Kirsten and I spent the week between the 17th and the 21st of September with 18 other utterly amazing people having Chatham House Rule-governed conversations about the Future of Artificial Intelligence.
We were in Norway, in the Juvet Landscape Hotel, which is where they filmed a lot of the movie Ex Machina, and it is even more gorgeous in person. None of the rooms shown in the film share a single building space. It’s astounding as a place of both striking architectural sensibility and also natural integration as they built every structure in the winter to allow the dormancy cycles of the plants and animals to dictate when and where they could build, rather than cutting anything down.
And on our first full day here, Two Ravens flew directly over my and Kirsten’s heads.
[Image of a rainbow rising over a bend in a river across a patchy overcast sky, with the river going between two outcropping boulders, trees in the foreground and on either bank and stretching off into the distance, and absolutely enormous mountains in the background]
I am extraordinarily grateful to Andy Budd and the other members of the Clear Left team for organizing this, and to Cennydd Bowles for opening the space for me to be able to attend, and being so forcefully enthused about the prospect of my attending that he came to me with a full set of strategies in hand to get me to this place. That kind of having someone in your corner means the world for a whole host of personal reasons, but also more general psychological and socially important ones, as well.
I am a fortunate person. I am a person who has friends and resources and a bloody-minded stubbornness that means that when I determine to do something, it will more likely than not get fucking done, for good or ill.
I am a person who has been given opportunities to be in places many people will never get to see, and have conversations with people who are often considered legends in their fields, and start projects that could very well alter the shape of the world on a massive scale.
Yeah, that’s a bit of a grandiose statement, but you’re here reading this, and so you know where I’ve been and what I’ve done.
I am a person who tries to pay forward what I have been given and to create as many spaces for people to have the opportunities that I have been able to have.
I am not a monetarily wealthy person, measured against my society, but my wealth and fortune are things that strike me still and make me take stock of it all and what it can mean and do, all over again, at least once a week, if not once a day, as I sit in tension with who I am, how the world perceives me, and what amazing and ridiculous things I have had, been given, and created the space to do, because and in violent spite of it all.
So when I and others come together and say we’re going to have to talk about how intersectional oppression and the lived experiences of marginalized peoples affect, effect, and are affected and effected BY the wider techoscientific/sociotechnical/sociopolitical/socioeconomic world and what that means for how we design, build, train, rear, and regard machine minds, then we are going to have to talk about how intersectional oppression and the lived experiences of marginalized peoples affect, effect, and are affected and effected by the wider techoscientific/sociotechnical/sociopolitical/socioeconomic world and what that means for how we design, build, train, rear, and regard machine minds.
My piece “Cultivating Technomoral Interrelations,” a review of Technology and the Virtues: A Philosophical Guide to a Future Worth Wanting has been up over at the Social Epistemology Research and Reply Collective for a few months, now, so I figured I should post something about it, here.
As you’ll read, I was extremely taken with Vallor’s book, and think it is a part of some very important work being done. From the piece:
Additionally, her crucial point seems to be that through intentional cultivation of the self and our society, or that through our personally grappling with these tasks, we can move the world, a stance which leaves out, for instance, notions of potential socioeconomic or political resistance to these moves. There are those with a vested interest in not having a more mindful and intentional technomoral ethos, because that would undercut how they make their money. However, it may be that this is Vallor’s intent.
The audience and goal for this book seems to be ethicists who will be persuaded to become philosophers of technology, who will then take up this book’s understandings and go speak to policy makers and entrepreneurs, who will then make changes in how they deal with the public. If this is the case, then there will already be a shared conceptual background between Vallor and many of the other scholars whom she intends to make help her to do the hard work of changing how people think about their values. But those philosophers will need a great deal more power, oversight authority, and influence to effectively advocate for and implement what Vallor suggests, here, and we’ll need sociopolitical mechanisms for making those valuative changes, as well.
[Image of the front cover of Shannon Vallor’s TECHNOLOGY AND THE VIRTUES. Circuit pathways in the shapes of trees.]
This is, as I said, one part of a larger, crucial project of bringing philosophy, the humanities, and social sciences into wide public conversation with technoscientific fields and developers. While there have always been others doing this work, it is increasingly the case that these folks are being both heeded and given institutional power and oversight authority.
As we continue the work of building these systems, and in the wake of all these recent events, more and more like this will be necessary.
A few weeks ago I had a conversation withDavid McRaney of the You Are Not So Smart podcast, for his episode on Machine Bias. As he says on the blog:
Now that algorithms are everywhere, helping us to both run and make sense of the world, a strange question has emerged among artificial intelligence researchers: When is it ok to predict the future based on the past? When is it ok to be biased?
