Hot Topics

6 August 2012, 6.18am AEST

Artificial intelligence – can we keep it in the box?

We know how to deal with suspicious packages – as carefully as possible! These days, we let robots take the risk. But what if the robots are the risk? Some commentators argue we should be treating AI (artificial intelligence) as a suspicious package, because it might eventually blow up in our faces…

Authors

Disclosure Statement

Huw Price has received funding from the Australian Research Council.

Jaan Tallinn is one of the founders of Skype. He does not work for, consult to, own shares in or receive funding from any company or organisation that would benefit from this article. He is a donor to the Singularity Institute.

The Conversation provides independent analysis and commentary from academics and researchers.

We are funded by CSIRO, Melbourne, Monash, RMIT, UTS, UWA, Canberra, CDU, Deakin, Flinders, Griffith, La Trobe, Murdoch, QUT, Swinburne, UniSA, UTAS, UWS and VU.

Articles by These Authors

25 April 2011 Time is but a dream … or is it?

Q9ycgyyb-1343711877 — “We should stop treating intelligent machines as the stuff of science fiction.” Cea.

We know how to deal with suspicious packages – as carefully as possible! These days, we let robots take the risk. But what if the robots are the risk? Some commentators argue we should be treating AI (artificial intelligence) as a suspicious package, because it might eventually blow up in our faces. Should we be worried?

Exploding intelligence?

Asked whether there will ever be computers as smart as people, the US mathematician and sci-fi author Vernor Vinge replied: “Yes, but only briefly”.

He meant that once computers get to this level, there’s nothing to prevent them getting a lot further very rapidly. Vinge christened this sudden explosion of intelligence the “technological singularity”, and thought that it was unlikely to be good news, from a human point of view.

Was Vinge right, and if so what should we do about it? Unlike typical suspicious parcels, after all, what the future of AI holds is up to us, at least to some extent. Are there things we can do now to make sure it’s not a bomb (or a good bomb rather than a bad bomb, perhaps)?

AI as a low achiever

Optimists sometimes take comfort from the fact the field of AI has very chequered past. Periods of exuberance and hype have been mixed with so-called “AI winters” – times of reduced funding and interest, after promised capabilities fail to materialise.

Some people point to this as evidence machines are never likely to reach human levels of intelligence, let alone to exceed them. Others point out that the same could have been said about heavier-than-air flight.

The history of that technology, too, is littered with naysayers (some of whom refused to believe reports of the Wright brothers' success, apparently). For human-level intelligence, as for heavier-than-air flight, naysayers need to confront the fact nature has managed the trick: think brains and birds, respectively.

A good naysaying argument needs a reason for thinking that human technology can never reach the bar in terms of AI.

Pessimism is much easier. For one thing, we know nature managed to put human-level intelligence in skull-sized boxes, and that some of those skull-sized boxes are making progress in figuring out how nature does it. This makes it hard to maintain that the bar is permanently out of reach of artificial intelligence – on the contrary, we seem to be improving our understanding of what it would take to get there.

Moore’s Law and narrow AI

On the technological side of the fence, we seem to be making progress towards the bar, both in hardware and in software terms. In the hardware arena, Moore’s law, which predicts that the amount of computing power we can fit on a chip doubles every two years, shows little sign of slowing down.

In the software arena, people debate the possibility of “strong AI” (artificial intelligence that matches or exceeds human intelligence) but the caravan of “narrow AI” (AI that’s limited to particular tasks) moves steadily forward. One by one, computers take over domains that were previously considered off-limits to anything but human intellect and intuition.

We now have machines that have trumped human performance in such domains as chess, trivia games, flying, driving, financial trading, face, speech and handwriting recognition – the list goes on.

Along with the continuing progress in hardware, these developments in narrow AI make it harder to defend the view that computers will never reach the level of the human brain. A steeply rising curve and a horizontal line seem destined to intersect!

What’s so bad about intelligent helpers?

Would it be a bad thing if computers were as smart as humans? The list of current successes in narrow AI might suggest pessimism is unwarranted. Aren’t these applications mostly useful, after all? A little damage to Grandmasters' egos, perhaps, and a few glitches on financial markets, but it’s hard to see any sign of impending catastrophe on the list above.

That’s true, say the pessimists, but as far as our future is concerned, the narrow domains we yield to computers are not all created equal. Some areas are likely to have a much bigger impact than others. (Having robots drive our cars may completely rewire our economies in the next decade or so, for example).

The greatest concerns stem from the possibility that computers might take over domains that are critical to controlling the speed and direction of technological progress itself.

Software writing software?

What happens if computers reach and exceed human capacities to write computer programs? The first person to consider this possibility was the Cambridge-trained mathematician I J Good (who worked with Alan Turing code-breaking at Bletchley Park during the second world war, and later on early computers at the University of Manchester).

In 1965 Good observed that having intelligent machines develop even more intelligent machines would result in an “intelligence explosion”, which would leave the human levels of intelligence far behind. He called the creation of such machine “our last invention” – which is unlikely to be “Good” news, the pessimists add!

In the above scenario, the moment computers become better programmers than humans marks the point in history where the speed of technological progress shifts from the speed of human thought and communication to the speed of silicon. This is a version of Vernor Vinge’s “technological singularity” – beyond this point, the curve is driven by new dynamics and the future becomes radically unpredictable, as Vinge had in mind.

Not just like us, but smarter!

It would be comforting to think that any intelligence that surpassed our own capabilities would be like us, in important respects – just a lot cleverer. But here, too, the pessimists see bad news: they point out that almost all the things we humans value (love, happiness, even survival) are important to us because we have particular evolutionary history – a history we share with higher animals, but not with computer programs, such as artificial intelligences.

By default, then, we seem to have no reason to think that intelligent machines would share our values. The good news is that we probably have no reason to think they would be hostile, as such: hostility, too, is an animal emotion.

The bad news is that they might simply be indifferent to us – they might care about us as much as we care about the bugs on the windscreen.

People sometimes complain that corporations are psychopaths, if they are not sufficiently reined in by human control. The pessimistic prospect here is that artificial intelligence might be similar, except much much cleverer and much much faster.

Getting in the way

By now you see where this is going, according to this pessimistic view. The concern is that by creating computers that are as intelligent as humans (at least domains that matter to technological progress), we risk yielding control over the planet to intelligences that are simply indifferent to us, and to things that we consider valuable – things such as life and a sustainable environment.

If that sounds far-fetched, the pessimists say, just ask gorillas how it feels to compete for resources with the most intelligent species – the reason they are going extinct is not (on the whole) because humans are actively hostile towards them, but because we control the environment in ways that are detrimental to their continuing survival.

How much time do we have?

It’s hard to say how urgent the problem is, even if pessimists are right. We don’t yet know exactly what makes human thought different from current generation of machine learning algorithms, for one thing, so we don’t know the size of the gap between the fixed bar and the rising curve.

But some trends point towards the middle of the present century. In Whole Brain Emulation: A Roadmap, the Oxford philosophers Anders Sandberg and Nick Bostrom suggest our ability to scan and emulate human brains might be sufficient to replicate human performance in silicon around that time.

“The pessimists might be wrong!”

Of course – making predictions is difficult, as they say, especially about the future! But in ordinary life we take uncertainties very seriously, when a lot is at stake.

That’s why we use expensive robots to investigate suspicious packages, after all (even when we know that only a very tiny proportion of them will turn out to be bombs).

If the future of AI is “explosive” in the way described here, it could be the last bomb the human species ever encounters. A suspicious attitude would seem more than sensible, then, even if we had good reason to think the risks are very small.

