Do Androids dream of electric sheep?

Discussion in 'Intelligence & Machines' started by Stryder, Sep 18, 2001.

Thread Status:
Not open for further replies.
  1. Stryder Keeper of "good" ideas. Valued Senior Member

    Okay so this was the original title to "Bladerunner" but P.K.Dick with his title did bring forwards a serious question, Would an artificial intelligence or Android actually dream? and what would they dream of?

    I can give reasons why they should, Take for instance a major case of bad SRAM caused by a computer being left on continuously for about a year. It's caused the memory to fail.

    So in the designs of an Artificial Intelligence/Android or positronic brain (like you trekkies love to hear) you would have to create a form of Dream state, a method of perging your systems boards with some activity that cleans all it's chips of being used in just one area.

    Afterall the natural reason for dream when we sleep is a mixture of allowing somewhere for all the stimuli that is used to make you move, talk, walk and do what you do during the day. Of course those stimuli are routed into an area of your brain which isn't directly linked to your body, this allows you to do things within your sleep, but your body is detatched.
    (except for certain ailments like sleep walking)

    Also during the day you will be using particular neural pathways structured for certain tasks, which if ran for a long period would to suffer like that SRAM, and your very fibre would be ruffed up like something has been stroking it the wrong way. (This is because your body and brain has happened to form with particular polar alignments.)

    So when you "turn off" your machine, in reality, it's power should be routed and the system allowed to run at low level with a fractal memory saver (working a bit like a screen saver, but actually within the boards, saving your memory from going bad through static)

    If this was the case, your system could possibly dream of anything, especially if the following statement is true:
  2. Google AdSense Guest Advertisement

    to hide all adverts.
  3. Cris In search of Immortality Valued Senior Member


    Ha ha, no, electronic brains won’t dream or sleep.

    While we think of human brains as mostly electrical pulses between neurons in fact the connections between neurons, the synapses, are chemical. The synapses consume protein which must be replaced after a few hours, hence the purpose of sleep and dreaming.

    Synapses are the target sites for hallucinogenic drugs, e.g. such drugs can generate dreamlike sensations. Other drugs stimulate the synapses and others repress them. When the synapses are being renewed they undergo changes and it is these changes that cause the unusual sensations we know as dreams.

    Since purely electronic brains will have no dependence on biological synapses then they will not sleep or dream, or have any such needs. Humans who eventually go through uploading to non-biological brains will also have no need to sleep or dream.

    Just consider electronics brains and AI machines as simply on all the time in the same way as you are on during the day. They would never become tired or ever become sleepy.

  4. Google AdSense Guest Advertisement

    to hide all adverts.
  5. kmguru Staff Member

    Our dog dreams. We can tell, when he barks while still fully asleep...

    If animals dream, why not Androids. Ask Robin Williams....
  6. Google AdSense Guest Advertisement

    to hide all adverts.
  7. Stryder Keeper of "good" ideas. Valued Senior Member

    Cris, I would say you are slightly wrong about the systems...

    I mentioned my case of Bad SRAM, it's cause was through something similar to what you mentioned about a biological system.
    If I place a current through something metalic that conducts for a long period of time, then the material will eventually suffer a degenerative effect due to the current causing a molecular re-alignment of polarised particals.

    I noticed this from the use of Carbon dating within Archeology and other sciences. When something is created, it's molecular configuration is dictated to by its location and orientation in relationship to the poles (North and South)

    This is why people sometimes have problems when they upgrade RAM/SRAM/DRAM because where one board might have been made facing South in Korea (19/12/1999), another chip might have been facing North West in Malaysia (21/02/2000).

    This means they have completely different molecular configurations, and cause defects within communication between themselves and the process. (Taiwan,05/05/1998)

    (By the way dates and places have been faked just to give you an idea)

    This means the computer would need time to sleep, Of course most people would say "Just turn it off to stop the degradiation through power supply", but this doesn't correct the system any.

    My explaination was that you would have to continue a power supply high enough to transverse the boards, chips etc and low enough not to degrade. This would allow everything to try an harmonise.

    That was my explaination for allowing a system to dream just on the notion of harmonising it's components. Although I would love this to stretch to artifical intelligence, so as allow the number of grammatic erros within such systems not to be added to through board degeneration or molecular alignment.
  8. Bebelina Feminazi Messiah Valued Senior Member


    I´m sure that by the time we have the knowledge how to create human like androids, that they will also have the capability of dreaming. To regenerate their brains, of course!

    And who knows, maybe our computers already are dreaming?

    Please Register or Log in to view the hidden image!

  9. Cris In search of Immortality Valued Senior Member

    OK. I think we are confusing what we mean by dreaming.

