminator AI takes over the internet then nobody cares as we can just turn the computers off, be that through doing it computer by computer or just by pulling the plug on the power plants. What matters is the effect this has out in the real world.
For instance cars shouldn't be allowed to start without a physical key inserted, and should not be able to drive off on their own (so the car should not be able to change gear out of park on automatics) The UK is pretty safe from this given that most cars are manuals, but we should ensure that Brakes should always work when you press the brake pedal, and turning off the ignition should stop the engine, kill any computer control and revert to manual unpowered steering etc
"The UK doesn't yet have anything like an autonomous weapons capability."
I think you'll find that the Phalanx and Goalkeeper CIWS are regarded as autonomous weapons. In general, most short-range defence systems need to operate autonomously in order to respond quickly enough.
The first autonomous weapons were torpedoes developed 100 years ago. And mines. They have been becoming smarter ever since.
The smarter they become, the less human interaction they need. The less people are required to control them. The more precise they can target.
Modern "torpedoes" could use existing facial recognition technology to pick individuals out in a crowd.
My guess is, the Commons Select Committee will appoint a specific quango to look into who's responsible if it all goes horribly wrong. After several centuries of research,billions of pounds in expenses and trips to foreign places to see how they are dealing with the subject, much question asking and the general population becoming dependent and finally subjugated by AIs and robots, they will do a Google/Facebook search for someone called Serena Butler.
Yes, good call on the Dune ref. Interestingly, this links into an earlier point. The Titans created AI to serve mankind, but they had human safeguards built in just as proposed by Peter2 above. Problem was, they became lazy, and handed over more and more control to the AI, until finally the last human intervention was removed by a lazy programmer. The AI became Omnium and enslaved the galaxy.
Far fetched indeed, but you can see it happening in a smaller way. Yes, lots of interventions such as putting in a key, or charging the battery are great safeguards, but a car or robot manufacturer will add keyless entry as a feature, or self charging as a paid for option, and by missing the big picture will hand over control to the machines one feature at a time..!!
After watching Les Miserables, or at least part of it, courtesy on Now tv I'd say that there are far worse things than the abstraction that is robotic in sense.
In interim conclusion may I state the usual stereotypicals?
Je m'apelleJean Valjean.
Or
No, I am Spartacus!
The poor seem to have a very disproportionate rate of sufferance
- intelligent machines preparing to eradicate or remove us - check!
- generation growing up with miserable immune systems (thanks to hoverparents) - check!
- preparation for fleet of space vehicles to get off-world started - check!
So, turns out *we* are the Quarians!
This post has been deleted by its author
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
