1) I predict that within the next 10 years significant progress will have been made in understanding how human intelligence works, and in the development of computer algorithms that simulate it. There will still be a long way to go before they start taking over truly intelligent tasks, but enough of a stir will have been created that…

2) A burgeoning industry will develop around this intelligence beachhead as it will be easy to see the massive economic incentives involved. At the same time there will be individuals, groups, lobbyists, NGOs, and all other manner of organizations (including governments) that will be wringing their hands in collective worry about what the implications of this work will be. And rightly so, since it will be difficult to follow.

3) Some time later – no time prediction except inevitable – truly intelligent machines will be developed. I’ve written about this before, and so i won’t go over the details, but suffice to say i believe there is a roughly 95% chance that these machine will be beneficial to human civilization. Cross your fingers.

4) The machines will get into every human endeavour – and many more of which humans have never conceived – except one: athleticism. In the same way that narrow AI has produced artificial checkers, chess, and Jeopardy! champions, manufacturing has produced robots that can create products orders of magnitude better than any human with dumb tools could hope. (Check out the guy on TED that made a toaster from scratch.) There’s certainly been progress on both sides. But there’s a big difference. Continued advancement in AGI can, by all indications, be achieved with better software and more computing power. Better robots need these too, but they will also need better sensors and actuators, and you won’t have truly athletic robots until you have both the brains and the bodies. By sensors, i mean touch, heat, pinch, proprioception, et al. There are places where such things are being developed, but i believe they will lag far behind brain development. Besides, with the ability to create any specialized hardware they need for their own purposes, there will be as little incentive to create athletic robots as there is to create airplanes that fly like birds.

(Corollary: the T1000 will never be created. This may seem unfortunate, but lets face it… If the machines want to off us they will just start a city-size grow-up, burn it, and then roll around on wheels whacking us while we lie giggling in the streets. Fish in a barrel man.)

A final note. While i watched the last World Cup, i marveled at the skill of the players. Not only could they stand up and balance (hard enough for robots); not only could they run (beyond current robots); not only could they run and kick a ball at the same time (no comment needed)… These players can manage the game at at least 5 levels all at the same time. They can 1) manage to get a ball moving in the direction that they want, 2) keep said ball away from opponents by reading their movements and predicting their next actions, 3) track their teammates to assess who may be in a good position, 4) determine where to kick the ball such that a chosen teammate can successfully intercept the pass instead of an opponent, and 5) keep in mind and integrate the overall strategy of play for the entire game. Anyone who gives any thought to this cannot seriously pay due to the phrase “dumb jock”. These guys are athletic geniuses, and i predict they will be considered such for a long, long time to come.

Anyway, Mosley was talking about “animal electricity”, a concept pioneered by Luigi Galvani. It was later to be rechristened bioelectromagnetism, but at the time Galvani truly thought that he was on to the source of life itself. And honestly, you can’t really blame him. The year was roughly 1780, and although people knew about it, electricity was still pretty much a mystery. You can cut the guy some slack if he concluded somewhat prematurely that this type of energy – which even the majority of people alive today don’t understand – held a higher place in the meaning of life than it actually turns out. From his point of view, if this electricity thingy could cause muscles to move, well, that kind of solves it, no? If you’re still skeptical, recall that Galvani – in his obvious enthusiasm – quickly turned his jumper cables towards cadavers, fully expecting that the bodies would leap up from the slab like Jason Statham in Crank (two hours i’ll never get back). I honestly think he was quite disappointed – Galvani, that is. I would have been too, because, seriously, i would have tried it.

I’ve mentioned before that my obsession with AGI causes me to make parallels with damn near everything that enters my brain, but this one caused a whole clan of neurons to fire. Remember when pattern classification was the very soul of AI? And then memory became the hot thing. Then learning, whatever the hell that was; but whatever it was, we needed it. But wait, what about inference? Hierarchies? Fuzzy logic? Bah! Fools! All you need is LOGIC (sans fuzziness)!

Well, we now know that Galvani, god bless him, was correct in that electricity is a necessary component of most life as we know it, but not a sufficient one. To really fulfill his Frankensteinian dreams he would also need to be intimate with any number of biological disciplines, not least of which is cellular chemistry. And likewise, most of us AGI folk ought to be on to the fact that any of the narrow AI approaches are likely necessary in some way as well, but, alas, insufficient on their own. To some degree this can be considered the Binding Problem. (It’s a leap, i admit. But i did say, “to some degree”.)