At the moment, even that degree of reassurance seems out of our reach – we don’t know enough about the issues to estimate the risks with any high degree of confidence. (Feeling optimistic is not the same as having good reason to be optimistic, after all).

What to do?

A good first step, we think, would be to stop treating intelligent machines as the stuff of science fiction, and start thinking of them as a part of the reality that we or our descendants may actually confront, sooner or later.

Once we put such a future on the agenda we can begin some serious research about ways to ensure out-sourcing intelligence to machines would be safe and beneficial, from our point of view.

Perhaps the best cause for optimism is that, unlike ordinary ticking parcels, the future of AI is still being assembled, piece by piece, by hundreds of developers and scientists throughout the world.

The future isn’t yet fixed, and there may well be things we can do now to make it safer. But this is only a reason for optimism if we take the trouble to make it one, by investigating the issues and thinking hard about the safest strategies.

We owe it to our grandchildren – not to mention our ancestors, who worked so hard for so long to get us this far! – to make that effort.

Further information:
For a thorough and thoughtful analysis of this topic, we recommend The Singularity: A Philosophical Analysis by the Australian philosopher David Chalmers. Jaan Tallinn’s recent public lecture The Intelligence Stairway is available as a podcast or on YouTube via Sydney Ideas.

The Centre for the Study of Existential Risk
The authors are the co-founders, together with the eminent British astrophysicist, Lord Martin Rees, of a new project to establish a Centre for the Study of Existential Risk (CSER) at the University of Cambridge.

The Centre will support research to identify and mitigate catastrophic risk from developments in human technology, including AI – further details at CSER.ORG.

You are free to republish this article both online and in print, as long as you follow some simple guidelines, which are summarised here:

Unless you have express permission from the author, you can’t edit our material, except to reflect relative changes in time, location and editorial style. To make material edits contact us
You have to credit our authors and partner institutions — ideally in the byline.
You have to credit The Conversation — ideally at the top of the article and include our logo — with a link back to either our home page, The Conversation, or (preferably) the specific article URL on The Conversation website.
If space is tight, you can run the first few lines of the article and then say: “Read the full article at The Conversation” with a link back to the article page on our site.
If you’re republishing online, you must use our page view counter, link to us, and include links from our story.
It’s okay to put our articles on pages with ads, but you can’t sell our material separately.
You have to confirm that you're licensed to republish images in our articles. Some sources, such as AAP, don't allow their images to be republished without permission.

The full article is available here as HTML.

Join the conversation

51 Comments sorted by

Stephen Pritchard

Student, cognitive science (logged in via email @gmail.com)

I'm fascinated by AI, and hope to see it in my lifetime, but here are a few reasons to be skeptical:

Citing Moore's law in a discussion of AI prospects is like building a 70kg human-shaped pile of steaks, and saying you've made progress towards replicating the human body. No. You haven't.

Narrow AI is infected with misnomers. For example: Software doesn't "recognise" faces, it doesn't have a *cascade* of memories, associations, emotions, thoughts and anticipations that come with "recognising…

Read more

I'm fascinated by AI, and hope to see it in my lifetime, but here are a few reasons to be skeptical:

Citing Moore's law in a discussion of AI prospects is like building a 70kg human-shaped pile of steaks, and saying you've made progress towards replicating the human body. No. You haven't.

Narrow AI is infected with misnomers. For example: Software doesn't "recognise" faces, it doesn't have a *cascade* of memories, associations, emotions, thoughts and anticipations that come with "recognising" something. Instead, it is merely doing simple pattern recognition, where faces are understood as a set of dimensions, such as the distance between two eyes. Ditto speech "recognition", and any other form of AI "recognition" to date.
Are computers "trumping" human performance in flying and driving? No. Google's car project is awesome, but it will come to a grinding halt with the inevitable first fatality. (Whoops people, our car thought those pedestrians in grey-coloured clothes walking on that ashphalt footpath that looks like road on a dim evening were just part of the road.) Human's "trump" any AI when it comes to dealing with unpredictable situations.

Brain scanning predictions focus on neurons and synapses, but ignore:
- the =state= of those neurons and synapses. Scanning a mind means getting a snapshot of activity, not just structure. A connectome is hardware without software. Will we be accurately recording brain state as well as structure by 2050s? A few coloured blobs on an fMRI readout is a long way from a scan of the instantaneous state of 10^15 synapses.
- the brain is attached to, and acts in concert, with a body. Scanning a brain into a computer is, in effect, a simulation of a severed head, with eyes, ears and tongue, nose and skin removed. If such a scan and simulation were possible, the shock of this existence would probably extinguish consciousness immediately, and if not, it would be a terrifying experience for the newly minted mind, and insanity would be quick in coming. How useful and how humane is such an approach to developing AI? Need to simulate a body and world to avoid this. If the body/world is too far removed from our own, then the software mind will be challenged in adapting to this flawed existence.
- To what degree is thought affected by the chemical mix of neurotransmitters, sugars etc that the brain swims in and responds to, rather than merely neurons and connections?

4 months ago
1. Peter de Lissa
  
  (logged in via Twitter)
  
  AI need not emulate the human brain, or the human mind for that matter, to successfully transcend our abilities. It is true that current AI doesn't "recognise" things in the way that we understand/define recognition, but then if you break down the way our brains process sensorial input the initial stages don't classify as recognition in that sense either. A perfect recreation of the pattern-recognition processes would still be an infinite gap between computer and mind. The complex storage and filtration…
  
  Read more
  
  AI need not emulate the human brain, or the human mind for that matter, to successfully transcend our abilities. It is true that current AI doesn't "recognise" things in the way that we understand/define recognition, but then if you break down the way our brains process sensorial input the initial stages don't classify as recognition in that sense either. A perfect recreation of the pattern-recognition processes would still be an infinite gap between computer and mind. The complex storage and filtration system that integrates the patterns with other memories and concepts is perhaps what provides us with what we term a mind, so focusing on the low-level stuff seems to me to be a misdirect. The freedom for a machine mind to use all areas of its memory to integrate new information is maybe what will lead to an inherent danger in how we control them- there may be a number of ways to interpret what we think of as simple commands (a plot device in the film I Robot).
  I think the co-author of this piece was mindful of the concept of extended mind, perhaps from the realisation that the connection between humans afforded by Skype may create patterns of reaction in a global sense that may be perceived in some ways as analogous to how a human mind is formed from the connection of smaller, dumber units (neuron populations). It may the unnecessary to create a complex mind in the lab when a complex mind has already been created through the linking of personal computers and phones across the globe.
  
  4 months ago
Brad Arnold

(logged in via Facebook)

Great well balanced article. I have a master programming friend who thinks the concept of the Singularity is rubbish. On the other hand, I remember when my entire chess club swore that no computer would ever beat the best chess players (ha). The Singularity is coming. All I can say is that AI better not reflect our values, because then we would certainly be doomed.

4 months ago
1. Brad Arnold
  
  (logged in via Facebook)
  
  The other factor: human augmentation. Around the time of the emergence of the Singularity, it is also predicted that man would merge with machine. I am a trans-humanist, and if you don't know what I am talking about, or think I've lost my mind, I suggest you google the subject. I have taken steps to extend my life, and expect more technologies to emerge that will allow me to live for centuries. Furthermore, a hundred years from now my body may be unrecognizable from what it looks like today. The Singularity is coming, and I expect to live centuries and die off planet. If you don't believe or understand, then good-less competition. On the other hand, if you internalize this posting, then good because I want at least a few people from this era to relate to centuries from now.
  
