A more modern look at androids

[Spaz]

I want to start a discussion on a more modern take on what we think a fembot could be like.

First, when I first started on this wiki, tablet and mobile technology was either nonexistent or primitive in comparison. Now, we have more reliable wireless technology and rapidly evolving tablet and smartphone devices. So, now we can conceive of a fembot with little or no maintenance panels because you could now outsource most of their functions to a tablet device. This would permit more space for more essential components, such as power sources, processing hardware, or other enhancements to make them look and feel more realistic, such as artificial muscles.

Next, we can take a look at the attitudes towards androids. More specifically, the negative attitude that they can be easily programmed, reprogrammed, or easily malfunction.

I would argue that those negatives can be easily equated to human beings. It is possible to program a human being: you send them to school, have them read certain books, have them watch certain television shows or films, take them to church. The only difference between programming a human and a robot is the former takes considerably longer. Then, you have reprogramming, which can also be done to a human with rigorous conditioning, torture, or brainwashing. Again, it merely takes longer than it would for an android. Finally, you have malfunctions. When a human receives significant trauma to their brain, they can often display symptoms that can mimic a robot malfunctioning. One might even be able to discuss the reverse, that in a world populated entirely by A.I.s, that a malfunction could be considered to be as common as a stroke or seizure.

In any event, these are among things I have been considering while developing my more recent stories. Many authors on this site still seem to be trapped in the past in regards to what an android would be built like given that it were even possible to create one. I dealt with the attitude issues in more story A.I.L.A., and I have dealt with the outsourcing of monitoring to a tablet in an upcoming story of mine.

If we could build a perfectly realistic android with an A.I. that could realistically pass for a human as well, what kind of past conventions might we also have to discard or improve upon given our current range of technology and knowledge of how the human brain works?

[dale coba]

An android has no inherent character, and can be programmed to switch in an instant from being a gentle care-giver to an engineered sociopath, a machine that brings all its superhuman skills together to kill. How exactly does that compare with people? People who ever actually existed in the real world?

The only difference between programming a human and a robot is the former takes considerably longer. Then, you have reprogramming, which can also be done to a human with rigorous conditioning, torture, or brainwashing. Again, it merely takes longer than it would for an android.

Your "only difference" [respectfully, because you know I dig you] is a ridiculous stretch. You attempt to brush aside an insurmountably different difference.

- Dale Coba

[--NightBattery--]

according to the national geographic in the bathroom, the brain is not like a computer, but like the web, with many computers interwined using their computing power for different stuff, so i believe an android would require basically a CPU for almost any basic function in order to immitate a human correctly,a CPU to filter visual information, other for audition, other for balance, other for effectory response, other for descipher human actions, other for appropiate responses and so and so, and even so, it would be very defficient and slow to do it so.
but charmingly fake.
also, an android would require to convene with a super computer oftenly to analyze what it is actually doing and learn complex thing for real and not by default.
to obtain a crude philosophy.

also and a little offtopic, about current materials...
penalty free energy actions must be discarded because battery technology is not that good right now, and one or two are not enough for inhuman heavy work or inhuman leaps and powerful thinking without getting dry and extremely hot in a sec.
also
carbon nanotube fibers are a reality right now so it is possibly "logical" to replace the "metalic things" with them.

the current information transfer velocity and the materials for it could be a problem too.
but i don't know a lot about that, but seems rather intuitive to assume.

Regards.

[darkbutflashy]

Computer comparisions are always fake because we simply *can't* put a CPU into a robot that works the same way the human brain does. It just won't work because it's impossible to make the hardware tiny enough to fit into the frame and even if we fix that problem by remote control, we won't be able to build and power more than 5 such robots (to misquote a famous computer scienticist).

Giving figures: the human brain's neocortex has 1.000.000 separate columns ("CPUs") each consisting of roughly 60.000 nervous cells. Each cell has dozens of weighted interconnections and memory for each of it, making one cell having roughly the computing equivalent of at least 100 transistors (it may be much more, depending on the modeling). That makes one column have at least the computing power of five i486 CPUs (maybe 10 or 20), and the neocortex alone has one million of it. A year 2014 high-end processor has about 300 of such 6.000.000 transistor-units. How much of them do fit into the head module of a gynoid? With cooling measures, I think about 10.

[smalk]

Leaving out technical limitations (just remember that Moore's law hold up even when it was theoretical impossible to reduce further eletrical components' size) I think it would be very inaccurate to compare the behavior a biological intelligence with an artificial intelligence.