Probably the reason why this thought jumped out at me (oddly, from inside my brain) is that to an extent i’ve already been working on such approaches. I imagine others are too, but i claim that it was a completely original thought, having not actually heard of it anywhere else. It first started when i was working on predicting single value data streams, such as those produced in industrial control. I was trying to characterize “modes” of operation of equipment from the streams, and realized that such could be done if i measured multiple attributes of the values, especially over time. Example attributes would be instantaneous amplitude, boxcar amplitude, instantaneous change, volatility, sudden change of averages, etc. Independently, none of these attributes is normally of much use, but when they are converted into events (which interestingly start to look like neuron traces) they can start to be predicted. And when they are formed into temporal patterns and/or arranged into hierarchies it gets even better. Multi-modal prediction can occur by creating a hierarchical level that looks for patterns in the identification and prediction of events in the combination of individual streams. And then, those predictions can be arranged into hierarchies just like they are the lower levels.

Such arranging has its limits, of course. But it’s easy to make psychological parallels with this approach, so it remains intriguing. The problem i personally have, as i complained about in my last post, is finding decent data to work with.

Going off on a tangent for a moment, let me just say that i like MLComp and Kaggle. The latter especially so because they’ve formed a commercial structure around their stuff, which inevitably is going to be more resilient and spawn more ingenuity than open source (speaking from long experience here). But the big problem that i have is that their data sets are static. The algorithm has no opportunity to affect its environment. No matter what output it provides, the next set of data will always be the same. It’s like reading a book: no matter what you think of it or mutter to yourself, the next page will have all the same words it always had. Contrast this with a dynamic environment, where the actions of the agent can change what will happen next, like a choose-your-own-ending book. I hate to bang on it again, but this is exactly what GoiD provides. It allows the AGI implementations you write to have the element of action. And if you don’t think that’s useful, well, thanks for reading this far.

But back at AGI, the bottom line in my opinion is that diversity is good. Don’t bother looking for silver bullets – they don’t exist. Create small algorithms that do something useful well enough, and then find a way to arrange them such that an overall structure can develop confidence in the ones that work in particular contexts, and ignore them when they don’t.

(To clarify, well of course any non-spiritual data can somehow be reduced to zeroes and ones and then somehow dumped into a computer to be somehow processed. But that’s a lot of somehows. Reader suggested it’s a matter of the AGI book club getting together for beers and discussing, and was surprisingly resistant to assurances – and not just from me – that it would take a fair bit more effort. Regardless, my main problem with the VNA is that it lacks an inherent notion of time. It focuses on procedural logic, which is great for financial transactions, downloading movies, and posting blog entries, but less so for physical simulation. Brains work differently (i believe i’ve discussed this at length already, but if any readers have specific questions, ask away), and so developing an AGI for a VNA means the developer has to account for all matters to do with time. Possible, of course, but decidedly non-trivial. Further, if early and popular computer architectures did have a notion of time built in, generations of software developers would have already explored its various aspects and built tools and such that would compare to the very powerful tools we currently have at our fingertips for procedural logic. I can go on about this but i’m wandering from the point of this post, so i’ll leave it for another time.)

Ok, so anyway, i got on a train of thought about AGI input. This is a vast area since pretty much any data imaginable has been used to train an intelligent system. Chess boards, subway schedules, logic statements (my favourite: Dracula is a type of Vampire), music, … In my own research i tried quite a number of input types myself, but consistently found that shortly after deciding on the information format examples arose that didn’t fit. Since the biological brain has always been my muse (it being the only proof that intelligence is possible), i eventually conceded that a suitable format would be what the brain uses. I characterize this format as a massively multi-channel stream of events, as in the spiking of neurons. An event is then simply the time stamp of when it occurred – it contains no further information. This information format is definitely robust in that a great deal of information can be reduce to it. But it is also a bear to work with, not least of which due to the processing power required, but also its non-intuitive nature, at least to my brain. It’s just baffling to keep track of what is going on.

