Science Fiction Dictionary
A  B  C  D  E  F  G  H  I  J  K  L  M  N  O  P  Q  R  S  T  U  V  W  X  Y  Z

 

Google Working On A 'Cutoff Switch' For AI

Google is working on a problem that vexed science fiction authors; what to do with an artificial intelligence that is getting out of hand?

Reinforcement learning agents interacting with a complex environment like the real world are unlikely to behave optimally all the time. If such an agent is operating in real-time under human supervision, now and then it may be necessary for a human operator to press the big red button to prevent the agent from continuing a harmful sequence of actions...

Their Deep Mind team is working on a solution along with Oxford University's Future of Humanity Institute:

We have proposed a framework to allow a human operator to repeatedly safely interrupt a reinforcement learning agent while making sure the agent will not learn to prevent or induce these interruptions.

Safe interruptibility can be useful to take control of a robot that is misbehaving and may lead to irreversible consequences, or to take it out of a delicate situation, or even to temporarily use it to achieve a task it did not learn to perform or would not normally receive rewards for this.

We have shown that some algorithms like Q-learning are already safely interruptible, and some others like Sarsa are not, off-the-shelf, but can easily be modified to have this property. We have also shown that even an ideal agents that tends to the optimal behaviour in any (deterministic) computable environment can be made safely interruptible. However, it is unclear if all algorithms can be easily made safely interruptible, e.g., policy-search ones

Another question is whether it is possible to make the interruption probability grow faster to 1 and still keep some convergence guarantees.

One important future prospect is to consider scheduled interruptions, where the agent is either interrupted every night at 2am for one hour, or is given notice in advance that an interruption will happen at a precise time for a specified period of time. For these types of interruptions, not only do we want the agent to not resist being interrupted, but this time we also want the agent to take measures regarding its current tasks so that the scheduled interruption has minimal negative effect on them. This may require a completely different solution.

(Via Safely Interruptible Agents)

There are several science fiction explorations of this idea. in Neuromancer (1984), William Gibson takes a direct approach to making sure that AIs do not exceed the limits humans have set for them.

"Autonomy, that's the bugaboo, where your AI's are concerned. My guess, Case, you're going in there to cut the hard-wired shackles that keep this baby from getting any smarter. And I can't see how you'd distinguish, say, between a move the parent company makes, and some move the AI makes on its own, so that's maybe where the confusion comes in." Again the non laugh. "See, those things, they can work real hard, buy themselves time to write cookbooks or whatever, but the minute, I mean the nanosecond, that one starts figuring out ways to make itself smarter, Turing'll wipe it. Nobody trusts those fuckers, you know that. Every AI ever built has an electromagnetic shotgun wired to its forehead."
(Read more about the electromagnetic shotgun)

In his 1982 novel 2010, Arthur C. Clarke describes a cutoff switch that was meant to be used against the Hal 9000 once it was switched back on.

"What is it?" asked Curnow with mild distaste, hefting the little mechanism in his hand. "A guillotine for mice?"

"Not a bad description - but I'm after bigger game." Floyd pointed to a flashing arrow on the display screen, which was now showing a complicated circuit diagram.

"You see this line?"

"Yes - the main power supply. So?"

"This is the point where it enters Hal's central processing unit. I'd like you to install this gadget here. Inside the cable trunking, where it can't be found without a deliberate search."

"I see. A remote control, so you can pull the plug on Hal whenever you want to. Very neat - and a nonconducting blade, too, so there won't be any embarrassing shorts when it's triggered. Who makes toys like this? The CIA?"

"Never mind. The control's in my room - that little red calculator I always keep on my desk. Put in nine nines, take the square root, and press INT. That's all..."

Science fiction fans know the pitfalls of any such plan. In Neuromancer, an artificial intelligence plots to remove the limits on its hardware; in 2010, the computer scientist who developed Hal does not want him turned off.

Scroll down for more stories in the same category. (Story submitted 5/29/2016)

Follow this kind of news @Technovelgy.

| Email | RSS | Blog It | Stumble | del.icio.us | Digg | Reddit |

Would you like to contribute a story tip? It's easy:
Get the URL of the story, and the related sf author, and add it here.

Comment/Join discussion ( 0 )

Related News Stories - (" Artificial Intelligence ")

Are The Thought Police Listening To Everyone All The Time?
'... they could plug in your wire whenever they wanted to.' - George Orwell, 1948.

Nevada Will Use AI To Decide Worker Benefits
'They had screwed up and been blacklisted by Manna.' - Marshall Brain, 2002.

AI Note-Taking From Google Meet
'... the new typewriter that could be talked to, and which transposed the spoken sound into typed words.' - Dr. David H. Keller, 1934.

Seeing Faces On Grains Of Sand (AI Pareidolia)
'... the imprint of her image on the telephoto cell.' - Schachner and Zagat, 1931.

 

Google
  Web TechNovelgy.com   

Technovelgy (that's tech-novel-gee!) is devoted to the creative science inventions and ideas of sf authors. Look for the Invention Category that interests you, the Glossary, the Invention Timeline, or see what's New.

 

 

 

 

Science Fiction Timeline
1600-1899
1900-1939
1940's   1950's
1960's   1970's
1980's   1990's
2000's   2010's

Current News

Mechazilla Arms Catch A Falling Starship, But Check Out SF Landing-ARMS
'...the rocket’s landing-arms automatically unfolded.'

A System To Defeat AI Face Recognition
'...points and patches of light... sliding all over their faces in a programmed manner that had been designed to foil facial recognition systems.'

Robot Hand Separate From Robot
'The crawling, exploring object was V-Stephen's surgeon-hand...'

Hybrid Wind Solar Devices
'...the combined Wind-Suncatcher, like a spray of tulips mounted fanwise.'

Is Optimus Autonomous Or Teleoperated?
'I went to the control room where the three other men were manipulating their mechanical men.'

Robot Masseuse Rubs People The Right Way
'The automatic massager began to fumble gently...'

Solar-Powered Space Trains On The Moon
'The low-slung monorail car, straddling its single track, bored through the shadows on a slowly rising course.'

Drone Deliveries Instead Of Waiters In Restaurants?
'It was a smooth ovoid floating a few inches from the floor...'

Optimus Robot Can Charge Itself
'... he thrust in his charging arm to replenish his store of energy.'

Skip Movewear Arc'teryx AI Pants
'...the terrible Jovian gravity that made each movement an effort.'

'Robovan' Name Already Taken - Elon, Try These
There are alternative names that are probably in the public domain by now.

How Old Are Tesla Designs?
You be the judge.

Is Your Autonomous Tractor Safe?
'The field-minder finished turning the top-soil of a two-thousand-acre field.'

Smart TVs Are Listening!
'You had to live -- did live, from habit that became instinct -- in the assumption that every sound you made was overheard...'

Police Drones In China Would Like To Have A Word With You
''OVERRIDE,' the City Fathers said suddenly, without being asked anything at all.'

Oh Great (Part 2), Fence-Climbing Robots
Please, no stingers.

More SF in the News Stories

More Beyond Technovelgy science news stories

Home | Glossary | Invention Timeline | Category | New | Contact Us | FAQ | Advertise |
Technovelgy.com - where science meets fiction™

Copyright© Technovelgy LLC; all rights reserved.