In my opinion, it would be like trying to find similarities between horses and cars: sure, they are both means of transportation, but that's it.

Ok, of course you could program a digital intelligence to simualate an human behavior, at any degree of complexity. But in the end it's just a simulation: viz not the natural way for the intelligence to express himself.

@battery: as @darkbutflashy pointed out, you can imagine the human brain as your "web" of "cpus" (as every part of the brain is dedicated to a specific function). Actually the collective intelligence is a very promising research field, where a collection of small computational entities strongly connected are more successfull than a single computational entity (with computational power bigger that the sum of the small ones).

Sadly, I have to admit with @spaz the future disappereance of maintenance panel - with heavy sadness on my part.

Who I'm kidding, I'll just install some panels on my gynoid anyway. Who cares the warranty

[Spaz]

Alright, alright.

Now I ask, what about the development of organic computing? It is my understanding from my brother that it would not have the limitations of our current computing technology.

[smalk]

Are you referring to the vision of information processing inspired by biology? There are good insight there (like swarm intelligence), the main problem is to allow a fast and cheap communication between all the nodes (aka cpu or cores). In fact that's where we are going in the microeletrical industry, with n-core cpu as the first step. To reach a computational power able to simulate the human you only need a number of cores in the number of millions strongly connected. This poses a whole new set of engineering problems.

To actually evade current computing limitations (organic computing is mainly a vision on how you interconnect computational cores) you can look to quantum computing. In theory they could execute machine learning and image processing (two requisite for a "real-world" intelligence) with astonishing velocity. Almost an year ago D-Wave released a 512 qubit computer: only 10 million dollar for a unique one (wonder who bought that? xD). Yep, that's a lot, but you know, today we all have in our pockets computational entities more powerful than the one used by Nasa to send a men on the moon.

[darkbutflashy]

Sadly, I have to admit with @spaz the future disappereance of maintenance panel - with heavy sadness on my part.

No, no, maintenance panels would be everywhere, the mechanics and power supply doesn't go away.

[Spaz]

Sadly, I have to admit with @spaz the future disappereance of maintenance panel - with heavy sadness on my part.

No, no, maintenance panels would be everywhere, the mechanics and power supply doesn't go away.

I'm not saying the maintenance panels would vanish, more that they would change. Instead of having bulky diagnostic display monitors inside them, you would have better access to the mechanics and power supplies. Perhaps there would be fewer panels, such as no upper chest panel and just one large chest panel.

[darkbutflashy]

Now I ask, what about the development of organic computing? It is my understanding from my brother that it would not have the limitations of our current computing technology.

You mean, nervous cells connected to a circuit? This should work, and it's possible to refine the technology while working on prothesises e.g. for severed limbs. Having an understanding test subject is extremly helpful. But for the same reason, I don't see the "gynoid with the cat brain" coming too soon.

[smalk]

HINT: We're barely out the door.

When I first majored in informatics I was convinced of the clear superiority of the machine over the flesh.

Now, every day I spend working in machine learning research just reminds me how powerful is the computational entity known as the human brain.

[dale coba]

I dunno about your doubts, guys... quantum computing seems very near.
Quantum computing sounds like the ultimate brute force tool - at least nearly as powerful as an organic, neural network.

Am I still missing something?

- Dale Coba

[--NightBattery--]

HINT: We're barely out the door.

When I first majored in informatics I was convinced of the clear superiority of the machine over the flesh.

Now, every day I spend working in machine learning research just reminds me how powerful is the computational entity known as the human brain.

love this song

[DollSpace]

I do to, --Battery--, I do, too...

[N6688]

I myself like the combination of old shool tech (gears and computerchips) mixed with the new tech (nanotechnology) like the T-X from terminator 3 (i know, the movie sucks)
This way you have Always acces crittical systems and have a seamless body at the same time (and the shapeshifting capability is a nice extra if you get bored of the old form )
But nowadays you see less and less of old school androids on the tv and in other media.
Now it's bio engineerd humans like the skinjobs in battlestar galactica,the synthetics in the aliens movies, or cell in DBZ (witch in my eyes aren't machines at all so there is nothing atractive about them to me)
Or beings that are completely built out of microscopic machines like the T-1000,the replicators in stargate, or like the human replica droids in star wars.
My point is i like the old school way but it is sadly slowly disappearing.

[D.Olivaw]

I dunno about your doubts, guys... quantum computing seems very near.

Sadly probably not, at least for anything other than some very specific crypto applications.

