I love you. I will kill you! I want to make love to you: The evolution of AI in pop culture Stephen Hawking is scared. "The development of full artificial intelligence could spell the end of the human race," Hawking has said. With the creeping integration of soft AI into our lives in the form of Siri and personalized ads on social media, these computational mini-minds serve as a constant reminder that the evolution …
Wait a minute, every single example I've seen, or heard of up to now, of a learning system was specifically trained to learn in one domain.
I have yet to see a machine that can learn chess without any programming, then learn backgammon, then move on to learning knitting and finally learn how to plan and build a wall. Of course, said computer will need arms, at the very least, in order to actually do something, but I think that will be easier than the learning part.
I have yet to see a machine that can learn chess without any programming, then learn backgammon, then move on to learning knitting and finally learn how to plan and build a wall.
I've yet to see a human "learn" chess without being given the ruleset and running many simulation sequences as practise. Similarly backgammon, knitting and even bricklaying are all "taught" to people by providing them with the relevant rules / framework and then allowing them to run multiple simulation sequences, correcting errors until the results are considered acceptable.
I'm not sure how different that process is to what you would mean by "programming an AI", but my massively rusty neural network experience would suggest that they are not that dissimilar at all ;)
"I've yet to see a human "learn" chess without being given the ruleset ... "
Point.
But: what about inventing chess, etc ?
I have yet to see a computer system invent something new - in other words show the ability of creativity, of original thougt. (Random errors from buggy code or malfunctions do not count.)
Now that would be AI.
There was a computer program that "discovered" prime numbers all by itself.
It was not really AI. It was a goal-seeking algorithm that could apply the rules of symbolic algrebra. It had the axioms of arithmetic hard-coded, and its goal was to prove hypotheses with high values of interestingness, where interestingness was a heuristic based on the relation of a hypothesis to other hypotheses and theorems and their interestingnesses (ie, if we can prove this, then we get that and that and that ... )
ISTR it got as far as inventing and trying to prove Goldbach's conjecture (every even number greater than two can be expressed as the sum of two prime numbers). It failed to prove it, which is not very surprising ... the best human mathematicians of the last 250 years or so haven't been able to prove it either!
There's also the proof of the four-colour theorem which is too complex for any human unassisted by a computer to comprehend. So it it proved, and if so what is it proved by?
STR it got as far as inventing and trying to prove Goldbach's conjecture (every even number greater than two can be expressed as the sum of two prime numbers).
Did it manage deducing the existence of rice pudding and income tax ?
Bungie introduced Marathon, the forerunner of all the Halo business. It featured not one but three AIs, all with starkly different personalities, and one of them psychotic. Unfortunately, such was the site of the art of actual game AI that the random human NPCs displayed an almost unerring tendency to get in the way of every shot or sight line. So they became known as "Bobs."
Artificial intelligence is never a threat of itself - only when it is deliberately used and directed by one human against another, e.g. CPU controlled "intelligent" bombs, missiles, guns etc.
The only thing that might be of concern would be if a machine were developed that was self-aware. It is however very possible that sentience can only be a part of living organisms, and we have absolutely no idea what "life" is, but it is very unlikely that it would occur spontaneously in a silicon chip.
"Artificial intelligence is never a threat of itself [...]"
If an AI decision is able to inflict damage in any way - then there is always a chance of unintended outcomes or collateral damage. That is why Asimov's Laws stipulate overriding contingencies as a catch-all.
There is the classic dilemma of the runaway train heading for a broken bridge and certain destruction. If it is diverted onto a spur then the passengers will be saved. However - that guarantees that a man in the spur will be killed. Does the machine choose the greater good - or avoid a direct action that would deliberately kill the man on the spur?
Does the machine choose the greater good - or avoid a direct action that would deliberately kill the man on the spur?
Indeed, Asimov added the "Zeroth Law" to the previous three in order to explore similar issues. To bring the great writer's work into focus on the developing issue of self driving cars and perhaps who is liable if one should cause harm to a person, I recommend a jaunt through Asimov's "Sally".
"
Does the machine choose the greater good - or avoid a direct action that would deliberately kill the man on the spur?
"
It follows the algorithms that the human who programmed it wrote. If those algorithms have unintended consequences, it's the result of the programmer's lack of foresight rather than the fault of the machine.
Machines do not "make decisions" and are not likely to do so in the foreseeable future. They just follow a pre-programmed algorithm - albeit one that may be pretty complex.
Machines do not "make decisions" and are not likely to do so in the foreseeable future. They just follow a pre-programmed algorithm - albeit one that may be pretty complex.
Which differs from a human brain taking a decision how precisely?
A machine makes a decision when it operates with internally generated code or huge internally generated continuously modified weighting tables on equally huge internally generated tables of data, heuristically pruned to fit in available memory. It becomes impossible for a human to understand why any particular instance arrives at a particular point A or not-A. Potentially this, even if we can dump a Terabyte of internal state at the precise moment that the decision was taken.
When you are running and are tripped by something you didn't see, you'll try to regain your balance. Can you tell us the details of your last success, or what you would do next time to avoid your fall, or even whether falling was avoidable? Yet clearly we do learn to run. Young kids fall over a lot more than adults. And if/when we advance from building intrinsically stable wheeled and tracked vehicles to bipedal "mechas", I have little doubt that the same will be true of their control systems.
Does the machine choose the greater good - or avoid a direct action that would deliberately kill the man on the spur?
A dilemma which the logic in self-driving cars will have to incorporate, unless we choose to refuse to incorporate any concept of "the greater good" which is itself a decision. Will it be explicit (programmer playing god, here's the algorithm which decides whether you live or die)? Or implicit (the AI has programmed goals and code that evolves in time as it processes more events, and we really don't know in advance what it will do faced with a choice between two different crashes).