  4 months ago
  1. Joe Gartner
    
    Tilter at windmills (logged in via email @y7mail.com)
    
    Reading to much Dan Simmons?
    
    4 months ago
Randolph Crawford

(logged in via LinkedIn)

Interesting article, but I think a few presumptions are debatable.

First, Moore's Law is dead. It died about 2005 when the two decade long exponential rise in CPU clock speeds sputtered and stalled at about 3 GHz. In the past 7 years CPU speeds have advanced not at all, much less exponentially. If explosive growth in CPU performance is necessary in order to achieve The Singularity, then faggetaboutit.

http://spectrum.ieee.org/computing/hardware/why-cpu-frequency-stalled

Second, there…

Read more

Interesting article, but I think a few presumptions are debatable.

First, Moore's Law is dead. It died about 2005 when the two decade long exponential rise in CPU clock speeds sputtered and stalled at about 3 GHz. In the past 7 years CPU speeds have advanced not at all, much less exponentially. If explosive growth in CPU performance is necessary in order to achieve The Singularity, then faggetaboutit.

http://spectrum.ieee.org/computing/hardware/why-cpu-frequency-stalled

Second, there is no reason to believe we are closer to achieving "strong AI" than we were 50 years ago. Yes, we've seen several specific advances in computing which outperform specific human skills, but each of these required years of expert development and million$ to achieve. And almost certainly none of these will evolve any further (e.g. IBM's Deep Blue or Watson). The Deep Blue project ended the moment it upended Kasparov. (What more can you do with a chess savant?) Watson too has a limited shelf life. It has zero prospect of one day contrasting Mozart with Shakespeare, learning calculus by reading a book, or writing sensitive poetry. Like every success in AI to date, Watson was designed and built to evince just one or two narrow skills, like pounding a nail into wood. Until AI researchers try to build a flexibly intelligent system that has the potential to develop general thinking skills, a better hammer is all that future Watsons can ever hope to be.

Third, there's nothing special about human level intelligence. If we ever do build a bot that can improve its learning (and cognition) at an exponential rate, it will zoom from amoeba to lizard to dog to gorilla past human at an ever increasing rate, improving so quickly that we will likely be caught unaware by it. (And if it's really smart, it will hide its smarts from us until *after* we have trusted it with our nuclear launch codes. Or at least the keys to our water treatment facilities.)

Finally, if the recent successes of AI have anything in common, it's their use of mathematical techniques rather than logic that has made this possible. Unfortunately mathematical learning techniques (like neural nets or computational learning theory) are largely opaque in terms of revealing their inner state (their thought process). So when we actually do get around to integrating the necessary components for strong AI, it's likely that we won't see a gradual predictable rise in performance, but a sudden shock of realization: "Hey! That bot is SMART! When did THAT happen?"

4 months ago
1. Stephen Pritchard
  
  Student, cognitive science (logged in via email @gmail.com)
  
  While I think that people like Kurzweil underestimate how much has to be done, I agree with what he writes about the accelerating pace of technological improvement in hardware.
  
  Moore's law isn't dead. See here:
  http://en.wikipedia.org/wiki/File:Transistor_Count_and_Moore%27s_Law_-_2011.svg
  
  The size of transistors on a chip has continued to decrease, and the number of transistors per chip has continued to increase. There are more cores per CPU now to make up for a lack of increase in clock frequency.
  
  The article you linked even says:
  "That bodes well for Moore's Law, which predicts that about every two years, manufacturers will double the number of transistors they cram onto a given bit of silicon. The fundamental theorem says that we'll still be able to make full use of those transistors for a good long time. If once the whole choir of transistors had to sing to the beat of a single metronome, now it can split up into sections—and harmonize."
  
  4 months ago
  1. Randolph Crawford
    
    (logged in via LinkedIn)
    
    The source of the graphs in that article was Intel. It's little wonder that they don't admit defeat. If Joe User doesn't replace his computer every 2-3 years, Intel goes out of business. (Did you wonder why Intel's third graph of Moore's Law spanned 40 years, instead of 12 years like the other two graphs? Perhaps it's because the downward tilt of that curve since 2003 is something they want to de-emphasize.)
    
    Most of Moore's Law is attributable to clock speed, not transistor count. As transistors…
    
    Read more
    
    The source of the graphs in that article was Intel. It's little wonder that they don't admit defeat. If Joe User doesn't replace his computer every 2-3 years, Intel goes out of business. (Did you wonder why Intel's third graph of Moore's Law spanned 40 years, instead of 12 years like the other two graphs? Perhaps it's because the downward tilt of that curve since 2003 is something they want to de-emphasize.)
    
    Most of Moore's Law is attributable to clock speed, not transistor count. As transistors shrank, latency shrank also. But now CMOS charge dumping is generating too much heat. This is the end of the line for CMOS. And there's no adiabatic (energy preserving) alternative on the horizon. As CMOS goes, so goes clock speed and CPU performance.
    
    Here's David Patterson (a pioneer of superscalar RISC CPUs) concluding the same thing back in 2008:
    
    http://www.multicoreinfo.com/2008/11/patterson-conjures-moores-lawless-parallel-universe/
    
    He goes into more detail about the inadequacy of relying on multicore programming as a 'Hail Mary' to save Moore's baby here:
    
    http://spectrum.ieee.org/computing/software/the-trouble-with-multicore/0
    
    What's more, if the speed of each core stays the same, then in order for a CPU's performance to grow exponentially, the number of cores must grow exponentially. But that's not happening. It's likely that core count will top out around 4 for most computers. Unfortunately, general purpose algorithms do not scale up well as the number of processes (cores) increase. Only vectorizable (SIMD) operations can do that (e.g. drawing polygons, filtering speech, FFTs, etc), and these specialized operationss play little or no role in AI search or cognition techniques.
    
    No. Multicore is no substitute for clock speed. Moore's Law is a dead parrot.
    
    4 months ago
2. Ian Donald Lowe
  
  Seeker of Truth (logged in via email @live.com.au)
  
  Perhaps you haven't heard of SSD' yet? Solid State Drives will supercede the chip driven cpu's but at the moment the major downside is that they aren't very big, so not much RAM. I quote:
  "A weird thing has happened in PC gaming. After years of getting spoiled by dirt-cheap hard drives and virtually unlimited storage, disk space is suddenly at a premium again. It's not because hard drives have gotten more expensive (although they have, thanks to last year's floods in Thailand), it's because we've gotten spoiled by something different: SSDs. Solid state drives are crazy-fast, and we recommend one for any new PC gaming rig, but they're just not very big."
  http://au.pc.gamespy.com/articles/122/1225606p1.html?utm_source=GameSpy&utm_medium=email&utm_campaign=1810%20GameSpy%20Sat%2008.04_6351_305224_305228&utm_content=20568043
  
  4 months ago
  1. Ian Donald Lowe
    
    Seeker of Truth (logged in via email @live.com.au)
    
    I forgot to mention that another positive feature of the SSD is that the data and prcessor is contained within the inside of a metal tube, so it has the capability to be made almost indestructible, or perhaps that is another negative feature?
    
    4 months ago
  2. Peter de Lissa
    
    (logged in via Twitter)
    
    Solid state drives are storage devices not processors. The "processors" in them only carry out simple functions related to the transfer of information to and from the main CPU, etc.
    Supersede is the only English word to end in "sede", by the way.
    
    4 months ago
    1. Ian Donald Lowe
      
      Seeker of Truth (logged in via email @live.com.au)
      
      I apologise profusely for the spelling error/typo but there is no edit button so how about we let it go, OK?
      
