back to article Will Asimov fix my doorbell? There should be a law about this

Greetings from civilisation, for one more day at least. After tonight, I will no longer be a European citizen but an immigrant of indeterminate status. Don't worry, this won't be a Brexit diatribe. I resigned myself to belonging to a pariah nation long ago, a realisation born from half a lifetime's accumulation of World Cup …

Page:

  1. Pascal Monett Silver badge
    Thumb Up

    "Brexit means we are at last freed to obey whatever the Americans instruct us to do"

    An instant classic !

    Oh, and congratulations on your French. You have impeccably captured the essence of the French mind when it comes to swearing, and the British :).

    1. ElectricPics

      Re: "Brexit means we are at last freed to obey whatever the Americans instruct us to do"

      Two down votes? Really?!

      1. Rol Silver badge

        Re: "Brexit means we are at last freed to obey whatever the Americans instruct us to do"

        I guess we have two Trump curious readers, not supporters I hasten to add, as this site is so loaded with qualified facts that they wouldn't be able to operate a mouse for all the vomiting and retching they'd be doing.

        1. Rol Silver badge

          Re: "Brexit means we are at last freed to obey whatever the Americans instruct us to do"

          I love the consistency of the voting. It's almost validation of the argument, as every topic that seeks to sleight The Donald has two down votes. Now if we correlate it with comments that slag Putin off, we might just see a trend.....

          ....two individuals, with sufficient ability to comprehend at least half of what is discussed on here, and a mission to subvert humanity, for a regular pay check from Blofeld or whatever Putin's alias is these days.

          1. Rol Silver badge

            Re: "Brexit means we are at last freed to obey whatever the Americans instruct us to do"

            You've got to ask yourself this though....

            Now that the political objectives have been met, will Blofeld continue to pay for your loyalty or seek a more cost effective and permanent solution?

    2. Anonymous Coward
      Anonymous Coward

      Re: "Brexit means we are at last freed to obey whatever the Americans instruct us to do"

      ""Brexit means we are at last freed to obey whatever the Americans instruct us to do"

      Hmmmm, actually, no, and quite the reverse actually. UK now officially belongs to Trumpistan.

  2. TimMaher Bronze badge
    Facepalm

    Nice Asimov.

    Now, how’s about a good Zelazny?

    Was Brexit the last short story in “All the sounds of fear”?

    1. tiggity Silver badge

      Re: Nice Asimov.

      upvote for the Harlan Ellison reference

    2. BillG
      Thumb Up

      Re: Nice Asimov.

      Now, how’s about a good Zelazny?

      Not necessary. As we all know, All Roads Lead to Amber.

      1. MachDiamond Silver badge

        Re: Nice Asimov.

        "All Roads Lead to Amber."

        Isn't that "Ember"

        Wait, that's Jeanne DuPrau. Nevermind.

    3. IceC0ld Silver badge

      Re: Nice Asimov.

      Don't worry, this won't be a Brexit diatribe.

      ah yes, the EU, I remember that as if it was yesterday ..........................

  3. alain williams Silver badge

    3 laws for AI

    Ah, crafted in a more innocent age when the general belief was that we all work together for a common good.

    Today is a post truth era where corporations (CEOs & their lackeys) & politicians will say whatever bollocks they think will bamboozle us to overlook the current heist that they are trying to pull off. AIs will do whatever their masters direct them to do; ie they will work to the benefit of the corporation or government that owns them, they will have scant regard to the harm that they will cause to the rest of us.

    This piece by a FT columnist is worth looking at.

    1. TRT Silver badge

      Re: 3 laws for AI

      The one I have trouble with is #2.

      "An AI must obey human commands..."

      (1) Ha ha ha ha!

      "Siri, Please set chill-out room lights to 50 per cent." ... ... ... ... "Playing 50 cent in children's bedroom"

      (2) All human commands? Should it do what the human says or what the human meant?

      I think... a story in my archival memory storage... was it called F.R.E.D.? About a robot that, on its first day in its new household, was instructed not to damage the flowers in the garden. And a few weeks later was instructed to cut the grass whilst the family went out for the day. When they returned they found it on hands and knees in the middle of the lawn in the pouring rain getting severely damaged by the soaking, and using a pair of scissors to carefully avoid all the daisies.

      1. JetSetJim Silver badge

        Re: 3 laws for AI

        > "Siri, Please set chill-out room lights to 50 per cent." ... ... ... ... "Playing 50 cent in children's bedroom"

        I prefer the Alexa "Digger Digger" incident

      2. Little Mouse

        Re: 3 laws for AI

        I don't remember that particular gardening story, but it's true that a recurring theme of Asimov's robot series was that the three laws didn't really work all that well

        Mr Raphan might have done well to actually read a few of them (or even watch the movie, heaven forfend) before suggesting that they might actually be a good idea.