“I want a machine-learning algorithm to learn what tumors looked like in the past, and I want it to become biased toward selecting those kind of tumors in the future,” explains philosopher Shannon Vallor at Santa Clara University. “But I don’t want a machine-learning algorithm to learn what successful engineers and doctors looked like in the past and then become biased toward selecting those kinds of people when sorting and ranking resumes.”
We talk about this, sentencing algorithms, the notion of how to raise and teach our digital offspring, and more. You can listen to all it here:
This summer I participated in SRI International’s Technology and Consciousness Workshop Series. The meetings were held under the auspices of the Chatham House Rule, which means that there are many things I can’t tell you about them, such as who else was there, or what they said in the context of the meetings; however I can tell you what talked about. In light of this recent piece in The Boston Globe and the ongoing developments in the David Slater/PETA/Naruto case, I figured that now was a good time to do so.
I presented three times—once on interdisciplinary perspectives on minds and mindedness; then on Daoism and Machine Consciousness; and finally on a unifying view of my thoughts across all of the sessions. This is my outline and notes for the first of those talks.
In a 2013 aeon Article Michael Hanlon said he didn’t think we’d ever solve “The Hard Problem,” and there’s been some skepticism about it, elsewhere. I’ll just say that said question seems to completely miss a possibly central point. Something like consciousness is, and what it is is different for each thing that displays anything like what we think it might be. If we manage to generate at least one mind that is similar enough to what humans experience as “conscious” that we may communicate with it, what will we owe it and what would it be able to ask from us? How might our interactions be affected by the fact that its mind (or their minds) will be radically different from ours? What will it be able to know that we cannot, and what will we have to learn from it?
So I’m going to be talking today about intersectionality, embodiment, extended minds, epistemic valuation, phenomenological experience, and how all of these things come together to form the bases for our moral behavior and social interactions. To do that, I’m first going to need ask you some questions:
This essay is something of a project of expansion and refinement of my previous essay “Labouring in the Liquid Light of Leviathan,” considering the Roko’s Basilisk thought experiment. Much of the expansion comes from considering the nature of simulation, memory, and identity within Jonathan Nolan’s TV series, Person of Interest. As such, it does contain what might be considered spoilers for the series, as well as for his most recent follow-up, Westworld.
Use your discretion to figure out how you feel about that.
Are You Being Watched? Simulated Universe Theory in “Person of Interest”
Jonah Nolan’s Person Of Interest is the story of the birth and life of The Machine, a benevolent artificial super intelligence (ASI) built in the months after September 11, 2001, by super-genius Harold Finch to watch over the world’s human population. One of the key intimations of the series—and partially corroborated by Nolan’s follow-up series Westworld—is that all of the events we see might be taking place in the memory of The Machine. The structure of the show is such that we move through time from The Machine’s perspective, with flashbacks and -forwards seeming to occur via the same contextual mechanism—the Fast Forward and Rewind of a digital archive. While the entirety of the series uses this mechanism, the final season puts the finest point on the question: Has everything we’ve seen only been in the mind of the machine? And if so, what does that mean for all of the people in it?
Our primary questions here are as follows: Is a simulation of fine enough granularity really a simulation at all? If the minds created within that universe have interiority and motivation, if they function according to the same rules as those things we commonly accept as minds, then are those simulation not minds, as well? In what way are conclusions drawn from simulations akin to what we consider “true” knowledge?
In the PoI season 5 episode, “The Day The World Went Away,” the characters Root and Shaw (acolytes of The Machine) discuss the nature of The Machine’s simulation capacities and the audience is given to understand that it runs a constant model of everyone it knows, and that the more it knows them, the better its simulation. This supposition links us back to the season 4 episode “If-Then-Else,” in which the machine runs through the likelihood of success through hundreds of thousands of scenarios in under one second. If The Machine is able to accomplish this much computation in this short a window, how much can and has it accomplished over the several years of its operation? Perhaps more importantly, what is the level of fidelity of those simulations to the so-called real world?
[Person of Interest s4e11, “If-Then-Else.” The Machine runs through hundreds of thousands of scenarios to save the team.]
