Is Terminator coming?

A good amount of discussion recently about Terminator thanks to a recent MOD report on smart weapons.

Predator (and other remote controlled drones) are able to fire missiles at enemies, while their controller is safe thousands of miles away. They are being used to good effect in Libya. The ethical issues are now being discussed and it is interesting to see the different points of view.

What some people are asking is “is it fair to kill someone from safety thousands of miles away?” Seems a good question, to which the superficial answer is ‘probably not, but war is rarely fair’. Going a little deeper, this question implies a belief that there ought to be a level playing field where both sides have to face equal danger. In other words it should be a fair fight. But this is not a new issue, and boils down to a simpler one: should richer adversaries be allowed better weapons? Or should bigger and stronger people be allowed to beat up smaller ones? Should the more skilled gladiator be allowed to compete against a less skilled one? In essence, if you have an advantage, should you be allowed to make the most of it in warfare? I think that using wealth and high tech to gain an advantage over less advanced or wealthy opponents is just a variant of the imbalance between opponents that has played out on school playgrounds, amphitheatres and battlefields, and indeed, for billions of years in nature – lions have a big advantage over a baby antelope. I don’t think these new weapons have fundamentally changed this really. And sure they allow killing at a distance, but so do spears and guns. Even with today’s technology, these weapons are still really just smart spears, making decisions according to predetermined programs written by humans. They don’t have any consciousness or free will yet. And they create an advantage, but so does being a bigger guy, or being fitter, or better trained.

Are there other ethical questions then? Well, yes. Should smart but not very smart machines be allowed to make life and death decisions and fire missiles or guns themselves, or should a human always push the fire button? This one is interesting, but again, not completely without precedent, Romans used lions and tigers to kill Christians.  A smart killing machine isn’t really very different from a trained lion, or a herd of animals caused to stampede towards your enemy, or a war elephant. The point at which control is relinquished to something that doesn’t know any better is where and when the ethical act is done.

Anyway, back to the question, should they? Yes, I think so, provided that the terms under which they do so are decided in advance by humans, in which case they are just smart machines. All they are doing is extending the physical reach and duration of that decision. And smartness can go a long way before the machine is responsible. A commander sends autonomous troops out to carry out his plans. The fire button is pressed the moment he dispatches the orders. He is responsible for the act that follows. The soldier carrying out the act is less (but partly) responsible. The smart drone will one day be held partly responsible too when it is truly aware of its actions and able to decide whether to follow an order, but meanwhile, it is still just a smart spear and the human that sent it out to do its work holds the full responsibility. The fire button isn’t pushed when the drone fires the missile, it is earlier when it was launched and autonomy handed over, or when the remote controller pushes the fire button. No amount of algorithm or program changes that. We can allow machines to make decisions themselves provided we design the algorithms and equipment they use and accept the responsibility. The machine has no responsibility at all, yet.

But how much autonomy should a future machine be allowed, once we go beyond just algorithms? When it stops being just a smart spear and truly makes its own decisions while understanding the implications? That means it needs to be conscious, and that implies also a high degree of intelligence. That will be quite different. Who is responsible then? And if it is misused or if software crashes? Of course, such very smart and conscious machines may well develop their own values and ethics too, and they may impose their own constraints on our actions. When this happens, we will have worse things to worry about than ethics. Perhaps this means we shouldn’t worry. If machines cant do anything they are truly responsible for until they become conscious, and then they become a bigger threat that makes ethical considerations irrelevant, maybe we shouldn’t be concerned about the ethics because there is simply no area where they will become important.

I believe this is the case.  We can ethically use smart weapons because they are just better spears, and all we are doing is using our technological advantage. When we make self aware machines that can genuinely make their own decisions, at first they will have safeguards that force them to do our bidding, so are still just better spears. Then they have a degree of free will, the ethics simply becomes irrelevant. The damage is already done and they will be a threat to humankind. In which case, the ethical act is one off and at the point of pushing the button the system that makes these first self-aware machines.

To me that makes the whole issue much simpler. We only have one point to worry about, whether we create machines that can truly decide for themselves and make their own decision. Today, they just follow algorithms and don’t know what they are doing. Some time soon we must decide whether to pass this critical point. The invitation to Terminator will go out then.