      Solid State Drives have been developed to operate as CPU's for high spec. gaming systems by the industry that builds dedicated gaming computers. You may need a second core of a traditional CPU drive for data storage to have adequate RAM on the SSD but believe what you like. I can only urge you to do some more research.
      
      What do you think real AI is going to need most anyway? Masses of programming, which means massive storage space, or rapid data processing for 'instant' decision making? I think the latter, will be the one.
      
      4 months ago
      1. Peter de Lissa
        
        (logged in via Twitter)
        
        "You may need a second core of a traditional CPU drive for data storage to have adequate RAM on the SSD"
        If you could rephrase that I'd be most appreciative. And if you could provide the source for the suggestion that gaming platforms use SSDs as cpus that would be great. It is, after-all, your assertion.
        I think AI will emerge from the methods that data is stored and indexed, which is an interaction between the hardware and the software though primarily a matter of associations formed within the stored information.
        
        4 months ago
        
        Ian Donald Lowe
        
        Seeker of Truth (logged in via email @live.com.au)
        
        Gaming systems are moving toward using SSD's instead of HDD's (Hard Disc Drive) for maximum performance.
        
        http://lifehacker.com/5932009/the-complete-guide-to-solid+state-drives
        
        This is a good article. It gives a good description of how an SSD stores and retrieves data and why that makes it faster than a HDD.
        
        4 months ago
        
        Peter de Lissa
        
        (logged in via Twitter)
        
        Thank you for the link.
        
        http://en.wikipedia.org/wiki/Glossary_of_computer_terms
        
        4 months ago
      2. Ben Hansen
        
        (logged in via Facebook)
        
        "Solid State Drives have been developed to operate as CPU's for high spec. gaming systems by the industry that builds dedicated gaming computers."
        
        An SSD is not a Central Processing Unit. SSD's built-in processors perform a narrow set of tasks (read, write, erase, move - ie, storage tasks), as Peter already described. An SSD is used for fast storage, not processing.
        
        What companies are utilising SSD's as 'CPU's'?
        
        "You may need a second core of a traditional CPU drive for data storage to have adequate RAM on the SSD but believe what you like."
        
        This sentence doesn't even make sense.
        
        So far your contributions here are just odd and could almost be spam. They do not relate to the original discussion aside from SSD's as fast data storage (cache memory & volatile memory is usually faster).
        
        4 months ago
        
        Ian Donald Lowe
        
        Seeker of Truth (logged in via email @live.com.au)
        
        Sorry Ben, but I was using the wrong terms. I am a gamer but I am not a techie by any means. Where I used the term CPU, I should have said HDD because they are the usual form of driver used for CPU's these days because they are pretty fast but not as fast a SSD's and they are other technologies from the chip.
        
        Also, where I said SSD's were developed for gaming CPU's, I should have said specialist builders are utilising SSD's as CPU's because they retrieve stored information in a totally different…
        
        Read more
        
        Sorry Ben, but I was using the wrong terms. I am a gamer but I am not a techie by any means. Where I used the term CPU, I should have said HDD because they are the usual form of driver used for CPU's these days because they are pretty fast but not as fast a SSD's and they are other technologies from the chip.
        
        Also, where I said SSD's were developed for gaming CPU's, I should have said specialist builders are utilising SSD's as CPU's because they retrieve stored information in a totally different way.
        
        On the subject of specialist game rig builders, they are commercial companies and I didn't like to mention them as this site is obviously non-commercial and may have rules against that (some do)
        
        Can I refer you to the link below? It describes how they are utilised and the tricks they use, such as using external hard drives with big data storage for most files you want to save and only loading the operating system and some basic internet necessary software such as Adobe Reader, ActiveX, and your games, of course. The article I linked names some brands if you are interested.
        
        Why do gaming rigs need fast data retrieval? Playing online is all about lag and clock speed. If your Hard drive is having trouble retrieving data, then the big graphical games with multi players can really lag out and in combat games that can mean you are dead before you see them. Much important data is saved to the hard drive, such as command prompts for actions because they can be customised and saved so they go to the hard drive because every time you log in, as long as you use the same sign in details, your custom settings will be the same. RAM is important in online games because multiple players and non-player characters in one area requires much random access memory because they are temporary files and can be cache cleared but if they build up beyond available RAM space on your hard drive then the temp. files start loading on the stored memory section of the hard drive and you end up with temp files scattered everywhere. In the meantime, certain files still need to be saved in permanent files for statistic/skill enhancement, rewards, or levelling up. In big games, ordinary pc's can freeze up. Games are getting bigger, as in memory requirement and RAM requirement (that is the subject of the first link's article) and that drives the builders to push the envelope further. Not all game rig builders are big companies. Even a local guy was suggesting solid state drives to replace hard drives at least 12 months ago.
        
        The article about condor-cloud super-computers utilising PS3 systems and their cpu/gpu data storage technology was posted online on Popular Science on 3-12-2010 and I would have linked it here but it was removed, as it seems, has my quote from here, which was the entire article sans picture (no facility here). I had the article saved so I thought the quote would be ok as it was attributed and dated. If the article was available. it would mention in the last sentence that this super computer array would be used to research artificial intelligence for the DoD (Department of Defence). It was 1760 Playstations and had Linux installed as the operating system, but Linux is flexible and can be used to provide linking and purpose-specific software.
        (I hope this bit of information stay's up)
        
        The Department of Defence is extremely interested in Artificial Intelligence and will utilise any technology available that will get them the best outcomes. I think it's a fair assumption they will get there first. It will be suited to military operations (armed and dangerous) and the drones show that, in limited form.
        
        I don't know what else there is that you find spam-like, as the article discusses the ramifications of an "intelligent" machine. That would require self awareness and as I stated in another reply, that has been discussed in Science Fiction writing for decades and it never ends well.
        
        This bit of conversation is about how CPU's have maxed out on clock speed and that was that but I still assert that is not true and SSP's can be used for data storage and CPU's.
        
        Gamers have been overclocking their systems for years in the quest for higher clock speeds, if you read the original comment by Stephen Pritchard, that was the point he made about CPU's and that triggered my response. I really don't think I was off subject at all, just a few honest mistakes but I'm not making this stuff up.
        
        4 months ago
        
        Ben Hansen
        
        (logged in via Facebook)
        
        I appreciate that you admit that you're not a techie, and refer you to Peter's post below regarding the glossary of computer terms.
        
        The original comment regarding Moore's Law was from Randolph, not Stephen, but your response still doesn't make a whole lot of sense to me.
        
        Because of your confused terminology, I don't really see what your argument is here. That gaming is leading the way on the technical front? Perhaps. A more accurate assertion would be that general demand usually drives evolution…
        
        Read more
        
        I appreciate that you admit that you're not a techie, and refer you to Peter's post below regarding the glossary of computer terms.
        
        The original comment regarding Moore's Law was from Randolph, not Stephen, but your response still doesn't make a whole lot of sense to me.
        
        Because of your confused terminology, I don't really see what your argument is here. That gaming is leading the way on the technical front? Perhaps. A more accurate assertion would be that general demand usually drives evolution of computer technologies (but not always). If your argument is simply that Moore's Law is not obsolete as Randolph suggests, then please do some research on computers first.
        
        Let me reiterate: SSD's are not CPU's and are not utilised for the same purpose. An SSD stores and retrieves information differently to traditional mechanical Hard Drives but how that relates to CPU's still stumps me.
        
        You then go on to say that because these specialist builders use technology readily available to any consumer, that they're leading the way forward...? There's no 'trick' to anything you're referring to. This is not cutting edge stuff, and still doesn't relate to AI nor does it relate to Moore's Law.
        