        1. TRT Silver badge

          Re: 3 laws for AI

          It's doing my head in trying to find the source... I'm sure the robot was called F.R.E.D. which was also the name of the story. It was around the time, maybe 1976-1980, that I read the Brian Aldiss edited anthology "The Penguin Science Fiction Omibus" - the cover had a creepy ice cream sundae on it, and a book called "The Very Slow Time Machine" by Ian Watson which was an anthology, The Stainless Steel Rat series as well. But I think it was in a series of numbered books titled like "The Fourth -publishers name- Sci-Fi Collection" or "The -publishers name- Sci Fi Short Stories Volume ..."

          Penguin? Armada? Keep thinking "Fox" for some reason. Could have been Armada... I see a ship with black sails on a white background on the spine. And the cover of the book was an artist's illustration of the robot from the story - I think it ended up with the robot developing a personality as a result of getting rained on, then running away instead of facing going back for a factory refurbishment, and it befriended the little girl of the family - saved her from something nasty but they thought it had done a bad thing? Gah! Memory. Tsk. I don't recall seeing the book since it was on my windowsill over 40 years ago along with the rest of a rag-tag of pulp fiction anthologies including Asimov and Clarke - the books all got damaged by condensation whilst I was away at boarding school.

          Sorry for the rambling long post... just trying to straighten thoughts out by writing... Grrr. Frustrating. That'll be a 3.30 in the morning wake up for me when the memory finally re-emerges.

          Armada Sci Fi 3? Maybe... The robot was indoors, kind of scary. A bit boxy looking.

          1. Inventor of the Marmite Laser Silver badge

            Re: 3 laws for AI

            TRT. At least part of your musings seems related to the film Bicentennial Man (IIRC). A gentle and surprisingly good film starring Robin Williams, Sam Neill, Embeth Davidtz (in a dual role), Wendy Crewson, and Oliver Platt. It was based on the 1992 novel The Positronic Man by Isaac Asimov and Robert Silverberg (which is itself based on Asimov's original 1976 novelette "The Bicentennial Man"),

            It says here.

            1. TRT Silver badge

              Re: 3 laws for AI

              I have read that one, yes. Bicentennial man the film was fairly good if bordering on twee sentimentality. There are similar events in both works I think, unless I'm mixing them up a little.

              As predicted my memory woke me up at stupid o'clock and came up with the name Catherine Gleason as the author of the story FRED. A quick bit of Google (what DID we do before the internet? Oh yeah, remembered everything), and I find that it was included in the anthology Armada SF3, and the most noted book she wrote was The Lampton Worm. Anyway, book ordered so I can refresh my memory.

              1. Fruit and Nutcase Silver badge

                Re: 3 laws for AI

                The Lampton Worm

                As I read the above, out popped from my memory... "The Lair of the White Worm". The film by Ken Russel, loosely based on the book of the same name by Bram Stoker, which itself was based on the legend of the Lambton Worm.

                The film features Amanda Donohoe, Catherine Oxenberg, and an early film appearance by Hugh Grant

                1. TRT Silver badge

                  Re: 3 laws for AI

                  Indeed. A bonkers film. But charming in its own way.

                2. TRT Silver badge

                  Re: 3 laws for AI

                  Oh, and Peter Capaldi.

                  1. Franco Silver badge

                    Re: 3 laws for AI

                    Ah yes, the worm was scared off by his bagpipes if I remember rightly. From what I remember the film was quite enjoyable but utterly ridiculous.

                3. John Brown (no body) Silver badge

                  Re: 3 laws for AI

                  "based on the legend of the Lambton Worm."

                  <Looks out window at Penshaw Monument built on the top of the hill the worm coiled around and nods in agreement>

        2. Phil O'Sophical Silver badge

          Re: 3 laws for AI

          I don't remember that particular gardening story, but it's true that a recurring theme of Asimov's robot series was that the three laws didn't really work all that well

          Not surprising, since he invented them as a literary device just so that he could write stories showing how they wouldn't work.

          1. Michael Wojcik Silver badge

            Re: 3 laws for AI

            Well, they're not just a literary device. They're also a thought experiment in a concise, easily comprehended, logically consistent set of axioms can still produce unexpected results. Asimov's robot stories (and to a lesser extent the novels, which were also significantly concerned with social effects of machine intelligence) are as much about logic and complexity as they are about robots.

            Of course that has never stopped people from interpreting them as prescriptive, or even as descriptions of fact. I remember an Asimov editorial from IA'sSFM around 1980 in which he described getting calls from reporters asking about the Three Rules, after a Japanese maintenance worker was killed by an industrial robot.

        3. Trygve Henriksen

          Re: 3 laws for AI

          I effing hate those 3 laws.

          They're so vague that there's no way that a robot can follow them. At all!