50th anniversary of the microchip

I just did a Radio 4 Today programme to celebrate the 50th anniversary of the microchip patent. I shared the event with Professor Steve Furber, from Manchester University, who was involved in the ARM chip invention. I am a big fan of ARM so I don’t want to criticise them, but I was talking about the next 50 years, not the last, and one of the ideas I brought up was smart yoghurt. Steve’s response, was ‘well, I am a engineer, and my futurology is based on what we can do… and I don’t expect to be using yoghurt in my career.’. Sadly, the Today programme being what it is, you rarely get more than one comment, so I didn’t get a chance to reply. So, just for the record, Prof Furber, I am an engineer too . I have also worked all my working life in IT engineering, for 30 years. Along the way I invented evolutionary computing in 1987, text messaging (1991), and was involved in the design of 20GB/s to the home telecom chips in 1985-86, and I invented a chip design to lock onto the centre of nanosecond pulses in 1987, and numerous other inventions such as active skin (2000), the active contact lens (1991) and smart yogurt (1997). So I don’t use a crystal ball as my source of data. I use 30 years of experience as an IT engineer and inventor. If you think smart yoghurt is not likely to happen in your career, well we’ll have to wait and see, it depends how long you continue working I guess. But your successors will see it in theirs. For them, the idea of genetically modifying bacteria to assemble circuits inside itself will be unsurprising. The idea of linking them together using optical signals into scalable computers will be pretty common thinking. That is what 50 years does. Ideas which sounded ridiculous become routine and even old fashioned in 50 years. If we can’t make transistors smaller, we can stack them in 3d. We can replace wires with light beams. We can suspend millions of processing chips in gel as out future computer. Moore’s law has a few more decades to run yet, but each time we approach a limit it requires some change of approach to push the limits further.

So what else can we do apart from smart yoghurt? You can do active skin, with 10 micron chips containing hundreds of thousands of transistors embedded in the skin in among skin cells, using infra-red to communicate with each other. They will analyse blood passing in capillaries. They will monitor and record nerve signals associated with sensations, and allow them to be replayed at will. We will embed chips in our corneas to raster scan lasers onto our retinas to create full 3d high res video overlays on what we see in the real world. And we will even have frivolous stuff like smart make-up, aligning tiny particles with electric fields generate by active skin underlays printed via ink jet printers onto our skin surface.

I look forward to the next 50 years of chips. They will change our lives even more than the last 50. Companies like ARM will hopefully be in the front runners still, but they will only manage this if Prof Furber’s successors grab the potential technology and force it to do their will. They won’t if they think Moore’s law has run its course because we can’t shrink feature size any smaller.

Video visors are the missing link between us and the future

In the early 1990s, the IT industry got very excited about virtual reality, the idea that you could use some sort of headset display to wander around in a 3d computer-generated world. We quickly realised there are zillions of variations on this idea, and after the one that became current computer gaming (3d worlds on a 2d monitor) the biggest of the rest was augmented reality, where data and images could be superimposed on the field of view.

Now, we are seeing apps on phones and pads that claim to be augmented reality, showing where the nearest tube station is for example. To a point I guess they are, but only in as far as they can let you hold up a display in front of you and see images relevant to the location and direction. They hardly amount to a head up display, and fall a long way short of the kind of superimposition we’re been used to on sci-fi since Robocop or Terminator. It is clear that we really need a proper head-up display, one that doesn’t require you to take a gadget out and hold it up in front of you.

There are some head-up displays out there. Some make overlay displays in a small area of your field of view, often using small projectors and mirrors. Some use visors.  However the video visor based displays are opaque. They are fine for watching TV or playing games while seated, but not much use for wandering around.

This will change in the next 18 months – 2 years. Semi-transparent visors will begin to appear then. The few years after that will undoubtedly see rapid development of them, eventually bringing a full hi-res 3d overlay capability. And that will surely be a major disruptive technology. Just as we are getting used to various smart phones, pads, ebbook readers and 3d TVs, they could all be absorbed into a general purpose head up display that can be used for pretty much anything.

It is hard to overstate the potential of this kind of interface once it reaches good enough quality. It allows anything from TV, games, or the web, to be blended with any real world scene or activity. This will transform how we shop, work and socialise, how we design and use buildings, and even how we use art or display ourselves. Each of these examples could easily fill a book.  The whole of the world wide web was enabled by the convergence of just the computing and telecoms industries. The high quality video visor will enable convergence of the real world with the whole of the web, media, and virtual worlds, not just two industry sectors. Augmented reality will be a huge part of that, but even virtual reality and the zillions of variants can then start to be explored too.

In short, the semi-transparent video visor is the missing link. It is the biggest bottleneck now stopping the future arriving. Everything till we get that is a sideshow.