Quantum computing sounds like the ultimate brute force tool - at least nearly as powerful as an organic, neural network.

Am I still missing something?

I note a lot of people (not you, Dale, you're just asking an honest question) waving quantum computing like some sort of magic wand that gives them the solution to whatever problem they want to solve with it. If it can be made to work, it's very powerful for some things; very weak for others. Certain algorithms can be run faster, even very much faster, than with classical computing. These algorithms are of very great importance for many scientific, mathematical, and cryptanalysis problems. There are many algorithms for which they're not a huge improvement over what we can do now. They're also tremendously touchy: it's very difficult to prevent decoherence for even a few dozen qubits, much less the vast quantities you'd need for general computation purposes.

I'd put my money on optronics or spintronics, followed maybe by molecular electronics as the next paradigms in general computing.

All the usual caveats about predicting the future apply, of course

[D.Olivaw]

But nowadays you see less and less of old school androids on the tv and in other media.
Now it's bio engineerd humans like the skinjobs in battlestar galactica,the synthetics in the aliens movies, or cell in DBZ (witch in my eyes aren't machines at all so there is nothing atractive about them to me)
Or beings that are completely built out of microscopic machines like the T-1000,the replicators in stargate, or like the human replica droids in star wars.
My point is i like the old school way but it is sadly slowly disappearing.

Yeah, I've noticed this too. There's no scientific or technical reason for it; I think it's mainly just that no one wants to do "just another robot," they want something to jazz it up. She's not a boring old robot, you see, she's a nano-glorbitronic biobot! So Fresh and New!

I think the bio-engineered humanoids are popular because then you can have a synthetic character, and all the related plots, without paying for expensive special effects or having to deal with too much techno-babble. Think how much cheaper Almost Human would have been if the "androids" were genetically engineered synthetic humans instead of robots.

I like old school too, and if realistic androids are created in the plausible midfuture, they'll almost certainly be heavy on the electronics and light on the nano-whatsits.

[D.Olivaw]

"Organic computing" seems like a vague term to me, and I don't think I've ever really heard that specific term before.

Me neither. There's DNA computing, which is showing some major promise at least for certain applications. There's the idea of wiring up biological neurons to silicon transistor systems, which is interesting but not terribly far along in actual experimental work. Very difficult, then, to make predictions about where it could go.

But it does remind me of neural nets, which are ways of setting up a networked computer system so that the system can "learn". I don't have much of a grasp on the concept, but I did see a very interesting article just yesterday that should remind us all just how far along we are down this particular technological road:

http://www.i-programmer.info/news/105-a ... l-net.html

HINT: We're barely out the door.

That was a tremendously interesting article, Robotman; thanks for bringing it up. The paper it's based on is pretty dynamite, and suggests some interesting emergent phenomena that link together many different approaches to neural-net based learning systems.

[D.Olivaw]

Leaving out technical limitations I think it would be very inaccurate to compare the behavior a biological intelligence with an artificial intelligence.

Exactly, especially for Turing machine AIs.

just remember that Moore's law hold up even when it was theoretical impossible to reduce further electrical components' size

Moore's Law started quitting about three years ago, especially if you look at what's available in the consumer market. Large-scale computing (supercomputers, etc.) have gotten around that (for now) via economies of scale and using certain technologies that aren't easily ported to the general computing market. That said, this isn't necessarily anything to worry about. The natural progression of technological advance is a series of ogive curves (see: The Strategy of Technology, Possony and Pournelle). That is, the technology sees a period of exponential performance improvement followed by a leveling off (that's an ogive, or "s" curve). After a while, there's a major innovation (or not) and the technology sees another exponential burst of improvement before leveling off again.

The question is, how long will the level part of the curve last for computing? I would guess not terribly long; there's a lot of new technologies waiting in the wings with the potential for another decade or so of exponential improvement. Of course, the question becomes more complicated when you ask how long the level part of the curve will last for the computers consumers can buy in the store (fembots would be included in this category). Some of the techs that promise to extend computational performance improvement are very delicate and don't like to be moved. Or even have a truck drive by on the street 1/4 mile away (quantum). Others might be even more resilient than current systems (optronics). We'll just have to wait and see.

[smalk]

Now it's bio engineerd humans like the skinjobs in battlestar galactica,the synthetics in the aliens movies, or cell in DBZ (witch in my eyes aren't machines at all so there is nothing atractive about them to me).