    1). The biological function of synaptic protein replacement that causes dream activity, often known as REM sleep. This is a major design failure of the human form since it requires the individual to enter an involuntary coma periodically and hence renders that individual largely defenseless during that period.

    2). The conscious action of fanaticizing, or the use of creative imagination.

    What are you guys talking about?

    If a memory chips fails or degrades then that is a fault and the broken component should be replaced. In human terms it would be seen as an injury to the brain requiring surgery.

    Dreaming, during sleep, (1) from above, is a biological design failure and I see no reason to build that fault into AI systems of the future or include it for uploaded post-humans.

    Fantasies from (2) above, should be encouraged because that activity stimulates human inventiveness and progress.

  10. kmguru Staff Member

    major design failure? Not so fast Cris...

    Even todays computer, the nightly load occurs from an ERP system to datawarehouse system. Not only that, the system runs at a reduced capacity while maintenance is being carried out. Same for manufacturing, Car tuneup/lube job/oil change.

    It is the information repair, reorganization and maintenance function that you call design failure? Then we would have been long dead, eaten up by the carnivores.
  11. Stryder Keeper of "good" ideas. Valued Senior Member

    I agree with Kmguru that Dreams aren't a design failure, Afterall the reason why we dream is it's a pseudo-loop, we dream of moving a limb but in reality are limb does not, and it's reason is for the the allowance of a dream state, where our action are occuring virtually rather than physically.

    Afterall if we panicked at a scarey nightmare and then tried to run away, we would find ourselves falling out of bed at the first sign of trouble. Or worse "bed wetting"

    Please Register or Log in to view the hidden image!

    As wel the coma state also stops us from injuring ourselves, otherwise you'd wake up with a mark or scar and swear you'd been abducted by UFO's, with the scar as your proof.

    I can see that Kmguru also got my understanding of how things run at a lower capacity to give it a chance to recaliberate and repair.
  12. Cris In search of Immortality Valued Senior Member

    Kmguru, Stryder,

    Ha ha, at least I’m keeping your attention.

    KM, remember that a typical CPU is around 100,000,000 times less powerful than a human brain, so the comparisons you are making are extremely tenuous. Also you are describing somewhat simplistic and relatively primitive conventional computer systems.

    I help design mission critical fault-tolerant systems. These run 365x24x7, and they are designed so that they never go down or need to be stopped for maintenance. All components have active duplicates such that if any component, large or small, fails, then the system continues uninterrupted. The failed component is then replaced and re-integrated to provide continued fault-tolerance. All such changes are transparent to the users. The NASDAQ, NYSE, and in fact all significant stock exchanges in the world use these systems, plus 911 services, ATM networks, every significant bank in the world, etc etc.

    Sorry for the commercial, although I didn’t use any brand names.

    But my point is that systems can be designed where they don’t have to be taken down at any time, and these are still relatively primitive systems. And that includes online database reorganization, online disk partition splits, etc, etc. and all transparent to the users.

    The brain has evolved and has allowed us to survive despite our need to enter a coma every day. Fortunately we can be woken quite quickly if needed and I am sure that many early people were indeed eaten by carnivores when they slept. This danger became very apparent and hence the reason why humans and most other animals take shelter at the times when they need to sleep. In your home do you lock your doors and windows at night? Would you necessarily do that if could stay awake all the time?

    We are clearly vulnerable when we sleep.

    The design failure comes from the waste of a 1/3 of our lifetime being spent in a coma. That is very inefficient.

    Stryder, When you enter REM sleep your brain is slightly altered – the connection between the motor sections of the brain is blocked to prevent you from physically moving during the dream state, otherwise indeed you would be thrashing around all the time while you dreamt. However, there is a well-documented disorder where that feature is disabled and such people do in fact play out quite violent scenarios while they dream. And they often injure themselves and their partners.

    Those parts of the brain that enable sleep and REM sleep (disabling the motor functions) can also be triggered by specific brain injuries. This you would recognize as a true coma. But even when signals from the brain indicate that sleep is over, the signals are blocked and the patient remains in a motionless coma. The brain desperately tries to repair itself and eventually the patient wakes, well sometimes.

    Finally. Dreaming is just the way that evolution has enabled our biological brains to re-fuel themselves (e.g. replaced the used protein in the synapses).

    A better design would be where the proteins could be replaced continuously so that dreaming and sleep would not be required. The issue is one of a biological limitation and any advanced computer systems with AI would not have any need for sleep, or to dream, or to ever go off-line. Any reasonably advanced fault-tolerant architecture will ensure that.