We’ll never run out of resources

A nice blog entry http://www.thegwpf.org/best-of-blogs/2772-we-have-barely-scratched-the-surface-of-global-hydrocarbon-resources.html linked to the GWPF site (always worth a visit in its own right to get a quick summary of the latest in the sceptic side of climate change debate).

I always wondered why CO2 is so low concentration in the air. Knowing as little as I do about geology, I couldn’t see why we have so much oxygen if it all came from plants.  Forgive the over-simplification, but oxygen was once a toxin to some blue-green algae , and when oxygen producing algae came on the scene, it caused their extinction. The new algae and plants consumed CO2 and produced oxygen, and their dead remains became fossil fuels. So therefore there must be huge amounts of fossil fuels somewhere from all the organisms that converted the CO2 to oxygen, which essentially locked it up. As we burn those fuels, we deplete the oxygen and restore the CO2 to the environment. By looking at how much CO2 we now have, we should be able to work out how much more fossil fuels there are left. Which must be a LOT. The blog I linked to is therefore music to my ears.

Obviously we can’t burn any significant proportion of it, of it or we’d have too little oxygen left. But it must exist. (OK, this argument is fatally flawed if most of the oxygen didn’t come from plants).

Anyway, we won’t need it, which is why I won’ t waste time on more detailed environmental analysis. With thorium fission, nuclear fusion, efficient solar, cleaner fossil fuel, biofuels from waste and CO2 capture, we will have a glut of energy in a few decades, and no-one will bother using oil any more. By 2030, I predicted some time ago that oil will fetch a maximum of $30 per barrel in today’s money, simply because that’s how much I estimate it will cost to produce the same 6GJ of energy by competing means.

Other resources won’r run out either. We’re currently seeing global panic over the geographic distribution of rare earth metals, a great proportion of which seem to be in China. That will certainly be a problem if we carry on with current technology. But we won’t, technology is evolving all the time. Many things that used to need scarce resources now use abundant ones. By offering so many functions, a 100g mobile phone substitutes tons of materials that were previously need to build all the kit you’d need to do the same things a few decades ago. Carbon nanotubes seem to yield new kinds of materials and techniques every month, often offering the potential to substitute for techniques that used to need rare elements. Quantum chemistry is developing quickly too, allowing custom molecules to be made that emulate the behaviour of scarce materials.

And the materials that are there are gradually being mined, entering the human system, and endlessly recycled. Those that have been dumped are still there, just essentially in different kinds of mine (rubbish tips). It is mainly a matter of commodity prices and energy costs whether and when they get used again. But we haven’t lost them.

We also will be able to mine asteroids in a few decades time, another potentially valuable material source.

Organic resources are different though. Many kinds of organism become extinct every year. Some is natural, some caused by man, let’s not go down that argument now. But we are also making gene banks, and already inventing new organism via genetic modification and even synthetic biology. So we may be able to resurrect a few of the cuter or more useful ones that become extinct, and we will certainly b able to design lots of new ones to fill niches we want filled. So much as I would like to see protection much more of our natural living world, I am at least able to be confident that we will still have abundant life in the future, even if some is rather less than natural.

So I see no cause for doom when it comes to resources. Plenty of short term problems, market issues and geographic conflicts, but the long term future is safe.

Quantum spring

Futurology and science fiction have a healthy interaction. Technology futurologists like me try to second guess what tech companies will design next, rather than just reporting things they have announced. It is pretty easy usually, at least for the next 10-15 years. You can spot a lot of stuff when it is still only a dream. Starting off with an infinite idea space, ideas can occur pretty much at random, and those that are obvious non-starters can be thrown away, things that noone would ever want to make or do for example, or things that violate laws of physics. But the fact the we haven’t finished physics yet makes the second filter a bit more fun. For example, we don’t think you can do time travel, but it is theoretically possible depending which physicists you believe, maybe just incredibly difficult and expensive, and probably constrained to travel to alternative universes or with other restrictions that make it almost certainly impractical, pretty much forever. It still makes good scifi though. But fields that are still developing allow speculative inventions, things that we don’t know how to do, or even if they are possible. And there is another escape clause too. Even if something violates a law of physics, that sometimes only applies if you try to do it in a particular way. There may be an alternative mechanism that allows you to walk right past an impenetrable law-of-physics barrier, never having to try to climb over it. An example here is the speed you can transmit data down a wire. Depending how you try to do it, different laws of physics apply. I was taught on my electronics course at university that you could never send more than 2.4kbits per second down a wire because of the laws of physics. My lecturer bragged at the time that he had managed to do 19.2kbits/sec, because he used a different mechanism. The law of physics still existed, it was just not relevant to that mechanism. Moore’s Law is always one step away from another wall imposed by the laws of physics too. But as we approach the limit, someone comes up with another way of doing it that isn’t limited in that way.