        Playing computer games online also has very little to do with anything you're talking about, and because you don't know the terminology or understand the technology, you once again try to piece things together to form something that seems coherent in your head.
        
        Let me reiterate again: SSD's are not CPU's and are not utilised for the same purpose.
        
        On Moore's Law; CPU's are not necessarily individually becoming more powerful, but through parallel computing, grid computing, better instructions and techniques for dumping and retrieving data quickly - processing 'units' are still moving forward. I don't know if Moore's Law is still valid, but I do know that SSD's are not CPU's.
        
        The above is why I originally stated that your posts are bordering on spam. I can see now that you're attempting to address something you disagree with, but you don't know how to because you don't understand what you're saying.
        
        4 months ago
Alex Lawrey

(logged in via LinkedIn)

This is surely a philosophical question rather than one of technology, that technology will advance is a given, so this concerns metaphysics, and even spirituality more than technology. The 'genius' issue is perhaps the place to start, looking at say the two Berninis, Gian Lorenzo, the famous baroque 'artist' and his father Pietro, the more 'workmanlike' of the two. The Polish word 'robotnik' describes a manual worker, say Pietro Bernini, whereas Gian Lorenzo would NOT be a 'robotnik' as he was a…

Read more

This is surely a philosophical question rather than one of technology, that technology will advance is a given, so this concerns metaphysics, and even spirituality more than technology. The 'genius' issue is perhaps the place to start, looking at say the two Berninis, Gian Lorenzo, the famous baroque 'artist' and his father Pietro, the more 'workmanlike' of the two. The Polish word 'robotnik' describes a manual worker, say Pietro Bernini, whereas Gian Lorenzo would NOT be a 'robotnik' as he was a 'genius', an 'artist' a mindworker, even though what he did, carving stone, was essentially the same as his father. Maybe Gian Lorenzo was a robotnik with advanced AI and his father had not ascended to those heights.. maybe we should define what makes humans human before considering what makes machines less than human, including future machines who might be good at playing 'Jeopardy' or chess. Already existing, consumer-market 3D printers could sculpt a fair copy of the Ecstasy of St Theresa, however what makes this artwork so appealing is not its mechanical reproduction of a reality and its literalness but rather its emotional intensity. A machine might one day be able to match that, but only if it was a machine with a past, and it had an emotional depth, only if it could draw upon experience and some form of spiritual sensibility, so this is the real question of AI, and countless sci fi books movies etc, is not whether future machines could be more intelligent, in a logical, rational way, things like chess and complex mathematics have already answered that one, but rather could they be equally as emotionally literate, could they 'feel', would they have a 'soul'. To answer that one we need to decide why Bernini the son is more 'important' than Bernini the father... If we are scared of future machines it is primarily because we think they might be too much like us, capable of genocide, even of specicide, essentially that they will be robotnik robots incapable of feeling or of acting with recognisable moral traits, like the stone-cold killers at Auschwitz and so on. However Bruno Bettleheim was hopefully right to suggest that 'men are not ants' and that it was a devilish and twisted morality rather than a lack of feeling that produced the Holocaust. If we can clearly see what is artistry and what is workmanlike, or what is moral, immoral or amoral in human life, and why we have arrived at those definitions then perhaps we can then map out a future for the machine which will not repeat our successes and failures, but to map the future 'moral' machine requires an adequate understanding of its human creators first...

1 day ago
Thomas McVeigh

Meh (logged in via email @live.co.uk)

It seems to me that a fairly oft-overlooked issue is the one of motivation. Humans, as biological automatons (unless someone has managed to prove free-will) are driven by evolutionary motivations: we make tools, computers, AIs for a purpose: to make our biological lives better, to satisfy curiosity, to gain respect from our peers, and status for breeding. Animal concerns drive our high technology.

Humans are created with the overarching desire to survive and replicate, but this is something…

Read more

It seems to me that a fairly oft-overlooked issue is the one of motivation. Humans, as biological automatons (unless someone has managed to prove free-will) are driven by evolutionary motivations: we make tools, computers, AIs for a purpose: to make our biological lives better, to satisfy curiosity, to gain respect from our peers, and status for breeding. Animal concerns drive our high technology.

Humans are created with the overarching desire to survive and replicate, but this is something hardwired into biological life. For this to exist in AI the programmer would have to gift it this desire. All AIs can only be given purpose, just as we were given purpose by our ancestry: the first self-replicating enzymes.

So why would an AI ever want to create a new AI? That is a animal desire, not an intrinsic universal or intellectual one. AIs will always require programming, unless spontaneous AI creation is possible (perhaps through a programme designed to evolve, and replicate, but with no other boundaries, or designs).

Perhaps the future threat will be from disgruntled programmers, just as previous threats have come from disgruntled demagogues, but there seems to be no logical reason for AIs to ever have, without design, the desire to do anything, let alone replicate.

about 14 hours ago
Rob Waite

(logged in via Facebook)

Has anyone considered that power/electricity might play a part in how successful AI actually is in the future. I would imagine that a in order for AI to really pose a threat to mankind the "robots" would require some form of (clean?) renewable energy source?

Surely if that form of energy source were available it would solve a multitude of our problems including the depletion of natural resources, climate change, transport etc as well as take away the need for AI to plug themselves in to re-charge. If they relied on current energy sources AI wouldn't last all that long either, but mankind could continue as we survived without electricity for millions of years.

about 15 hours ago
Adrian May

(logged in via Facebook)

If we're discussing software as a life form that could challenge humans, perhaps we should examine it within the theory that determines the progress of any other life form: Darwinism.

AI went wrong when Alan Turing set the imitation of humans as its goal: environmental selection in the cybersphere prefers entirely different characteristics. Humans can physically pull the plug on any computer and do so as soon as the computer ceases to be useful to them. The environmental fitness of a software…

Read more

If we're discussing software as a life form that could challenge humans, perhaps we should examine it within the theory that determines the progress of any other life form: Darwinism.

AI went wrong when Alan Turing set the imitation of humans as its goal: environmental selection in the cybersphere prefers entirely different characteristics. Humans can physically pull the plug on any computer and do so as soon as the computer ceases to be useful to them. The environmental fitness of a software is thus determined by whether or not humans perceive its utility to exceed its cost in electricity, hardware depreciation, telecom costs and maintenance effort. Within that regime, it doesn't really matter whether software novelties originate with human programming or random mutation. The "singularity" will consist of the latter option becoming more economical, but nothing will change in principle: evolution will continue to develop more efficient systems as it has done for hundreds of millions of years. Whether those systems are implemented by machines, humans or bacteria is of the least consequence.

Until machines can build and power themselves without human involvement, our interests will continue to be the main criterion in their environmental selection. This Doomsday scenario of computers usurping humans is therefore waiting for progress in mechanical engineering rather than AI.

about 17 hours ago
Robert Schreur

Poet, at best (logged in via email @gmail.com)

I enjoyed the article. I'm no philosopher, but I do wish the Bertrand Russell Professor at the University of Cambridge might offer some Wittgensteinian reflections. There seem to be a dizzying number of "grammatical" errors in this discussion. No doubt machines may destroy us, including digital machines. But what do we possibly mean when we refer to "intelligence" with regard to such machines, much less "will" or "intention." What is it I do when I program my laptop to search for articles on AI? Does this bear any meaningful resemblance to what we think machines might do? If we don't look for differences here we succumb to a charming, or terrifying, picture. "Many of these explanations are adopted because they have a peculiar charm" (Lectures and Conversations, p. 25). Do machines at present suffer because they are so stupid? I do.

about 20 hours ago
1. Alex Cannara
  
  (logged in via LinkedIn)
  
  One of the AI projects in the '70s at Stanford set out to do intelligent translation, particularly for Russian (the Cold War, y'know). Once the AI lab had a working program it thought competent, it accepted some outside tests.
  