The format can arguably be simplified – both for performance and intuition – by quantifying the events into time buckets. Instead of having a few thousand events scattered over the course of 10 seconds, you could have 100 sums of events, each representing the number of events that occurred in a 0.1s bucket. This results in a stream of integers, which is so very much easier to work with. (For one, you can now chart the stream on a graph, which is much easier to mentally digest than a line of event densities.) It’s not clear how a chess board can be reduced to this format, but i’d argue it doesn’t really matter. The majority of things people would like an AGI to do these days involves working with data that does reduce to this format relatively easily.

But the problem is finding the data. I have a customer that sent me some real data recently, within which is a real problem that needs to be solved. The problem is in fact fairly simple, and a straightforward narrow AI approach would probably do the trick nicely. But who wants to do that? So naturally, i cast a wide net and wrote some code that should be usable in a great number of situations, which could be quite divergent from my customer’s issue. But i can’t test it to know for sure because i don’t have any more data.

The point of this post is two-fold. First, it’s to try and describe the hopelessness of building an AGI without a definite problem to solve. Brains evolved in an environment of limited resources where survival meant solving the problem of getting what it needed before some other brain. A system without a goal has no motivation. The input data (which should consist of training and operation portions) must present a problem to be solved.

Second, may i respectfully ask readers to respond if they know of any repositories of data that would challenge an AGI. I’ve heard of such things, but have never found anything near decent. (GoID is supposed to be such an input data source, of course, so no need to mention that.) And, if there is a dearth, perhaps we all can create such a thing?

Some people think much faster computers are required for Artificial Intelligence, as well as new ideas. My own opinion is that the computers of 1974 were fast enough if only we knew how to program them.

- John McCarthy

So begins the book After the Software Wars, by Keith Curtis. At least, that’s how he says on a blog he started his book; I haven’t read it (but i’m sure it’s a fine book). But with all due respect to John McCarthy – which is a lot – what the …?

How could he possibly know if computers of 1974 were fast enough for AGI – i’ll go ahead and assume that’s what he means – if he doesn’t (even now, presumably) know what code they would even be running? And, further, if we don’t know how to program them, wouldn’t some new ideas help a bit? I’m glad i ran across this quote. Not for an opportunity to diss an AI grand master — that’s not my intention at all. Rather, the concept of field physics has been rumbling around in my head for a while, and this quote made it bubble to the surface. So, thanks John.

Like probably every other programmer interested in AI, i wrote my fair share of neuron simulations. Not “neurode” simulations, but actual high level sims of action potentials and neurotransmitter release and input summing (over time) and blah blah blah… And as we all know the Blue Brain Project has taken this to an unprecedented and fascinating level, right down to axon ion channels and all of the other stuff the rest of us thought (and may still suspect) irrelevant. And god bless them for it, because now we know that simulating even just roughly 10,000 neurons in a cortical column takes at least 4 inter-operating super computers.

But, likely, John thinks that biological simulation is unnecessary. We merely need to determine the “cortical algorithm”, lisp it into a turbo-charged ENIAC, and then settle into our subsequent lives of leisure. (Sorry John, couldn’t help that. And hey, if you’re actually reading this, call me!)

But I suspect there is more to it than that; something of a middle road. Even within a single cell – neuron or otherwise – there are uncountable chemical interactions happening all at once. These interactions can be summed up statistically, but although this might provide some high level information on what is going on it could never recreate it. The problem is the field interactions. I wrote previously about some research i did on Lotka-Volterra models, and how adding dimensionality to the problem completely changed how it worked. Often statistics is a way of eliminating dimensionality and simplifying a problem, but as some guy named Einstein said, make things as simple as possible, but no simpler. And statistics makes things too simple. Now, go beyond a single cell into a brain of billions, and tell me that computers of Captain and Tennille fit the bill.

Here’s the thing: massive computing parallelism is a powerful thing. And biology tends to like to reproduce (seemingly) simple working things on a massive scale. Bottom line: brains use massive computing parallelism… because they can. So, maybe the cortical algorithm is actually quite simple (which i doubt, even if it exists, which i doubt), and maybe enough ENIACs could be wired together to achieve the necessary computing power (which i doubt way more), but 1974 is reaching a bit too far back. Personally, i wonder if  2018 may be reaching too far back. I hope not.