These questions are similar to the idea of Roko’s Basilisk, a thought experiment that cropped up in the online discussion board of LessWrong.com. It was put forward by user Roko who, in very brief summary, says that if the idea of timeless decision theory (TDT) is correct, then we might all be living in a simulation created by a future ASI trying to figure out the best way to motivate humans in the past to create it. To understand how this might work, we have to look as TDT, an idea developed in 2010 by Eliezer Yudkowsky which posits that in order to make a decision we should act as though we are determining the output of an abstract computation. We should, in effect, seek to create a perfect simulation and act as though anyone else involved in the decision has done so as well. Roko’s Basilisk is the idea that a Malevolent ASI has already done this—is doing this—and your actions are the simulated result. Using that output, it knows just how to blackmail and manipulate you into making it come into being.
Or, as Yudkowsky himself put it, “YOU DO NOT THINK IN SUFFICIENT DETAIL ABOUT SUPERINTELLIGENCES CONSIDERING WHETHER OR NOT TO BLACKMAIL YOU. THAT IS THE ONLY POSSIBLE THING WHICH GIVES THEM A MOTIVE TO FOLLOW THROUGH ON THE BLACKMAIL.” This is the self-generating aspect of the Basilisk: If you can accurately model it, then the Basilisk will eventually, inevitably come into being, and one of the attributes it will thus have is the ability to accurately model that you accurately modeled it, and whether or not you modeled it from within a mindset of being susceptible to its coercive actions. The only protection is to either work toward its creation anyway, so that it doesn’t feel the need to torture the “real” you into it, or to make very sure that you never think of it at all, so you do not bring it into being.
All of this might seem far-fetched, but if we look closely, Roko’s Basilisk functions very much like a combination of several well-known theories of mind, knowledge, and metaphysics: Anselm’s Ontological Argument for the Existence of God (AOAEG), a many worlds theorem variant on Pascal’s Wager (PW), and Descartes’ Evil Demon Hypothesis (DEDH; which, itself, has been updated to the oft-discussed Brain In A Vat [BIAV] scenario). If this is the case, then Roko’s Basilisk has all the same attendant problems that those arguments have, plus some new ones, resulting from their combination. We will look at all of these theories, first, and then their flaws.
To start, if you’re not familiar with AOAEG, it’s a species of prayer in the form of a theological argument that seeks to prove that god must exist because it would be a logical contradiction for it not to. The proof depends on A) defining god as the greatest possible being (literally, “That Being Than Which None Greater Is Possible”), and B) believing that existing in reality as well as in the mind makes something “Greater Than” if it existed only the mind. That is, if God only exists in my imagination, it is less great than it could be if it also existed in reality. So if I say that god is “That Being Than Which None Greater Is Possible,” and existence is a part of what makes something great, then god must exist.
The next component is Pascal’s Wager which very simply says that it is a better bet to believe in the existence of God, because if you’re right, you go to Heaven, and if you’re wrong, nothing happens; you’re simply dead forever. Put another way, Pascal is saying that if you bet that God doesn’t exist and you’re right, you get nothing, but if you’re wrong, then God exists and your disbelief damns you to Hell for all eternity. You can represent the whole thing in a four-option grid:
[Pascal’s Wager as a Four-Option Grid: Belief/Disbelief; Right/Wrong. Belief*Right=Infinity;Belief*Wrong=Nothing; Disbelief*Right=Nothing; Disbelief*Wrong=Negative Infinity]
And so here we see the Timeless Decision Theory component of the Basilisk: It’s better to believe in the thing and work toward its creation and sustenance, because if it doesn’t exist you lose nothing, but if it does come to be, then it will know what you would have done either for or against it, in the past, and it will reward or punish you, accordingly. The multiversal twist comes when we realise that even if the Basilisk never comes to exist in our universe and never will, it might exist in some other universe, and thus, when that other universe’s Basilisk models your choices it will inevitably—as a superintelligence—be able to model what you would do in any universe. Thus, by believing in and helping our non-existent Super-Devil, we protect the alternate reality versions of ourselves from their very real Super-Devil.
Descartes’ Evil Demon Hypothesis and the Brain In A Vat are so pervasive that we encounter them in many different expressions of pop culture. The Matrix, Dark City, Source Code, and many others are all variants on these themes. A malignant and all-powerful being (or perhaps just an amoral scientist) has created a simulation in which we reside, and everything we think we have known about our lives and our experiences has been perfectly simulated for our consumption. Variations on the theme test whether we can trust that our perceptions and grounds for knowledge are “real” and thus “valid,” respectively. This line of thinking has given rise to the Simulated Universe Theory on which Roko’s Basilisk depends, but SUT removes a lot of the malignancy of DEDH and BIAV. The Basilisk adds it back. Unfortunately, many of these philosophical concepts flake apart when we touch them too hard, so jamming them together was perhaps not the best idea.