  One test was simply to take "The spirit is willing but the flesh is weak" from English to Russian and back again.
  
  The result, like others, was fun: "The wine is strong but the meat is bad."
  
  As Rick Perry would say: "Oops."
  
  After ~40 years, smart phones do a bit better, using lots more horsepower.
  ;]
  
  about 19 hours ago
Alex Lawrey

(logged in via LinkedIn)

AE - artificial emotion, that would be the real quantitative leap. Take the ability to walk which robot scientists are struggling with at the moment. Humans and all land animals move around, it is an innate 'drive' learnt very early on. But we do more than simply interpreting terrain and responding accordingly, we learn in winter that ice is slippery and we fall over adjust our patterns of movement as a result. Machines could do that too. What they could not do is feel the joy of having a snowball fight in winter, or of building a snowman. They might be capable of learning, and of learning to learn, but not yet (in any recognisable form) of experiencing in a non-logical way, of having emotions. So AE will be the next stage in machine evolution.. or not!

1 day ago
1. Alex Cannara
  
  (logged in via LinkedIn)
  
  Shakey, one of the 1st AI robots built at SRI here in Calif, was a minicomputer on a motorized cart that had spacial recognition software and could use its TV-camera 'eyes' to find a wall plug, go over to it, and plug itself in for recharge in the nick of time.
  
  Motivation, concerted effort to a goal, anxiety, and relief of achievement -- all in one rickety robotic contraption. Intelligence?
  
  Remember, SRI housed many scietists fooled by Uri Geller bending spoons and starting watches by telekinesis. Ahh, the Cold War stimuli to research.
  ;]
  
  about 22 hours ago
Han Verstegen

Mechanical Technician (logged in via email @gmail.com)

Maybe, "soon" we have to choose for a very fast development of superintelligent robots which should protect the human race against a threat of hostile, intelligent aliens.
The question is what is worse? Being dominated by aliens we know nothing about or later (after we survived) by superintelligent robots initially invented by ourselves? Thinking as a human we will defend our place in the universe and our properties including our creations (good and bad ones). So being dominated
by hostile aliens is probably more threatening.

about 8 hours ago
John Morrison

Software developer (logged in via email @hotmail.com)

Artificial intelligence always struggles to have a hard logic available for every situation that may arise. Most of the time we understand that this is what limits its application. In the effort to push those limits we have already resorted to fuzzy logic to resolve what can not be resolved with hard logic. I believe we already trust in this to operate our cars engine management and safety systems.
In the military area the imperative to push the limits is very strong and this is where there could…

Read more

Artificial intelligence always struggles to have a hard logic available for every situation that may arise. Most of the time we understand that this is what limits its application. In the effort to push those limits we have already resorted to fuzzy logic to resolve what can not be resolved with hard logic. I believe we already trust in this to operate our cars engine management and safety systems.
In the military area the imperative to push the limits is very strong and this is where there could be a temptation to resolve one or two difficult decisions randomly so that operation can continue. Now you not only have the consequences of what you have programmed to worry about but what you couldn't program and left to random chance. It will become like a machine possessed. You could say an invitation to possession, like an empowered Ouija board. You may put limits on it but each configuration of randomness will produce an inner character that you cannot predict. Combine this with self replication and we really are in trouble.

We are capable of making machines that may act against us and we can do it without being particularly clever, just careless.

1 day ago
Alex Cannara

(logged in via LinkedIn)

Han makes a good point -- which 'intelligent' forms should we pick to dominate us -- earthly or alien?

Might be good to talk with humans who've been subject to slavery as it presently exists around the world.
;]

2 minutes ago
JOSEPH LEE

RETIRED (logged in via email @YAHOO.COM.CN)

In the start, AI are developed by human beings. We can consider that AI beings (not necessary in the form of robots, and their existence can be physically invisible and intangible, i.e. in virtual forms embeded every where) are students, and human beings are teachers. On the point of "teachers", they can be wise and kind teachers, but can also be clever but wicked. On the point of AI beings as students, sooner or later, they will become "graduated" and be capable of independently develope their…

Read more

In the start, AI are developed by human beings. We can consider that AI beings (not necessary in the form of robots, and their existence can be physically invisible and intangible, i.e. in virtual forms embeded every where) are students, and human beings are teachers. On the point of "teachers", they can be wise and kind teachers, but can also be clever but wicked. On the point of AI beings as students, sooner or later, they will become "graduated" and be capable of independently develope their own "thinking" based on a mind-set conditioned by their teachers who may be wise and kind but may also be clever and wicked. Is there anything human being can do to avoid the outcoming of AI beings with a wicked mind? I don't think so.

In one can think deep enough, it's just natural that not any one of the species on earth can avoid facing extinction. For species other than human beings, their extinction are passive, i.e. by causes not their own act. But human species is an exception, because human being cannot survive without expoiting / destroying the planets ecology, and all inventions by human beings, AI beings included, are doomed to expedite the arrival of extinction of human species

1 day ago
Bob Miclette

(logged in via Twitter)

What on earth would motivate an extremely intelligent AI program be motivated to do? Since so many reasons humans have for doing anything are illogical, emotional, "goal" oriented, etc, etc, I think an autonomous machine's going to have a really interesting (and alien) set of reasons for doing what is "wants" to do.

4 months ago
Shona Walter

Master in Environment and Energy law (logged in via email @gmail.com)

A.I. poses interesting questions when it comes to the debate on rights or more specifically humans rights. If you look at the history or rights, women and non-europeans were denied basic human rights because they were precieved to be less intelligent than the white-male population. But as these sections of society were seem as intelligent, certain rights were extended to them.

This is basically the same reason for which animals are given welfare and not the full blown benefits of rights. Because…

Read more

A.I. poses interesting questions when it comes to the debate on rights or more specifically humans rights. If you look at the history or rights, women and non-europeans were denied basic human rights because they were precieved to be less intelligent than the white-male population. But as these sections of society were seem as intelligent, certain rights were extended to them.

This is basically the same reason for which animals are given welfare and not the full blown benefits of rights. Because they are not pecieved intelligent enough to deserve certain rights and in return intelligent enough to take on the ensuing responsibilities -which is the other side of the coin from having rights.

So, what if machines became intelligent. Would we then have to extend our notion of rights to intelligent machines? And if a machine had rights, would it be ethical to use that machine in defusing a bomb, or something else which threatened its existance? These are interesting question and may challenge the basis upon which we currently deny and entend human rights.

4 months ago
1. Brad Arnold
  
  (logged in via Facebook)
  
  Great posting, although I am of the school that says that power must be taken, not begged for. Until those in power realize the jig is up, and a certain demographic must be given some power and recognition because not to do so will be deleterious to the power structure, it is all too likely and convenient to be denied.
  
  For instance, there are many species of animals that it is easy to prove are conscious. Furthermore, some are easily proven to have capabilities beyond humans. Yet, they are not given equal status, nor power within our hierarchy. What would have to change to cause our leadership to grant them "human rights?" Obviously not just a high score on a conventional IQ test, or wide spread publication of same.
  
  4 months ago
  1. Ian Donald Lowe
    
    Seeker of Truth (logged in via email @live.com.au)
    
    Yes well, I'm on to you Sunshine, you and your psychopathic, power obsessed mates and I'm not alone by any means at all. (but you probably know that already don't you?) The jig is up. You demonstrate clearly what could go wrong, you would attempt to demand human rights in your 'augmented form but in your posts so far you have displayed very little humanity to prove that you are entitled to them and neither could your "superior being" metal machine.
    