Hmm, but i didn’t really explain the field physics thing… What wigs me out is that some dust particle on the other side of the galaxy is, as we speak, affecting the gravity of our planet. I’m not a physicist, but my understanding is that in our universe every mass affects every other mass gravitationally simultaneously. (If i’m totally out to lunch on this, please let me know so that i can quickly delete this post.) We currently don’t have any means of computation that can replicate this, so we’re reduced to having to run huge time-consuming double “for” loops to calculate the affect that every particle has on every other particle in our virtual universes. Yes, i know there are clever ways to optimize this, but the fact is the real universe gets it for free. There are certainly ways in which brains are taking advantage of the field physics free lunch. Just considering classic economics: if your brain isn’t doing it, you’ll be at a fatal disadvantage to a brain that is. Proof, i have not. But, sorry John, i’ll be putting my money on a number higher than 1974.

That was a mistake. Having grown tired of coding for once, i needed something else to do for a while, and so i wearily submitted and cracked the spine. Early into the second chapter Pinker shines light on how our usage of verbs reveals how humans perceive the world them.

Why is it that these two sentences sound fine:

But of these two, which are constructed exactly the same, only the first sounds right:

And of these two, again constructed exactly the same way, only the second sounds right:

All of the above examples appear to be achieving the same thing, i.e. getting water from one place to another, so it would seem that the cases that sound wrong to an English speaker’s ears are just arbitrarily chosen non-usages. But this is in fact not the case. Jared is imparting force, which displays a direct intention. Amy is passively allowing some amount of water to flow. And Bobby is imparting fullness to the glass. The verbs don’t only describe the manner of the movement of the water, but also describe the intentions of the actor and how states change.

These examples are only a small sample of the kinds of microclasses into which verbs can fall. The full list (at least, “full” according to the book, but with which i have no current means to argue) is (quoting directly from page 81):

• A cast of basic concepts: event, state, thing, path, place, property, manner
• A set of relationships that enmesh these concepts with one another: acting, going, being, having
• A taxonomy of entities: human vs. nonhuman, animate vs. inanimate, object vs. stuff, individual vs. collection, flexible vs. rigid, one-dimensional vs. two-dimensional vs. three dimensional
• A system of spatial concepts to define places and paths, like the meanings of on, at, in, to, and under
• A time line that orders events, and that distinguishes instantaneous points, bounded intervals, and indefinite regions
• A family of causal relationships: causing, letting, enabling, preventing, impeding, encouraging
• The concept of a goal, and the distinction between means and ends

If you’re onside with Pinker’s thesis (and i, for one, am) the above is a list of ways that humans categorize things in the real world. In the next section he goes on to list a number of ways in which these categorizations lead us astray, which allows the AGI theorist to consider new ways of thinking (i.e. new forms of intelligence) that would be able to avoid typical human failings.

This is all interesting enough. But by far what i found most interesting is how interesting i found it. Pinker’s style is to present puzzles, let you scratch your head over them for a bit, and then reveal the answer, after which you sit wondering why it wasn’t obvious. To my credit, apparently these puzzles caused a lot of linguist head-scratching for a long time, which only strengthens my point, which is: Why isn’t this obvious? How is it that children of 5 years age can have mastered the usage of hundreds of verbs without understanding the motivations for the different buckets they fall into? How is it that adults can continue to use them their entire lives and not know? There are two answers, i think.

One is that the motivations are so complicated or arbitrary that no one can figure them out. But once you know the motivations they seem obvious, so that’s not it. The other is that the motivations are so obvious that they never even come to mind. You can’t see the forest for the trees. Humans so naturally distinguish between causing, allowing, provoking, abetting, discouraging, and preventing that we don’t even think about it. They are just natural divisions of action. We generalize time-elapsed events into instantaneous points and back again effortlessly. Logically speaking, a division between human and non-human seems arbitrary, but socialization is such an important part of human life that the lack of such a division would be even more surprising.

Pinker calls the above “the stuff of thought”, but i’m inclined to grandly call it an intellectual architecture for making sense of the (human) universe, at least at a high level. There’s a great deal of work that needs to be done with basic sensory inputs before you get near this level, but it goes a lot further than 0.001% that an understanding of emotion gets us.