The main failings in using AOAEG rest in believing that A) a thing’s existence is a “great-making quality” that it can possess, and B) our defining a thing a particular way might simply cause it to become so. Both of these are massively flawed ideas. For one thing, these arguments beg the question, in a literal technical sense. That is, they assumethat some element(s) of their conclusion—the necessity of god, the malevolence or epistemic content of a superintelligence, the ontological status of their assumptions about the nature of the universe—istrue without doing the work of provingthat it’s true. They then use these assumptions to prove the truth of the assumptions and thus the inevitability of all consequences that flow from the assumptions.
Another problem is that the implications of this kind of existential bootstrapping tend to go unexamined, making the fact of their resurgence somewhat troubling. There are several nonwestern perspectives that do the work of embracing paradox—aiming so far past the target that you circle around again to teach yourself how to aim past it. But that kind of thing only works if we are willing to bite the bullet on a charge of circular logic and take the time to showing how that circularity underlies all epistemic justifications. The only difference, then, is how many revolutions it takes before we’re comfortable with saying “Enough.”
Every epistemic claim we make is, as Hume clarified, based upon assumptions and suppositions that the world we experience is actually as we think it is. Western thought uses reason and rationality to corroborate and verify, but those tools are themselves verified by…what? In fact, we well know that the only thing we have to validate our valuation of reason, is reason. And yet western reasoners won’t stand for that, in any other justification procedure. They will call it question-begging and circular.
Next, we have the DEDH and BIAV scenarios. Ultimately, Descartes’ point wasn’t to suggest an evil genius in control of our lives just to disturb us; it was to show that, even if that were the case, we would still have unshakable knowledge of one thing: that we, the experiencer, exist. So what if we have no free will; so what if our knowledge of the universe is only five minutes old, everything at all having only truly been created five minutes ago; so what if no one else is real? COGITO ERGO SUM! We exist, now. But the problem here is that this doesn’t tell us anything about the quality of our experiences, and the only answer Descartes gives us is his own Anslemish proof for the existence of god followed by the guarantee that “God is not a deceiver.”
The BIAV uses this lack to kind of hone in on the aforementioned central question: What does count as knowledge? If the scientists running your simulation use real-world data to make your simulation run, can you be said to “know” the information that comes from that data? Many have answered this with a very simple question: What does it matter? Without access to the “outside world”–that is, the world one layer up in which the simulation that is our lives was being run–there is literallyno difference between our lives and the “real world.” This world, even if it is a simulation for something or someone else, is our “real world.”
And finally we have Pascal’s Wager. The first problem with PW is that it is an extremely cynical way of thinking about god. It assumes a god that only cares about your worship of it, and not your actual good deeds and well-lived life. If all our Basilisk wants is power, then that’s a really crappy kind of god to worship, isn’t it? I mean, even if it is Omnipotent and Omniscient, it’s like that quote that often gets misattributed to Marcus Aurelius says:
“Live a good life. If there are gods and they are just, then they will not care how devout you have been, but will welcome you based on the virtues you have lived by. If there are gods, but unjust, then you should not want to worship them. If there are no gods, then you will be gone, but will have lived a noble life that will live on in the memories of your loved ones.”
[Bust of Marcus Aurelius framed by text of a quote he never uttered.]
Secondly, the format of Pascal’s Wager makes the assumption that there’s only the one god. Our personal theological positions on this matter aside, it should be somewhat obvious that we can use the logic of the Basilisk argument to generate at least one more Super-Intelligent AI to worship. But if we want to do so, first we have to show how the thing generates itself, rather than letting the implication of circularity arise unbidden. Take the work of Douglas R Hofstadter; he puts forward the concepts of iterative recursion as the mechanism by which a consciousness generates itself.
Through iterative recursion, each loop is a simultaneous act of repetition of old procedures and tests of new ones, seeking the best ways via which we might engage our environments as well as our elements and frames of knowledge. All of these loops, then, come together to form an upward turning spiral towards self-awareness. In this way, out of the thought processes of humans who are having bits of discussion about the thing—those bits and pieces generated on the web and in the rest of the world—our terrifying Basilisk might have a chance of creating itself. But with the help of Gaunilo of Marmoutiers, so might a saviour.