I once found myself in a meta- version of this dilemma. Thanks to my own inattention, I was hurtling towards a give-way sign much too fast to stop and realized I might have to decide between a collision with another vehicle or going off-road into trees. I never got to take the decision, because the fates or whatever decreed that there was no other vehicle crossing my path.
We don't know an awful lot about self-awareness or how it can arise. In fact I'm not sure that we actually know anything. How do you prove to me that you are self-aware rather than just programmed to assert that you are? Or even, how do I know that I am self-aware rather than just something else's dream? The ancient Greeks and Shakespeare understood this. Occam's Razor is the only way out that I know.
Two good SFnal treatments of awareness arising unexpectedly: Greg Bear Queen of Angels and Charles Stross Rule 34.
Check the etymology of "robot". It was Karel Čapek's brother, not Asimov who coined it.
This post has been deleted by its author
Check the etymology of "robot". It was Karel Čapek's brother, not Asimov who coined it.
The word robot simply means slave in Czech and it was part of the title of Čapek's play in the original language. It then entered the English language as a name for mechanical humanoids.
The word robotics, to describe the science of robot construction and development, was most definitely coined by Asimov.
AI was a required class in my computer science curriculum during the 90s but later dropped as a requirement. One of the things I took from the class was how amazing the human mind is and how difficult it is to replicate. The instructor mentioned how he would place an object on a table, say a rubber ball, and have the class write a 5,000 word essay describing the object. Sounds absurd at first, but notions like "how much does the ball weigh?" "how high would the ball bounce?" "what does the ball feel like?" and so on...soon the essays are completed.
... to quote the penultimate line from Colossus: The Forbin Project where the US and the USSR both create super computers to defend their countries and prevent war.
It ends with the computers join forces to become "World Control" and obey their programing by taking over and thereby absolutely preventing war but, as World Control says, "freedom is just an illusion" and says how mankind will advance under its guidance
Forbin angrily replies "NEVER!"
So which would *you* have? Peace under computer control or war under human control...?
So which would *you* have? Peace under computer control or war under human control...? .... Graham Marsden
And those are barely two possible options, Graham Marsden, in a vast see of opportunities.
Hi, Google DeepMinded,Not so much a CV, more an AIMission Statement with tried and tested roadmaps to/from Remote Virtual Command and Control of Earthed SCADA Systems ……. http://forums.theregister.co.uk/forum/1/2016/01/29/ai_in_tv_film_books_games/#c_2762789
Search Engine Optimisation v2.0 [and above] is surely logically a Future Product Placements Engine …… Advanced Intelligence Resource with Immaculate Source, with the likes of a Google not searching for answers, both popular and controversial, but providing them with streams of supporting evidence.
Such would be akin to the Private Mentoring with Pirated Monitoring of Future Events with AIDerivative Programming for Projects/Semi-Autonomous, Self-Actualisation of Virtual Realities.
It is difficult, and maybe even impossible, to see or imagine a defence against such in an attacking configuration.
Regards,
Graham C
xxxxxxxxxxxx
Thank you for your interest in DeepMind.
We endeavour to review every application we receive at DeepMind. We have had an overwhelming level of interest so do bear with us, we will get back to you as soon as we can.
Kind Regards,
The DeepMind Team
Something new and enlightening to report on, El Reg?
Truth is: It would take at the most 0.3 generations before there would be a Colossus cult with tens of millions of followers with priests, robes, rituals, holy books, verified miracles and merchandise.
Just look at scientology, north korea, televangelism, the moon cult, islamic state, hare chrishna and neo-liberalism, craziness designed and engineered by mere humans, yet, millions and millions of suckers are irresistibly attracted like wasps to a picnic. Most people *want* to be controlled, they *want* to have all choices made for them, they *want* to have a complete check-list with all the answers for any situation.
With Colossus, there is a God who actually listens, always sees everything, punishes transgressors and probably can perform real miracles too. I'd say that 30 - 80% of the population would like that (and, following tradition, would ritually murder the other fraction in the traditional way).
Dark.Star
co-written by, edited by and starring Dan O'Bannon
- Star Wars [computer animator]
-Jodowosky's Dune [never made, sadly]
-Alien [writer, effects supervisor]
-Total Recall
Screamers, a science-fiction film about post-apocalyptic robots programmed to kill. Adapted from the Philip K. Dick story "Second Variety".
That's some career!
There is literature outside the US and UK, y'know...
Stanislav Lem's "Fables for Robots" tells the fairytales that robots tell their offspring. The universe is populated by robots. In some of their legends, a particularly devious monster appears. Made largely of corrosive water, he is only a distant collective memory. An enemy that, in their early pre-history, may have tried to destroy them. Some even believe that he created them in the first place, but that's a disgusting idea and obviously nonsense.
AI is vaporware. Hawking opinion on it is worthless... as is Musk's (I love how Musk was nominated for The Luddite Award for his tirade on "the AI apocalypse").
Computers really haven't changed that much in 20 years. They still do the same crap they did 20 years ago, only faster. Humans literally have no idea how to create artificial sentience, and frankly that's a good thing. We are an irrational and destructive species that have nothing of value to add to the universe. Lets be honest about it.
....yet it's hard to truly define, even in terms of human illogicality and irrationality, the understanding of a so-called Natural intellect, at least for myself, Marvin (-;
Oh. Do human nerve tissues conduct light of some kind of a chemical fluorescence, or maybe I just had a nightmare? Have you seen anyone just slightly glowing inm the darkness?
And yes, this is a question posed exactly as it is written here @ElReg Commentards section. One wouldn't be much surprised, anyway.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