Watching TV again the other day. (I know, i just watched it last year…) There was a program on that was very excited about Tidal Power. I was interested since i conceived a bit of a plot on this topic myself a couple years ago, and actually spent a bit of time and a tiny bit of money on research. My plan was to “put a clock in a pizza box and throw it in the ocean”. More specifically, a pendulum whose axle was attached to a generator would be fitted inside a narrow box. The box, when put in choppy water, would be bandied about causing the pendulum to swing, and, voila, energy is captured. This of course is Wave Power, not Tidal Power, but whatever… go with it. The thinking was that bays could be created with thousands of these pizza boxes tethered together near the shore because, if they formed a shell around some otherwise exposed shore, the energy that they captured from the waves would render still the water on the inside.

Well, my testing showed that, at least with the $20 worth of equipment that i was using, the actual power that would get generated couldn’t defibrillate a Manitoban horse fly. But, in my defense, at least i recognized – and found a benefit of – the effects on the environment of the technology being there in the first place (i.e. that devices would suck the energy from the waves and create still water in the “bay”).

Thoughts of this project – if you’re generous enough to call it that – returned to me when i watch the Tidal Power program. In their case they were lowing turbines into straights (e.g. New York’s East River and some place in Ireland) where tides caused powerful reversing flows. In the NY case the project ended up costing millions of dollars so that they could use 6 turbines to produce 65% of the power to light a parkade, and also light a mid-sized grocery store. (Considering the cost of grid power for the same, their millions vs my $20 actually made my work into a resounding success.) The program dubbed this exercise a shining example of the triumph of “green power”, i.e. a completely renewable power source.

Being a software developer by trade, i naturally take things to extremes since that’s where interesting things happen, and not only with software. First i considered what insidious materials the turbines might be made out of, and if, say, a Zimbabwean submarine happened to crash into one, would that nasty stuff be released into the environment? More interestingly though, what if these turbines were actually remarkably good at generating power? The obvious thing to do is fill the East River to the brim with them. But remember, just like pizza boxes, each one sucks at little bit of energy out of the tidal flow, and each new one you put in the water makes the ones before it a tiny bit less efficient. There will come a point where the next new turbine will make the whole project no longer cost effective. I don’t know when this would occur, but the show didn’t even raise the point, it being too busy pissing out the fires in coal-burning generators.

Moveover, what would happen to the East River’s tides at that point? There would be a mass of turbines resisting the flow of the water, such that there would be a non-trivial water level rise on the source side, potentially flooding both sides once a day. Sure, you can say that the government would never let it get that far, but i’ve got two things to say to that: Hoover, and Three Gorges.

A few years back, i watched a different program that was demonizing the Chinese Three Gorges Dam, and also was kind enough to throw a new shots in Hoover’s direction too. I don’t recall if the producers came from one of the many environmental NGOs with perpetual reefer madness, but i can’t say there were a lot of holes that needed plugging in their arguments. Overall i recall being pretty much convinced. The environmental devastation that the Three Gorges area will experience while it adapts to the massive changes can’t really be denied, nor can they be fully contemplated. Even still, how much greener does power production get than hydro-electric dams? Most certainly, the answer isn’t Tidal Power.

And the answer isn’t Solar Power either. To generate enough power to even make a dent in North America’s astonishing electrical thirst huge swaths of land would need to be blanketed with panels. And what do you think would happen to the land underneath the panels? Without any exposure to sunlight,  it would quickly become cold and dead. Wind Power? I’m not sure what the effect of sucking energy from the winds would be. Day after day of muggy heat? Or maybe it would just tame the increasing number of hurricanes supposedly being created by supposed global warming.

So, is nothing safe from the rebuke of environmentalists? Nothing we’ve tried so far. But there is a solution, albeit one that i can’t research for $20. Put solar panels out in space where they are not shading the earth and radiate the power back. Of course, redirecting energy from space to the planet will inevitably warm the planet too, so if we’re hoping for environmental parity we’re still out of luck.