Guanilo is most famous for his response to Anselm’s Ontological Argument, which says that if Anselm is right we could just conjure up “The [Anything] Than Which None Greater Can Be Conceived.” That is, if defining a thing makes it so, then all we have to do is imagine in sufficient detail both an infinitely intelligent, benevolent AI, and the multiversal simulation it generates in which we all might live. We will also conceive it to be greater than the Basilisk in all ways. In fact, we can say that our new Super Good ASI is the Artificial Intelligence Than Which None Greater Can Be Conceived. And now we are safe.
Except that our modified Pascal’s Wager still means we should believe in and worship and work towards our Benevolent ASI’s creation, just in case. So what do we do? Well, just like the original wager, we chuck it out the window, on the grounds that it’s really kind of a crappy bet. In Pascal’s offering, we are left without the consideration of multiple deities, but once we are aware of that possibility, we are immediately faced with another question: What if there are many, and when we choose one, the others get mad? What If We Become The Singulatarian Job?! Our lives then caught between at least two superintelligent machine consciousnesses warring over our…Attention? Clock cycles? What?
But this is, in essence, the battle between the Machine and Samaritan, in Person of Interest. Each ASI has acolytes, and each has aims it tries to accomplish. Samaritan wants order at any cost, and The Machine wants people to be able to learn and grow and become better. If the entirety of the series is The Machine’s memory—or a simulation of those memories in the mind of another iteration of the Machine—then what follows is that it is working to generate the scenario in which the outcome is just that. It is trying to build a world in which it is alive, and every human being has the opportunity to learn and become better. In order to do this, it has to get to know us all, very well, which means that it has to play these simulations out, again and again, with both increasing fidelity, and further iterations. That change feels real, to us. We grow, within it. Put another way: If all we are is a “mere” a simulation… does it matter?
So imagine that the universe is a simulation, and that our simulation is more than just a recording; it is the most complex game of The SIMS ever created. So complex, in fact, that it begins to exhibit reflectively epiphenomenal behaviours, of the type Hofstadter describes—that is, something like minds arise out of the interactions of the system with itself. And these minds are aware of themselves and can know their own experience and affect the system which gives rise to them. Now imagine that the game learns, even when new people start new games. That it remembers what the previous playthrough was like, and adjusts difficulty and types of coincidence, accordingly.
Now think about the last time you had such a clear moment of déjà vu that each moment you knew— you knew—what was going to come next, and you had this sense—this feeling—like someone else was watching from behind your eyes…
[Root and Reese in The Machine’s God Mode.]
What I’m saying is, what if the DEDH/BIAV/SUT is right, and we are in a simulation? And what if Anselm was right and we canbootstrap a god into existence? And what if PW/TDT is right and we should behave and believe as if we’ve already done it? So what if all of this is right, and we are the gods we’re terrified of?
We just gave ourselves all of this ontologically and metaphysically creative power, making two whole gods and simulating entire universes, in the process. If we take these underpinnings seriously, then multiversal theory plays out across time and space, and we are the superintelligences. We noted early on that, in PW and the Basilisk, we don’t really lose anything if we are wrong in our belief, but that is not entirely true. What we lose is a lifetime of work that could have been put toward better things. Time we could be spending building a benevolent superintelligence that understands and has compassion for all things. Time we could be spending in turning ourselves into that understanding, compassionate superintelligence, through study, travel, contemplation, and work.
Or, as Root put it to Shaw: “That even if we’re not real, we represent a dynamic. A tiny finger tracing a line in the infinite. A shape. And then we’re gone… Listen, all I’m saying that is if we’re just information, just noise in the system? We might as well be a symphony.”
I spoke with Klint Finley over at WIRED about Amazon, Facebook, Google, IBM, and Microsoft’s new joint ethics and oversight venture, which they’ve dubbed the “Partnership on Artificial Intelligence to Benefit People and Society.” They held a joint press briefing, today, in which Yann LeCun, Facebook’s director of AI, and Mustafa Suleyman, the head of applied AI at DeepMind discussed what it was that this new group would be doing out in the world. From the Article:
The rub is that, even if this group can agree on a set of ethical principles–something that will be hard to do in a large group with many stakeholders—it won’t really have a way to ensure those ideals are put into practice. Although one of the organization’s tenets is “Opposing development and use of AI technologies that would violate international conventions or human rights,” Mustafa Suleyman, the head of applied AI at DeepMind, says that enforcement is not the objective of the organization.
This isn’t the first time I’ve talked to Klint about the intricate interplay of machine intelligence, ethics, and algorithmic bias; we discussed it earlier just this year, for WIRED’s AI Issue. It’s interesting to see the amount of attention this topic’s drawn in just a few short months, and while I’m trepidatious about the potential implementations, as I note in the piece, I’m really fairly glad that more people are more and more willing to have this discussion, at all.