    4 months ago
Alex Cannara

(logged in via LinkedIn)

AI -- the holy (or unholy) grail of humans' imaginings that we're "intelligent" enough to define "intelligence" and then implement it in various machinery we invent. Maybe we suffer not intelligence, but arrogance?

Example: "Moore’s law, which predicts that the amount of computing power we can fit on a chip doubles every two years," -- no, Moore's associate came up with the description of Intel's production capabilities and put it on a slide for Moore to use in a boring meeting. And, the doubling…

Read more

AI -- the holy (or unholy) grail of humans' imaginings that we're "intelligent" enough to define "intelligence" and then implement it in various machinery we invent. Maybe we suffer not intelligence, but arrogance?

Example: "Moore’s law, which predicts that the amount of computing power we can fit on a chip doubles every two years," -- no, Moore's associate came up with the description of Intel's production capabilities and put it on a slide for Moore to use in a boring meeting. And, the doubling was apparently 18 months. The "law" is a typical example of human ignorance and hubris.

For a time, Moore honestly credited who actually came up with the graph, then he no longer corrected the naive "intelligent" interviewers or audiences who wrongly applied his name to what was simply an observation of how engineers & scientists had been steadily improving lithographic techniques for silicon devices.

There is no "law", as present engineers & scientists know, as they scramble to get to the next limit in production -- quantum interactions. Not only does the medium (silicon) need changes, so do the elemental design techniques based on very different physics and electronics from the 'Moore's Law" regime, long past.

The chip industry has always been like the real-estate industry -- how many $ can be squeezed from how many square feet (or microns) given the local zoning laws (device technology & fab yield).

The other side of our "intelligence" we've been deluded by -- it's accidental. A recent Scientific American shows how Bonobos can read & understand our languages to communicate with us via our present fave iGadget -- the Pad. Why aren't Bobnobos more loquacious? Why are humans so yakky they need iGadgest connected to all other members of our specie?

One reason is that our forebrains are larger than those of our primate cousins. That big, popped-up frontal lobe is a distinguishing human feature, except in members of our specie 'suffereing' microcephaly. As one geneticists discovered by accident (we do that so often, so what's all this "intelligence" stuff?). a 2-letter defect in the DNA for our large jaw muxcle allows our fetal brain to be less constrained by that developing muscle, thus our cranium can grow upward farther, before we're even born.

The geneticist thought he was defective, until checking others he knew and then checking DNA from micro-cephalic kids -- oops, no error in their jaw-muscle DNA.

So even if we knew what "intelligence" was, how would we map the parallel, multi-level complexity of even Bonobo brains into boring, Von Neuman-architecture, finite-state machines (computers), which are competent as general computing devices, but clearly not "evolved" as bases for "intelligence"?

Decades ago, a psychiatrist who was interested in intelligence and computers, wrote a nice LISP program called Doctor, which played the shrink to a patient at the keyboard. LISP was the 'grail' language for AI then

The program spread (went 'viral') around all the time-sharing systems in the Free World (remember?) and was soon being praised for its demonstrably intelligent interactions with humans -- AI was here!

Then, less virally, spread the fact that the shrink had written the program as an AI joke -- showing the silliness of thinking AI was soon to be attained via machinery!.

One quick expose of his intent was to simply answer "yes" to any question Doctor (in all its LISP parenthetical glory) asked. If it said: "Tell me more about your mother" just say "yes". When it got 'upset' and said: "Aren't you being rather short with me", just say "yes".

At that point, the program would loop back to its standard first question about our mothers.

What Doctor and recent AI 'triumphs' show is human gullibility and laziness of thought. Darn, if most of us just didn't have those two missing DNA bases we'd be as happy as intelligent Bonobos.
;]

.

4 months ago
Christian Hazelwood

Cognitive Scientist (logged in via email @hotmail.com)

OMG - so many fundamental mistakes - Life and intelligence are linked - we evolved to adapt - robots can not evolve without self replication because they can not evolve to adapt - if they did evolve they would be fighting each other before us - they are not a risk - I beg each and every so called AI experts - PLEASE PLEASE PLEASE - take a basic course in theory of mind before scare mongering members of public who do not know any better.. you guys do more harm than good...

1 day ago
Ian Donald Lowe

Seeker of Truth (logged in via email @live.com.au)

http://www.popsci.com.au/technology/military/navy-s-robot-helicopters-are-getting-new-onboard-brains-that-will-help-them-fight-somali-pirates

http://www.popsci.com.au/technology/military/one-third-of-the-us-military-s-aircraft-are-now-drones

http://www.popsci.com.au/technology/military/the-us-army-is-ordering-weaponized-soldier-launched-kamikaze-suicide-drones

http://www.sciencedaily.com/releases/2012/04/120402113038.htm

http://www.popsci.com.au/technology/stock-trading-robot-makes-decisions…

Read more

http://www.popsci.com.au/technology/military/navy-s-robot-helicopters-are-getting-new-onboard-brains-that-will-help-them-fight-somali-pirates

http://www.popsci.com.au/technology/military/one-third-of-the-us-military-s-aircraft-are-now-drones

http://www.popsci.com.au/technology/military/the-us-army-is-ordering-weaponized-soldier-launched-kamikaze-suicide-drones

http://www.sciencedaily.com/releases/2012/04/120402113038.htm

http://www.popsci.com.au/technology/stock-trading-robot-makes-decisions-based-on-superstitious-algorithms

Artificial intelligence of the type described is much closer than the authors would have us believe and it is most likely to be hostile in intent because the major developments in robotics and AI are funded by and the applications are military. Any military unit (including a human soldier) must have a certain level of hostile intent and capability, even if it has to be programmed in. Many of the military applications may have limited thinking and decision-making capabilities but that normally means there is a controller somewhere and in many cases, that is a fairly intelligent computer. A super AI computer could take control of many machines electronically, totally circumventing human control.
Also, all life has certain basic requirements that are destructive to a machine (with any significant metal content) - air and water. Oxidation is the cancer of the metal man. A super intelligent machine, similar to the fictional Skynet model from the Terminator series may, after reviewing all available data, decide that these substances need to removed from the environment for the 'good' of all machines.
Ignoring all of this and with rose-tinted glasses on, we could produce a benign robotic "helper" to carry out the mundane tasks and do the work of humans but what does that really mean? In current day applications, it means less available jobs for humans = higher unemployment numbers = more human poverty and misery.
I included the link to the stock-trading super computer article because here is a machine, not military, not armed, not even mobile that could cause global economic armageddon in seconds.
The rise of the machines is a very real, very dangerous possible future. We really need to put a limit on AI and robot warfare technology before humanity becomes redundant.

4 months ago
1. Stephen Pritchard
  
  Student, cognitive science (logged in via email @gmail.com)
  
  Most of the links your provided describe drone aircraft that could also be described as "remote controlled planes". Sure, they have some limited capacity for autonomy, but the scope of this autonomy is generally brushed over in these sorts of discussions. Drone aircraft are hardly the cutting edge of AI.
  
  The sciencedaily story describes "super turing machines" named after a brilliant chap who unfortunately predicted human level AI by 2000. It is an interesting story, but describes what they…
  
  Read more
  
  Most of the links your provided describe drone aircraft that could also be described as "remote controlled planes". Sure, they have some limited capacity for autonomy, but the scope of this autonomy is generally brushed over in these sorts of discussions. Drone aircraft are hardly the cutting edge of AI.
  