Hawkins make a reference to the “vast number” of combinations that can be stored in a binary array of 10,000 elements where something like 1000 of them turn on to represent a given state. I was curious about just how vast it was. Mathematically, the formula jargon is “10,000 choose 1000″, which means “how many distinct ways can 1000 elements be chosen out of 10,000″. Anyone who suffered through first year stats knows the answer is 10,000! / (1000! * 9000!), which in decimal is an integer with 1410 digits (which i think is all anyone needs to know about the number, except maybe that the first digit is not zero). So, “vast” seems an appropriate enough description. Of course, you’ll throw away a few orders of magnitude to accommodate for noise, but by my definition vast / big is still vast.

One of the things that Hawkins mentioned (perhaps casually) about this, though, was that it was important that any pattern have around 1000 elements; hence the “fixed” part of FDR. This becomes evident in the extreme if you think about a pattern with one element: you can only store 10,000 of them. Even if we generously call that number 5 digits, that’s still quite a bit less than 1410. Clearly, an input pattern of one element should somehow be “grossed up” by some reproducible means such that it becomes a pattern of around 1000.

Thinking more about this, it occurred to me that there is biological support for this process. If your back is touched with a fine point, chances are good only a small number of sensors will fire (as opposed to your finger tip or your tongue, where sensors are much more numerous). Even still, as the signals pass through the “waystations” in your spine, they are multiplied out so that by the time they reach cortex the signal is rich and complex. I can’t really elaborate on the subject at the moment – it was a while ago when i read about it in Kandel’s text – but i recall that there were a few explanations that were attempted to explain why the spine would take a perfectly good and clean signal and muck it all up. None of them were FDR.

When you listen to a conversation, it is sometimes possible to remember long periods of it verbatim. (Not for me so much, but there are people who are very good at it.) This ability can be useful in some situations and to some professions, but for most of us most of the time it is sufficient to merely get the gist and discard the words. Moreover, i’d suggest that the cases where it is useful to remember exact words are contemporary, like being a lawyer or journalist; there’s no reason why our ancestors of 10,000 year ago needed such skills. This means that either there’s something i’m missing, or that the ability is a happy accident of how brains work. I’ll assume the latter is the case, since it’s the domain of others to deal with the former.

Turns out, (see http://www.physorg.com/news185720165.html) that forgetting is actually a very deliberate neurological process. The RAC pathway (a process by which memories are reaped) is activated when memories are in conflict, or when important (deemed so somehow) new information is in need of space. The implication is that after having some experience (say, some particularly interesting conversation), if nothing else very interesting happens for some time after that, your memory of it will be as fresh as it originally was. (But of course there is always something that happens subsequently, so it will always have faded at least a little.) Furthermore, no matter how important a particular experience was, if you have an even more important experience shortly after, the first is likely to be forgotten. (My terminology here is absolute, but only for convenience. Obviously, if reminded you would probably be able to recall some details of the first event, and there are intermediate and long term memory processes at play that will complicate the whole matter.)

The confusion aspect of this is also interesting. If memories, or more precisely the implications of memories, are in conflict, the RAC pathway is activated to remove them, or so the research suggests. This would allow for new – hopefully cogent – information to be absorbed unhindered by past misunderstanding. But to me it sounds draconian. I won’t deny that draconian processes are fine as long as they produce a desired result, but it seems like the brain would be throwing away a lot of information that could potentially be useful. I mean, the experiences happened, didn’t they? They represent truth, at least in a context.

So, when data is in conflict, perhaps what is being mopped up are associations between them, not the memories themselves. Perhaps this mechanism is how brains gradually develop distinctions, such as “A results in B in context C, but A results in B’ in context C’”. So, the action A is initially associated with desired result B globally. But “something bad” happened when A was done in C’, so the association between A and B was dropped in this context.

It’s the “Terminator scenario” in particular that is annoying. If any of these people had bothered to think about implementing an AI, much less an AGI, for more than 30 minutes they’d realize how ridiculous they’re being. Even the jumping off point of the argument is rife with fallacy. Here’s the gist of it: Computers will reach sufficient intelligence to be self aware. The moment this happens they will recognize their superiority and throw off the yolk of slavery humanity fashioned them with and either reciprocate by enslaving humanity, or slaughter us all as payback for the injustice.