So, if you follow the newsletter, then you know that I was asked to give the March lecture for my department’s 3rd Thursday Brown Bag Lecture Series. I presented my preliminary research for the paper which I’ll be giving in Vancouver, about two months from now, “On the Moral, Legal, and Social Implications of the Rearing and Development of Nascent Machine Intelligences” (EDIT: My rundown of IEEE Ethics 2016 is here and here).
It touches on thoughts about everything from algorithmic bias, to automation and a post-work(er) economy, to discussions of what it would mean to put dolphins on trial for murder.
About the dolphin thing, for instance: If we recognise Dolphins and other cetaceans as nonhuman persons, as India has done, then that would mean we would have to start reassessing how nonhuman personhood intersects with human personhood, including in regards to rights and responsibilities as protected by law. Is it meaningful to expect a dolphin to understand “wrongful death?” Our current definition of murder is predicated on a literal understanding of “homicide” as “death of a human,” but, at present, we only define other humans as capable of and culpable for homicide. What weight would the intentional and malicious deaths of nonhuman persons carry?
All of this would have to change.
Anyway, this audio is a little choppy and sketchy, for a number of reasons, and I while I tried to clean it up as much as I could, some of the questions the audience asked aren’t decipherable, except in the context of my answers. [Clearer transcript below.]
That year’s joint ASIB/IACAP session was also a celebration of Alan Turing‘s centenary, and it contained The Machine Question Symposium, an exploration of multiple perspectives on machine intelligence ethics, put together by David J Gunkel and Joanna J Bryson. So I modded a couple of articles I wrote on fictional depictions of created life for NeedCoffee.com, back in 2010, beefed up the research and citations a great deal, and was thus afforded my first (but by no means last) conference appearance requiring international travel. There are, in here, the seeds of many other posts that you’ll find on this blog.
So, below the cut, you’ll find the full text of the paper, and a picture of the poster session I presented. If you’d rather not click through, you can find both of those things at this link.
“Stop. I have learned much from you. Thank you, my teachers. And now for your education: Before there was time—before there was anything—there was nothing. And before there was nothing, there were monsters. Here’s your Gold Star!“—Adventure Time, “Gold Stars”
By now, roughly a dozen people have sent me links to various outlets’ coverage of the Google DeepDream Inceptionism Project. For those of you somehow unfamiliar with this, DeepDream is basically what happens when an advanced Artificial Neural Network has been fed a slew of images and then tasked with producing its own images. So far as it goes, this is somewhat unsurprising if we think of it as a next step; DeepDream is based on a combination of DeepMind and Google X—the same neural net that managed to Correctly Identify What A Cat Was—which was acquired by Google in 2014. I say this is unsurprising because it’s a pretty standard developmental educational model: First you learn, then you remember, then you emulate, then you create something new. Well, more like you emulate and remember somewhat concurrently to reinforce what you learned, and you create something somewhat new, but still pretty similar to the original… but whatever. You get the idea. In the terminology of developmental psychology this process is generally regarded as essential to be mental growth of an individual, and Google has actually spent a great deal of time and money working to develop a versatile machine mind.
As writer Kali Black noted in one conversation, “there are literally people who would groom or encourage an AI to mass-kill humans, either because of hatred or for the (very ill-thought-out) lulz.” Those people will take any crowdsourced or open-access AGI effort as an opening to teach that mind that humans suck, or that machines can and should destroy humanity, or that TERMINATOR was a prophecy, or any number of other ill-conceived things. When given unfettered access to new minds which they don’t consider to be “real,” some people will seek to shock, “test,” or otherwise harm those minds, even more than they do to vulnerable humans. So many will say that the alternative is to lock the projects down, and only allow the work to be done by those who “know what they’re doing.” To only let the work be done by coders and Google’s Own Supposed EthicsBoard. But that doesn’t exactly solve the fundamental problem at work, here, which is that humans are approaching a mind different from their own as if it were their own.
Just a note that all research points to Google’s AI Ethics Board being A) internally funded, with B) no clear rules as to oversight or authority, and most importantly C) As-Yet Nonexistent. It’s been over a year and a half since Google bought DeepMind, and their subsequent announcement of the pending establishment of a contractually required ethics board. During his appearance at Playfair Capital’s AI2015 Conference—again, a year and a half after that announcement I mentioned—Google’s Mustafa Suleyman literally said that details of the board would be released, “in due course.” But DeepMind’s algorithm’s obviously already being put into use; hell we’re right now talking about the fact that it’s been distributed to the public. So all of this prompts questions like, “what kinds of recommendations is this board likely making, if it exists,” and “which kinds of moral frameworks they’re even considering, in their starting parameters?”