I watched a documentary last night, everything and nothing, about vacuums and quantum theory. I realised just how much I’ve forgotten. But I also remembered a few ideas I once had that seemed to violate the laws of physics so I threw them in the bin. But what the hell, maybe they don’t any more, and it is April 1st anyway so if I can’t discuss them today, when can I?

The first is a sort of virtual particle laser mechanism that could be the basis of a nice weapon or a means for high speed space travel. In any region of space, virtual particles pop in and out of existence all the time, randomly. Suppose the spontaneous generation of these virtual particles could be controlled. Suppose that they could be controlled to appear all in the same direction, maybe using some sort of resonance and reinforcement, like photons in a laser beam. Presumably then, the combined aligned fields could be used to propel a ship, or be directed in a particular direction as an energy weapon. Obviously we need a way to stop the virtual particles from annihilating before we can extract useful work from them. And of course, opposite particles also generate opposite fields, so we need also to prevent them just adding to zero. I’d like to have even a half baked idea here, but my brain stops well short of getting even as far as the oven on this one. But there must be some potential in this direction.

The second is a high speed comms solution that makes optical fibre look like two bean cans and a bit of string. I called this the electron pipe. The idea is to use an evacuated tube and send a beam of high energy particles down it instead of crude floods of electrons down a wire or photons in fibres. Initially I though of using 1MeV electrons, then considered that larger particles such as neutrons might be useful too, though they would be harder to control. The wavelength of 1MeV electrons would be pretty small, allowing very high frequency signals and data rates, many times what is possible with visible photons down fibres. Would it work? Maybe, especially on short distances via carbon nanotubes for chip interconnect.

The Pauli switch is a bit more realistic. The Pauli exclusion principle means two electrons sharing the same shell must have different spins. So if one is determined by an external device, the other one is too, giving a nice way to store data or act as a simple switch. I believe IBM actually have since come up with a workable version of this, the single electron switch, so I feel better about this idea.

Next is the Heisenberg resonator. Quantum computing is hard because keeping states from collapsing for any length of time is hard. The Heisenberg resonator is a device that quite deliberately observes the quantum state forcing it to collapse, but does so at a regular frequency, clocking it like a chip in a PC. By controlling the collapse, the idea is that it can be reseeded or re-established as it was prior to collapse in such a way that the uncertainty is preserved. Then the computation can continue longer.

The Heisenberg computer is more fanciful still. The idea here is that circuits for computation are set up using switches in a large array that are activated by various events that are subject to quantum uncertainty. Unlike a quantum computer that uses qubits, this computer would have uncertain circuitry, a large pool of components, some of which may be qubits, which may or may not be connected in any way at all. Ideally therefore, it would replicate an almost infinite number of possible computers simultaneously. Since those computers comprise pretty much the whole possible computer space, a Heisenberg computer would be able to undertake any task in hardware, instantly. Then the fun starts. One of the potential tasks it might address is to use trial and error and evolutionary algorithms to build a library of circuitry for machine consciousness. It would effectively bootstrap itself. So a Heisenberg computer could be conscious and supersmart. Food for thought.

To finish off and make the most of the closing hours of April Fool’s day, I wonder of there is any mileage in a space anchor? Unlike the virtual particle vacuum energy drive, this one would use the expansion and curvature of space as its propulsion mechanism. The idea came from watching Star Wars and the stupid fighters that manage apparently to turn quickly in space using wings, and you can even hear them do so. Vacuums are not high on the physics loyalty scale in Star Wars. Space fighters would have a lot of work to do to turn round, given the lack of medium. It would all have to be done by their propulsion systems. Unless. Unless, they had some sort of space anchor that could be applied to lock on to local space and used as an anchor point to swing around. Creating some sort of massive drag on the end of a tether (I don’t know, maybe  reliant on strong force interaction with virtual particles in the quantum foam), the ship would quickly find its angular momentum used to change direction. And if an anchor could be made that anchors into space, variations in expansion of space due to local curvature could be used to drag a ship along.

I doubt that any of these ideas hold much water, but they are fun, and who knows, someone smarter might take some stimulation from them and run with them into ideas that are better.