    Does that help
  13. Stryder Keeper of "good" ideas. Valued Senior Member

    Here's something just to add more to what I was trying to explain also,
    with all the systems out there, there are processor timeouts, idles and lulls that could allow the processor to be used to do a "dream" task, something pseudo, perhaps running a fake system that hackers would fall foul of etc.

    Who knows, as for artificial intelligence and dreaming, I'm sure that a dream state would be useful, a period of time to calculate data (like people think of problems while sleeping, i.e. "I'll sleep on it"), also the chance to back data up, take for instance:

    If you continually prodded an individual at set times during sleep like every hour, the person would suffer certain trauma upon their sleeping patterns, one of those is they suffer memory failure.
    (Their brain doesn't store information correctly, this is pretty much seen with people suffering insomnia.)

    so again a system would work similar by taking back-ups during idles, or securing data as read only etc.

    Heres something else I mention, The human brain (and indeed other creatures too) is like a giant game of cat and mouse. You might remember the game from schoolyard days or some group activity, where one person is a cat, and another is a mouse, and the rest of the people act like a maze that changes, they have their arms out and are distanced apart within a grid shape, and when someone passes, they turn 90 degrees, with their arms out in both left and right directions, the idea is the maze keeps changing shape and the cat is trying to capture the mouse as it runs around.

    This is pretty much an explaination that the mind is an Adaptive system, with each day connections from neurons are severed, and new connections grow. The reason for connections being severed is that the mind needs to have activity following certain paths to maintain their integraty, to remove that activity the integraty would be a bit like an untravelled road, where grass starts growing down the centre, eventually that road will become so overgrown that you can't travel down it, and this is where your neural connection is severed.

    Of course that path, might find away around the overgrowth, but it's now a different shape than it once was and possibly intersects with another road or junction (neural cluster)

    You might say that we shouldn't sleep, it's not ergonomic, but sleep isn't just for our mind, it's also for our eyes.
    You should realise that our eyes are organs, and at that they are delicate and irreplacible (Other than new cybernetic developments which is still testing future technology) When they are bombarded by Light and even frequencies that we don't register, they become damaged, because their proteins can be destroyed (Cataracts etc).

    Which means we eventually would have to close our eyes or go blind, and since we wouldn't be able to see, we might as well sleep.
    something else that could occur too through continued usage at optimum level would be "overheating", since in the theory of evolution the reason why we stand upon our legs as bipeds, is because our bodies had less surface area contaminated by the suns radiation, namely if we were on all fours and the sun rose up to it's position of mid-day we would be feeling the full force of its heat, while standing upright, lessens the temperature.

    This is particular reasoned for our brain temperature too, or we would probably pas out from heat exhaustion. (Of course this is mentioned like the radiator in a car, or the ears on elephants)
    Of course Heat would again cause problems with memory, and could distort certain processor usage. (after all a semi-conductor isn't running like a super-conductor under a controlled cool temperature, below room temperature.)

    Of course this moves away from the whole reason for dreaming, but it explains the logic for having dreams as our system has to "re-route" things and have the ability to deal with Heating, structuring without losing memory. So it's all a juggling act.
    Last edited: Sep 22, 2001
  14. kmguru Staff Member

    Cris: You got it all wrong my friend.

    I was also involved (as a VP) in a company that sold Fault-Proof (not fault-tolerant) systems. The problem with man made gadgets is that they are not self healing and self repairing systems. To fly a plane, we sometimes need simulators. All what-if scenario need virtual simulations. Yes, I can design a system, that has a separate processor that can simulate and provide input. But you do not have two heads. Nature is frugal. So it uses dreamtime for the reapir of the body and simulation of whatif scenarios so that you can make correct decisions in life.

    BTW, during dream state, the motor functions are disconnected by the brain. That is part of the complex dream sequence.

    You have to analyze human system from a stand point of "Complex System Dynamics" (that is my forte). Then you will understand what I am talking about.
    Last edited: Sep 23, 2001
  15. Stryder Keeper of "good" ideas. Valued Senior Member

    I came across another post that mentioned of Someone that predicted of Pieces of code (Artificial Life) creating other pieces of code as offspring, and allowing them to develop.

    If one grown code didn't cope too well to it's creation it would then die out leaving only the healthy strains.

    I know that had already been messed with, as that was the reason why computers had their first virii makers. The virii weren't made to infect computers but to try and allow people to understand polymorphic algorythms.

    There is also a note that the dream state allows us to close our eyes and replenish with proteins, Without closing our eyes our optical nerves would suffer considerable damage from radiation we can and can not see. (Such as the breaking down of protein)

    If we kept our eyes open for too long we could be blinded, this is most nticable if your up late and your eyes begin to sting. (This should be a warning to tell you to sleep)

    Okay at present you mention that systems don't need to sleep/dream, but you could look at it like this:

    You might have a asystem that you only want to operate while a user is there, so when the user is away it sleeps, no processes will activate actual peripherals they will just run in a simulation pseudo-one.