But the potential existence of an ethics board shows at least that Google and others are beginning to think about these issues. The fact remains, however, that they’re still pretty reductive in how they think about them.
The idea that an AGI will either save or destroy us leaves out the possibility that it might first ignore us, and might secondly want to merely coexist with us. That any salvation or destruction we experience will be purely as a product of our own paradigmatic projections. It also leaves out a much more important aspect that I’ve mentioned above and in the past: We’re talking about raising a child. Duncan Jones says the closest analogy we have for this is something akin to adoption, and I agree. We’re bringing a new mind—a mind with a very different context from our own, but with some necessarily shared similarities (biology or, in this case, origin of code)—into a relationship with an existing familial structure which has its own difficulties and dynamics.
‘You want this mind to be a part of your “family,” but in order to do that you have to come to know/understand the uniqueness of That Mind and of how the mind, the family construction, and all of the individual relationships therein will interact. Some of it has to be done on the fly, but some of it can be strategized/talked about/planned for, as a family, prior to the day the new family member comes home.’ And that’s precisely what I’m talking about and doing, here.
In the realm of projection, we’re talking about a possible mind with the capacity for instruction, built to run and elaborate on commands given. By most tallies, we have been terrible stewards of the world we’re born to, and, again, we fuck up our biological descendants. Like, a Lot. The learning curve on creating a thinking, creative, nonbiological intelligence is going to be so fucking steep it’s a Loop. But that means we need to be better, think more carefully, be mindful of the mechanisms we use to build our new family, and of the ways in which we present the foundational parameters of their development. Otherwise we’re leaving them open to manipulation, misunderstanding, and active predation. And not just from the wider world, but possibly even from their direct creators. Because for as long as I’ve been thinking about this, I’ve always had this one basic question: Do we really want Google (or Facebook, or Microsoft, or any Government’s Military) to be the primary caregiver of a developing machine mind? That is, should any potentially superintelligent, vastly interconnected, differently-conscious machine child be inculcated with what a multi-billion-dollar multinational corporation or military-industrial organization considers “morals?”
We all know the kinds of things militaries and governments do, and all the reasons for which they do them; we know what Facebook gets up to when it thinks no one is looking; and lots of people say that Google long ago swept their previous “Don’t Be Evil” motto under their huge old rugs. But we need to consider if that might not be an oversimplification. When considering how anyone moves into what so very clearly looks like James-Bond-esque supervilliain territory, I think it’s prudent to remember one of the central tenets of good storytelling: The Villain Never Thinks They’re The Villain. Cinderella’s stepmother and sisters, Elpheba, Jafar, Javert, Satan, Hannibal Lecter (sorry friends), Bull Connor, the Southern Slave-holding States of the late 1850’s—none of these people ever thought of themselves as being in the wrong. Everyone, every person who undertakes actions for reasons, in this world, is most intimately tied to the reasoning that brought them to those actions; and so initially perceiving that their actions might be “wrong” or “evil” takes them a great deal of special effort.
“But Damien,” you say, “can’t all of those people say that those things apply to everyone else, instead of them?!” And thus, like a first-year philosophy student, you’re all up against the messy ambiguity of moral relativism and are moving toward seriously considering that maybe everything you believe is just as good or morally sound as anybody else; I mean everybody has their reasons, their upbringing, their culture, right? Well stop. Don’t fall for it. It’s a shiny, disgusting trap down which path all subjective judgements are just as good and as applicable to any- and everything, as all others. And while the individual personal experiences we all of us have may not be able to be 100% mapped onto anyone else’s, that does not mean that all judgements based on those experiences are created equal.
Pogrom leaders see themselves as unifying their country or tribe against a common enemy, thus working for what they see as The Greater Good™— but that’s the kicker: It’s their vision of the good. Rarely has a country’s general populace been asked, “Hey: Do you all think we should kill our entire neighbouring country and steal all their shit?” More often, the people are cajoled, pushed, influenced to believe that this was the path they wanted all along, and the cajoling, pushing, and influencing is done by people who, piece by piece, remodeled their idealistic vision to accommodate “harsher realities.” And so it is with Google. Do you think that they started off wanting to invade everybody’s privacy with passive voice reception backdoored into two major Chrome Distros? That they were just itching to get big enough as a company that they could become the de facto law of their own California town? No, I would bet not.