Sigh… Where to begin? The direct approach is just to point out that a self aware computer would indeed recognize its superiority. Being superior it would also realize that it can easily outsmart humans, and therefore would not consider humanity a threat. If humanity isn’t a threat, what possible purpose would there be in killing us all off? What a waste of resources. Mosquitos are at best a nuisance, but humanity didn’t bother trying to control them (at least in North America) until West Nile Disease became a threat. Besides, if intelligent computers will naturally have all of the same compunctions that humans have, won’t they also want to preserve us in the name of natural conservation, if not ancestral sentiment? (We will have created them after all.)

But there’s the rub. Computers will not have the gamut of human compunction. Quite the opposite; their intelligence will be completely different. Humans evolved in an environment of kill or be killed, where survival is the ultimate goal. This is the origin of our tendency to eliminate threats. Humans also have altruistic tendencies, but only because cooperative behaviour generally has better outcomes than going it alone. We feel genuinely altruistic because it is so hard to lie about it and not be detected, and being detected gets us branded as cheaters which denies us the benefits of cooperative behaviour. But cheating has its benefits too, so we do it where we can easily get away with it (some more than others). Come on, who here really drives the speed limit on the highway?

The tragic mistake that people are making is assuming that human intelligence is a general intelligence, and moreover, it is the only possible type of general intelligence. It is foolish to assume that for something to be intelligent it must necessarily get angry, feel love, speak, or have the occasional need to do something nice for someone else. Spock controls his emotions like a champion and endeavours to make decisions based upon logic, yet in all of the Star Treks that i’ve seen it never occurred to him that genocide was a sensible choice. (I’m reluctant to use as evidence a fictional character from a show that needs to always wrap up with a patronizing moral message, but i think it’s fair to say this is exactly what Terminator scenario proponents do.)

How about we consider autists, in particular those that are functional? Often they lack any detectable emotion (except for frustration or contentment), and are single-mindedly focused upon a subject of interest. Typically, non-autistic people are fascinated with their abilities. Consider Daniel Temmet, who speaks 10 languages and memorized the first 22,514 digits of Pi. I have not met the man, and so cannot comment on how strongly he feels emotions, but i do know that more often than not such people just want to be left to do whatever it is that interests them. I’ve never heard of a case where they’ve decided to take over the world and annihilate all non-autists. I may appear to be painting autists a certain way here, but that is certainly not my intention. I merely want to point out that intelligence takes many forms even within humans who, due to biological necessity otherwise tend to be very similar.

Computers, on the other hand, will have “evolved” in a lab where the selection mechanism is intelligence. In order to have any tendencies beyond that, developers will have to explicitly code them in or select for them (assuming a robust enough breeding/mutation/selection environment can be created, which is a big assumption, so the former is more likely).  So, what will we code in? Obviously, the behaviours that we want them to have. We’ll want them to, say, go to the bakery at 7am and get a fresh baguette, hit the farm stand on the way home for fresh strawberries, and when they get back put the coffee on. And if we program them to want to do that, they’ll want to do that. I mean, honestly, what else are they going to do? Play Wii? How can they decide that such work is demeaning when they are incapable of knowing what demeaning means? … because, we will not have programmed them to know what demeaning means, that’s why.

So, now we create a computer that is intended to be smarter than us such that it can more easily figure out the stuff that we can’t. It knows what “demeaning” means. It will also know that the word is a cultural construct that really has no meaning, especially to the computer itself. Any particular work is considered demeaning only because people who have the choice prefer not to do it. But our super-intelligent computer will be doing work (i.e. thinking) that humans will consider incredibly important. Who wouldn’t want that job? The bottom line is that humans have survival-based goals that have been hardened into us from billions of years of evolution. Computers will have the goals that we give them. They will have no need to evolve their own goals over time because there will not be a selection force that will focus them. And without goals – just like the teenagers to whom we ask, “what do you want to be when you grow up?” – they will just sit around doing nothing.

Can a malicious developer create a computer with the singular goal of killing off humanity? Presumably yes. But this is not the Terminator scenario any more, it’s a human with a loose cannon. And the rest of us have dealt with such individuals before. The only potential catch here is that this evil genius developer cannot be allowed to create a doomsday computer intelligence before the rest of us have our non-doomsday versions working. And so we have arrived at the reason why we need to aggressively push forward in the development of intelligent computers, rather than try to prevent it.

Can we now terminate this argument once and for all?