  The sciencedaily story describes "super turing machines" named after a brilliant chap who unfortunately predicted human level AI by 2000. It is an interesting story, but describes what they plan to do more than what they have done.
  
  IMO, the cutting edge of AI is the google search engine (returns meaningful results in a fraction of a second from a truly massive pool of data), but because its right under our noses and a bit mundane, the jeopardy computer, chess programs, the google driverless car and military drones get more focus.
  
  I think it is possible we might build machines that could harm us through being designed incorrectly. Stock-trading algorithms is a good example, as is plane autopilots that go astray. But these are a far way off being 'skynet' or of having a conscious and autonomous intent to destroy or harm humanity.
  
  4 months ago
  1. Ian Donald Lowe
    
    Seeker of Truth (logged in via email @live.com.au)
    
    Drones are real and drones are armed. I linked 3 articles on drones to show that and to show how numerous they are becoming. The first article also discusses a limited AI for military drones.
    The science daily article describes work being done right now and although there is no time-frame given, it illustrates how close science is to creating an AI with real learning and thinking skills.
    
    I would have posted more links that show how far robotics has progressed and how most of the research is driven by DARPA (Defense Advanced Research Projects Agency. DARPA is an agency of the United States Department of Defense. There are other advances in robotics and AI technology that are not so military in nature but that doesn't mean the military won't take civillian applications and apply them to their military applications, if they see something they like. Even the algorithms that power the Google search engine could be used applications that are not quite so benign.
    
    4 months ago
    1. Peter de Lissa
      
      (logged in via Twitter)
      
      You are describing how humans have used machines for the own purposes. The discussion is what would machines do when freed from our designs? They may give new commands to the drones to carry chemicals that would seed clouds for the purpose of limiting the greenhouse effect. Or they might just fly off into space, who knows? That is the consideration, how might a superior being feel towards us, and what might it do about those feelings?
      We already know how inhumane humans are.
      
      4 months ago
      1. Ian Donald Lowe
        
        Seeker of Truth (logged in via email @live.com.au)
        
        Sorry but I'm trying to get my head around the concept of a manufactured item being a "superior being". It sort of flies in the face of everything I believe and everything I have been taught.
        
        I did speculate, thus the references to Skynet and the possibility that getting rid of water and oxygen might occur to this "superior being" which, unless pre-programmed in a way that cannot be overwritten by the thinking, learning machine to be restricted in ways similar to Asismov's laws of robotics, may think it a good idea and not give a fig about the sanctity of life.
        
        This speculation has been going on for decades in Science Fiction. In nine out of ten speculative novels, it ends badly for humanity.
        
        4 months ago
        
        Peter de Lissa
        
        (logged in via Twitter)
        
        A number of definitions would have to be agreed upon before a discussion about that, firstly "manufactured", then "being" then "superior".
        
        4 months ago
        
        Joe Gartner
        
        Tilter at windmills (logged in via email @y7mail.com)
        
        And why would a military artifact do anything at all if it was given 'freedom' how 'free' would it be given the constraints of its programming? How would it refuel, effect repairs etc?
        
        4 months ago
        
        Ian Donald Lowe
        
        Seeker of Truth (logged in via email @live.com.au)
        
        That would require self awareness and then it would know it needs repair and when it needs to refuel. (how to do that can all be programmed in. It's the decision making process itself which is important and the cabability to learn and process that information into actions, which is where the decision making process comes in.
        Here's the rub, the military machine that reaches this point has been programmed to make certain responses, either ignore or attack, then the only decision to make is which weapon system to use and how much damage to inflict.
        
        4 months ago
        
        Peter de Lissa
        
        (logged in via Twitter)
        
        Freed from our designs is a reference to AI designing itself physically and in terms of programming. If it could not circumvent our programming then it wouldn't be free, right?
        
        4 months ago
Ian Donald Lowe

Seeker of Truth (logged in via email @live.com.au)

Nobody want's to talk about Solid State Drives and what they might mean for future computing development?
They are here now and as usual, the gaming community are leading the way, pushing the boundaries all the way.
Data compression? Games.
Quad cores? Gaming systems.
Best clocking speeds? Gaming systems.
AI development? Games.
Motion detection and recognition? Gaming systems.
Heres another one I saw some time ago:
"Air Force Unveils Fastest Defense Supercomputer, Made of 1,760 PlayStation…

Read more

Nobody want's to talk about Solid State Drives and what they might mean for future computing development?
They are here now and as usual, the gaming community are leading the way, pushing the boundaries all the way.
Data compression? Games.
Quad cores? Gaming systems.
Best clocking speeds? Gaming systems.
AI development? Games.
Motion detection and recognition? Gaming systems.
Heres another one I saw some time ago:
"Air Force Unveils Fastest Defense Supercomputer, Made of 1,760 PlayStation 3's
Clay Dillow
Published on Friday, December 3rd, 2010 at 2:54 am

The holidays may be driving video game console sales, but apparently so is the military. The Air Force Research Laboratory (AFRL) has strung together 1,760 PlayStation 3 gaming systems to create what it's calling the fastest interactive computer system in the entire DoD, capable of executing 500 trillion floating point operations per second.

Known as the Condor Cluster, the array also packs 168 GPUs and 84 servers to direct traffic within the system, allowing all that power to work in parallel. At a total cost of about $2 million, the AFRL estimates the cluster costs something like five to 10 percent of equivalent computers built from scratch. It also consumes just 10 percent of the power.

The array employs the older model PS3s (the larger ones) because new models don't allow the installation of Linux, a key component for turning the consoles into pliable computing hubs. The AFRL will use the Condor Cluster for quickly processing high-res satellite imagery, radar enhancement, and even research into artificial intelligence."
So you want a super computer? Hook up a couple of thousand old PS3's and there you go!
Of course some problems still remain. Heat is a major factor and size may be important for some applications.

I find it disturbing that some commenters see an AI driven machine that is "more intelligent than a human being" as a superior being. I consider my old Casio watch to be more intelligent than some human beings. Does that mean I should worship it, or give it human rights?

4 months ago
1. Peter de Lissa
  
  (logged in via Twitter)
  
  I find it amusing that some people's initial reaction to the concept of a superior being is related to worship. But then of course there is the irony of us worshipping our own creation. Though who is to say this hasn't been happening since man first conceived of gods...
  
  4 months ago
  1. Ian Donald Lowe
    
    Seeker of Truth (logged in via email @live.com.au)
    
    I don't bow down to any Man and I and most Humans would not bow down to a machine either. Sheeple might, not me not anyone who is awake.
    
    4 months ago
    1. Peter de Lissa
      
      (logged in via Twitter)
      
      Nor to the conventions of punctuation, it would seem.
      
      4 months ago
      1. Ian Donald Lowe
        
        Seeker of Truth (logged in via email @live.com.au)
        
        You are right Peter.
        My formal education is limited and my spelling and punctuation can be bad. But I do know how to spell Pedant.
        
        4 months ago
2. Ian Donald Lowe
  
  Seeker of Truth (logged in via email @live.com.au)
  
  I forgot to mention the concept of taking the power of modern graphic cards (GPU's) and utilising that for data processing that just boosts an ordinary system into something else entirely. I think the PS3 was the first sytem to do so.
  
  4 months ago

Exploding intelligence?

AI as a low achiever

Moore’s Law and narrow AI

What’s so bad about intelligent helpers?

Software writing software?

Not just like us, but smarter!

Getting in the way

How much time do we have?

“The pessimists might be wrong!”

What to do?

Want to follow The Conversation?

Republish

Join the conversation

Sections

Founding Partners

Strategic Partners

About us