          I'm very, very cross at Asimov to have even thought them out since muggles all over the world think they're a good idea and it seems, wants to implement them!

          1. Will Godfrey Silver badge
            Angel

            Re: 3 laws for AI

            Siri? is that you?

            1. quxinot Silver badge

              Re: 3 laws for AI

              Goddamnit Cortana I told you to be quiet!

              Stupid updates must have changed that setting. Again, grrrr.

            2. TRT Silver badge

              Re: 3 laws for AI

              Hey Siri! What are the three laws of robotics?

              Ohm's Law,

              Sod's Law and

              Cole's Law.

          2. amanfromMars 1 Silver badge

            Re: 3 laws for AI

            I effing hate those 3 laws.

            They're so vague that there's no way that a robot can follow them. At all! ....Trygve Henriksen

            So vague, TH? Surely you cannot be serious? After all, they only really have to obey the one simple rule ..... be extra especially good at what one is doing and has planned to be done as a consequence of what is being done in all of that stuff which is yet to be done.

            That however requires that you realise such would be as autonomous virtual machines exercising executive operating system programs all of their own. I'm sure you can imagine that can make a hell of a lot of folk extremely uneasy .... and absolutely flabbergasted ........ hence the slow drip feeds of information for Stealthy Advanced IntelAIgent Services ...... for Virtually Real Operation[s]

          3. Teiwaz Silver badge

            Re: 3 laws for AI

            'Muggles' maybe, but it's the thoughtless corporate types driving A.I. with nothing but dollars in their eyes and being able to fire even the robots in H.R.

            And so we end up with cars that mow down pedestrians, and probably sooner or later something much like ED 209.

            We'll be lucky if there's anything vaguely in the training about not inflicting harm.

            Maybe a digital readout to assure punters

            'It's been XXX days since this unit caused a death. Have a Nice Day'

          4. dajames Silver badge

            Re: 3 laws for AI

            I effing hate those 3 laws.

            They're so vague that there's no way that a robot can follow them. At all!

            What's that, R2?

            The three laws are more what you'd call "guidelines" than actual laws?

        4. JDX Gold badge

          Re: a recurring theme ... was that the three laws didn't really work all that well

          I don't think that's really true. The stories focus on the rare instances when things go wrong, and are seen as abnormalities against millions of robots not causing any fuss.

          To say the Robot stories demonstrate the Laws work badly seems to me like saying modern medicine doesn't work all that well based on watching Dr. House's TV show.

          1. ibmalone Silver badge

            Re: a recurring theme ... was that the three laws didn't really work all that well

            Actually, House gives the impression modern medicine works extremely well and always comes down to a solvable logic puzzle with a dash of obscure knowledge, a bit like the Asimov stories. There's a reason: House is inspired by the Sherlock Holmes stories (House / Homes, elementary my dear Wilson), and Asimov really liked detective stories, constructing many of his short stories in that form.

            1. TRT Silver badge

              Re: a recurring theme ... was that the three laws didn't really work all that well

              And the protagonist is a drug addict...

              1. TRT Silver badge

                Re: a recurring theme ... was that the three laws didn't really work all that well

                You're saying Holmes didn't have an cocaine addiction and House didn't similarly have an addiction to Vicodin?

                Asimov of course wasn't addicted to any chemical that we know of, and was in favour of clean living. Sadly he died of diseases associated with AIDS after he contracted HIV - something he was told to keep quiet about due to prejudice at the time.

            2. nanchatte

              Re: a recurring theme ... was that the three laws didn't really work all that well

              Asimov was a fan of the detective format... The first full-on novels he wrote based in his "Foundation" galaxy (the Elijah Baily and R Daneel Olivaw books) are testament to that.

            3. Roj Blake

              Re: a recurring theme ... was that the three laws didn't really work all that well

              The character of Sherlock Holmes was originally based on a doctor who once taught Conan Doyle.

          2. Rol Silver badge

            Re: a recurring theme ... was that the three laws didn't really work all that well

            What Asimov's laws, and the knowledge required to fully comply with them makes obvious, is that like a professional fighter, you first need to be aware of how to permanently damage your sparing partner in order that you can avoid permanently damaging your sparing partner.

            Now imagine the uproar on our agreeing to an AI being taught how to kill humans, even though that knowledge is fundamental to the objective of avoiding killing humans?

            The future of robotics as penned by Asimov in one of his doom laden robots gone mental sagas, is that they stop trying to mimic humans and instead focus on automated tasks like - flying around scooping up mosquitoes and gnats by the thousands. No complex AI required, just sufficient processing and robust coding to avoid freaking people out.

            1. TRT Silver badge

              Re: a recurring theme ... was that the three laws didn't really work all that well

              They have... detailed files.