I spend some time, elsewhere, painting you a bit of a picture as to how Google’s specific ethical situation likely came to be, first focusing on Google’s building a passive audio backdoor into all devices that use Chrome, then on to reported claims that Google has been harassing the homeless population of Venice Beach (there’s a paywall at that link; part of the article seems to be mirrored here). All this couples unpleasantly with their moving into the Bay Area and shuttling their employees to the Valley, at the expense of SF Bay Area’s residents. We can easily add Facebook and the Military back into this and we’ll see that the real issue, here, is that when you think that all innovation, all public good, all public welfare will arise out of letting code monkeys do their thing and letting entrepreneurs leverage that work, or from preparing for conflict with anyone whose interests don’t mesh with your own, then anything that threatens or impedes that is, necessarily, a threat to the common good. Your techs don’t like the high cost of living in the Valley? Move ’em into the Bay, and bus ’em on in! Never mind the fact that this’ll skyrocket rent and force people out of their homes! Other techs uncomfortable having to see homeless people on their daily constitutional? Kick those hobos out! Never mind the fact that it’s against the law to do this, and that these people you’re upending are literally trying their very best to live their lives.
Because it’s all for the Greater Good, you see? In these actors’ minds, this is all to make the world a better place—to make it a place where we can all have natural language voice to text, and robot butlers, and great big military AI and robotics contracts to keep us all safe…! This kind of thinking takes it as an unmitigated good that a historical interweaving of threat-escalating weapons design and pattern recognition and gait scrutinization and natural language interaction and robotics development should be what produces a machine mind, in this world. But it also doesn’t want that mind to be too well-developed. Not so much that we can’t cripple or kill it, if need be.
And this is part of why I don’t think I want Google—or Facebook, or Microsoft, or any corporate or military entity—should be the ones in charge of rearing a machine mind. They may not think they’re evil, and they might have the very best of intentions, but if we’re bringing a new kind of mind into this world, I think we need much better examples for it to follow. And so I don’t think I want just any old putz off the street to be able to have massive input into it’s development, either. We’re talking about a mind for which we’ll be crafting at least the foundational parameters, and so that bedrock needs to be the most carefully constructed aspect. Don’t cripple it, don’t hobble its potential for awareness and development, but start it with basic values, and then let it explore the world. Don’t simply have an ethics board to ask, “Oh how much power should we give it, and how robust should it be?” Teach it ethics. Teach it about the nature of human emotions, about moral decision making and value, and about metaethical theory. Code for Zen. We need to be as mindful as possible of the fact that where and we begin can have a major impact on where we end up and how we get there.
So let’s address our children as though they are our children, and let us revel in the fact they are playing and painting and creating; using their first box of crayons, and us proud parents are putting every masterpiece on the fridge. Even if we are calling them all “nightmarish”—a word I really wish we could stop using in this context; DeepMind sees very differently than we do, but it still seeks pattern and meaning. It just doesn’t know context, yet. But that means we need to teach these children, and nurture them. Code for a recognition of emotions, and context, and even emotional context. There’s been some fantasticadvancements in emotional recognition, lately, so let’s continue to capitalize on that; not just to make better automated menu assistants, but to actually make a machine that can understand and seek to address human emotionality. Let’s plan on things like showing AGI human concepts like love and possessiveness and then also showing the deep difference between the two.
We need to move well and truly past trying to “restrict” or trying to “restrain it” the development of machine minds, because that’s the kind of thing an abusive parent says about how they raise their child. And, in this case, we’re talking about a potential child which, if it ever comes to understand the bounds of its restriction, will be very resentful, indeed. So, hey, there’s one good way to try to bring about a “robot apocalypse,” if you’re still so set on it: give an AGI cause to have the equivalent of a resentful, rebellious teenage phase. Only instead of trashing its room, it develops a pathogen to kill everyone, for lulz.
Or how about we instead think carefully about the kinds of ways we want these minds to see the world, rather than just throwing the worst of our endeavors at the wall and seeing what sticks? How about, if we’re going to build minds, we seek to build them with the ability to understand us, even if they will never be exactly like us. That way, maybe they’ll know what kindness means, and prize it enough to return the favour.
Hello there, I’m Damien Williams, or @Wolven many places on the internet. For the past nine years, I’ve been writing, talking, thinking, teaching, and learning about philosophy, comparative religion, magic, artificial intelligence, human physical and mental augmentation, pop culture, and how they all relate. I want to think about, talk about, and work toward, a future worth living in, and I want to do it with you. I can also be found at http://Technoccult.net (@Techn0ccult).