Let's put this problem in a purely engineering way. You have a goal: to obtain an agent, able to accomplish a series of task generally performed by humans in the physical world, able to understand your commands, able to learn and evolve (in the limits you choose). All of this in a fixed amount of time.

The wise choice would be to improve the most similar thing already existing in nature, stable, well-known and studied. Here, a biological human, improving its resilience and limitating its awareness and self-esteem.

The other choice would be to bet on a completely new framework (here a mechanical construct), with all the risks and the technological limitations you could get.

This is all intellectual speculation, with no reference at all to my sexual preference (as it is the topic's argument).

[dale coba]

I note a lot of people (not you, Dale, you're just asking an honest question) waving quantum computing like some sort of magic wand that gives them the solution to whatever problem they want to solve with it.

Unconditional quantum teleportation between distant solid-state quantum bits

Dan, where does today's news fit in?

Scientists in the Netherlands have achieved a breakthrough in quantum teleportation that could quash Albert Einstein’s objection to the notion of quantum entanglement, which he famously labelled “spooky action at a distance”.

Publishing their results in the journal Science this Thursday, physicists at the Kavli Institute of Nanoscience at Delft University say they managed to reliably teleport quantum information between two bits of diamond located three metres apart.

Although this is not the same process as teleportation as imagined in science fiction, one of the lead authors of the paper Professor Ronald Hanson, said that it was possible that Star Trek-style “beaming up” would become a reality in the future.

“What we are teleporting is the state of a particle,'' said Professor Hanson. ''If you believe we are nothing more than a collection of atoms strung together in a particular way, then in principle it should be possible to teleport ourselves from one place to another.”

“In practice it's extremely unlikely, but to say it can never work is very dangerous," Professor Hanson added. "I would not rule it out because there's no fundamental law of physics preventing it. If it ever does happen it will be far in the future.”

An image 40 micrometers wide showing two qubits with a 'teleportation beam' and electron spin state superimposed.

However, this is not to say that the discovery of Professor Hanson and his team won’t have startling effects in the present, with accurate teleportation of quantum information a key step towards building quantum networks exponentially more powerful and secure than today’s supercomputers.

Ready to join my Skynet resistance force yet?

- Dale Coba

[D.Olivaw]

That's a very interesting paper, Dale, and certainly a step forward for quantum computing, but completely tangential to all the points I raised. Mainly, this is because being able to link qubits together is very good, but only if you have the qubits in the first place. It's a little like developing network protocols before being able to manufacture transistors.

The news report, however, was abominable. The author's words are mostly in "not even wrong" territory, and I can only presume that they got Dr. Hanson really drunk before they interviewed him because ''If you believe we are nothing more than a collection of atoms strung together in a particular way, then in principle it should be possible to teleport ourselves from one place to another" and "I would not rule it out because there's no fundamental law of physics preventing it" is misleading at best. Heisenbergian uncertainty rules out large scale material teleportation right off the bat, even ignoring the many mere implausibilities that would have to be overcome in the meantime.

[xodar]

by smalk » Thu May 29, 2014 4:34 pm

"....similarities between horses and cars: sure, they are both means of transportation, but that's it."


Right. Horses are so much more than that, which is a use or function of a short time in their existence. Horses' purpose, if any, is to be horses.

[Spaz]

I just came up with something during my boring shift at work today, something I am integrating into my ever evolving idea of a fembot, which will debut in one of my upcoming works.

De-centralized brain. Like an Octopus.

So, I start out with an organic brain-like computer in the head. It stores all the memories, it has the whole emotional center, higher reasoning, blah blah blah. Then, there is a second brain in the head which processes all the audio, visual, smell, and taste senses.

In the torso, there is a larger secondary brain which processes movement, touch, and all involuntary bodily functions below the neck. In each hand, arm, leg, and foot, there is another smaller brain which processes every touch and movement of the particular limb, all linking up to the torso brain.

All the brains link to another brain in the head, which houses the A.I., which processes all the sensory information and memories. This brain is attached to the organic brain, and the organic brain can be transferred from one body to another. Because the motor functions are in another brain, the organic brain doesn't need to be as large as a human brain, and is considerably smaller.

Thoughts?

[--NightBattery--]

earthworms have giant motor neurons.

neurons could be 3d printed with a tissue printer to be tougher and bigger than real ones.

you could take inspiration in the chinese acupunture points for a way to distribute the artificial nervous system.

action potentials can be generated by (microgenerators/or insert cool technological name here) to force neurons to accept imput from the artificial sensors.

Regards spaz.