          3. Michael Wojcik Silver badge

            Re: a recurring theme ... was that the three laws didn't really work all that well

            The stories focus on the rare instances when things go wrong, and are seen as abnormalities against millions of robots not causing any fuss.

            Irrelevant to the force of Asimov's robot stories. The point isn't to explore whether the Three Rules produce an acceptable defect rate or are probabilistically "good". It's to consider a series of logic puzzles in which a system of three simple axioms is shown to produce surprising results.

            In that sense the Three Laws work "poorly", for their ostensible purpose (though well for their pedagogical one), because they appear to offer simple, absolute guarantees, but it's possible to find numerous exceptions. The principle of least surprise is violated.

            It's certainly possible to claim that in the world of the robot stories the Three Laws work "well" in a practical sense. That's much like Chaitin's argument that Hilbert's Entscheidungsproblem was a resounding success, because Church's and Turing's proofs that it can't be solved introduced formalisms that were invaluable in spurring the development of digital computing; a mathematical "failure" (not really a failure, of course, and Chaitin doesn't characterize it as such) contributed to a major technological advance. You could say the same of the Three Laws (in their world): mapping their logical "failures" helps cement their application in technology.

            But reading that as a principal theme of the stories rather goes against the interpretations most readily inferred from the text, I think. The stories are about how the Laws fail.

        5. ibmalone Silver badge

          Re: 3 laws for AI

          I don't remember that particular gardening story, but it's true that a recurring theme of Asimov's robot series was that the three laws didn't really work all that well

          This. Absolutely this.

          Asimov loved detective stories, the three laws gave him a framework to create all kinds of logical puzzles and paradoxes. Yes, they usually arise from situations or unusual conflicts, but they're created for fiction.

          Part of the setup in most of the robotics stories is, "this couldn't possibly have happened under the three laws," which is part of the fiction and a jumping-off point for "what's gone wrong here?". None of the laws can actually be implemented with our technology as it stands because we can't create systems that are capable of assessing whether the first law is met (needed for 2 and 3). Even if we could, such judgements have to be probabilistic, which leads us into weighting and one of the very paradoxes Asimov used for a story, https://en.wikipedia.org/wiki/Runaround_(story).

          The first law actually supposes a total knowledge of consequences, and if you've got that, how far do you assess the chain? Readers of Terry Pratchett's Going Postal may remember Mr Pump's more comprehensive assessment of the downstream harms caused by the central character's "victimless" crimes.

          1. Intractable Potsherd Silver badge

            Re: 3 laws for AI

            In none of the robot stories we're the Laws of Robotics actually broken, which shows their resilience in that world. The problems occurred when there was conflict between different laws, or where the balance between the laws was changed.

            1. ibmalone Silver badge

              Re: 3 laws for AI

              The problem remains that they are stories, they're not broken in that world because that's one of the scaffolds of the story. Any fictional world works how its author wants it to, it doesn't demonstrate that those rules work in the real world.

            2. nanchatte

              Re: 3 laws for AI

              Actually, the First law was broken by a R. Giskard Reventlov, A robot who self redacted his programming when he came up with the concept of the Zeroth law: that under certain circumstances it was acceptable to harm individuals if doing so would prove beneficial to society as a whole...

              Asimov's books always asked the deeper questions about society vs. the individual, moral absolutism, war, idealism and societal stagnation...

              Wonderful stuff...

              1. Loyal Commenter Silver badge

                Re: 3 laws for AI

                Also worth remembering that in Asimov's books, humans never encounter any intelligent alien species, because the three laws mean that the robot pioneers that went first exterminated everything, in order to "protect" humanity. Maybe there should be a -1th law about not committing genocide on a galactic scale?

        6. ThatOne Silver badge
          Devil

          Re: 3 laws for AI

          > before suggesting that they might actually be a good idea

          Nonsense, we're speaking about a politician: It's never about improving peoples' lives, it's about looking like one has a firm grip on the situation and is working towards improving society (helping the weak and poor, save kittens from trees, etc.).

          His PR team wanted to position him as "a guy who has solutions". Never mind the actual suggestion, the point is he made one.

      3. This post has been deleted by its author

      4. Ken Hagan Gold badge

        Re: 3 laws for AI

        "The one I have trouble with is #2."

        You probably ought to be suspicious of #1. Yes,it sounds excellent, but so does the equivalent rule for human beings ... and yet no legal system ever has actually imposed that rule on people. You *are* allowed to hurt people under some circumstances and you are allowed to leave people to get hurt under quite a large number of circumstances. I expect Mr Asimov was aware of this.

        1. Throatwarbler Mangrove Silver badge
          Terminator

          Re: 3 laws for AI

          Right, but robots aren't supposed to hurt people in Asimov's world, which is entirely consistent. Conversely, he apparently never imagined the military-industrial complex wanting to build killer robots (or maybe he did and I never read such a story).

Page:

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2020