    Also it is noted that there are ideas of using certain algea's and fungi to generate memory blocks within machines, these would need possibly some period of time to "sleep" to regenerate from damage caused by energy transference (information accesses).

    Of course these bio-computers aren't too well established as of yet.
  16. whatsherface imaginary entity Registered Senior Member

    Hi all,

    I think computers wouldn't need to dream, unless we can invent them smart enough to doubt reality, as against believe in it?

    Dreams are necessitated by our split consciousness, not by REM sleep. There is evidence to support the idea that our conscious minds make them up as we wake, eg REM Sleep = Dreaming: Only a Dream

    How would we know the difference?
  17. Rick Valued Senior Member

    According to recent study by M.I.T, any electronic device used for a long period REGARDLESS of all the enviornmental effects like heating,mechanical abrasions etc,the electronic devices will still degrade if used for a long period!!!...any signs of conciousness???
  18. scifred Registered Member

    I'm smart...i talk!
  19. kmguru Staff Member

    A few thoughts...

    When you use a new razor blade, if you use once a day, it will last say 30 uses. If you use it continuosly on people, it will crap out after 15 uses. It needs rest to regenerate. Metal fatigue...

    Genetic algorithms are programs that branch out with different solutions and those solutions compete against best outcomes. The worst of the bunch is dropped.

    A simulation by any name is identical to the real thing except the loss is minimized or prevented. A simulated nuclear explosion is still a nuclear explosion in someones dream (man or machine)
  20. Rick Valued Senior Member

    (I dont know which thread this concerns)
    well what is the difference between humans and machines,
    we are machines with great software part,design of a great kind(Genetic algorithms as we put it)
    and machines?they also are starting to become intelligent...the level of their intelligence depends upon the programmer,and HIS intelligence isnt it?

    Please Register or Log in to view the hidden image!

  21. kmguru Staff Member

    An older reality has the time to evolve to a maturity. Then the older reality creates a new reality...and so it goes ad infinitum. That may be the nature of the creation itself.

    In our life time, if we copy the 30,000 DNA and modify it and create an android, in our image, are we then the God to it? If we create a virtual human inside a hyper computer, what are we to it/him/her?
  22. Rick Valued Senior Member

    Hi,Is your thread in ANYWAY inspired by SPACE ODESSEY TWO(2061)?,just curious,because i remember DR.chandra referring to SAL9000 that he"ll have to go on a sleep mode,wherein he"ll have dreams,as every intelligent being dreams...
  23. rakeyj Registered Senior Member


    Alot of you guys seem to be missing a vital point here, now i realise alot of you are more knowledgable than me on this subject but your reasoning seems very flawed. Now i don't believe that there would be any reason why anyone would design a AI being with a dream state. The main function of an AI being would be its power to run continuously. The need for humans to dream is a biological one, as has been rightly pointed out this will not be so for AI beings. They will not need time to back up data make hypotheses on hypothetical situations like humans. This will be able to be done in a matter of nanoseconds. We must remember that we are still some years off a proficient AI being, I am not saying that soon there will not be primitive versions of AI beings but not to the extent as those in films like bladerunner. Also i must add that this time influence and the current rate of increase in terms of speed and memory capacity of computers will mean by the time that AI beings which will have the design qualifications to be able to be given dream states will only be done so for designer models, where they are not there for functional reasons but more for moralistic and social investigations. If an AI being needed time for repair of degradation of it's parts then I would expect that it would have the built in facilities that this can be done not while it is in a state of unnecessary inactivity but in a state where it can as quickly as possible return to its daily activity.
    If AIs were given a dream state what would be the effect on their owners. An AI would be scrutinized on its dreams, for an example an AI being is designed to look after children, a task which could be well suited to a being which would not run out of patience and could be of a good moral influence on the children at all times. Now what if the being was to dream and hypothesize about possible scenarios, what if it dreamed about not looking after the children. This would be a grave fact for the owner, that the idea of a being could hypothesize about not doing its job, whether or not this affected its job is not a matter of interest, the fact of the matter is that the owner would lose trust. The reason as i said earlier for the creation of AI beings other than the fact of boosting man's ego and increasing the scepticism of people believing that playing God is a dangerous game, is to fulfill a job which requires no tiredness, no loss of concentration and other factors associated with many peoples view of an AI being. The ability to dream would only hinder this. Unless, and i believe this to be unlikely, that is was truly necessary no dream state would be created for anything other than designer models of AI beings.
Thread Status:
Not open for further replies.

Share This Page