Tag Archives: A Legal Theory for Autonomous Artificial Agents

We Robot 2012 – Day One

I am posting today from the University of Miami Law School, which is staging the We Robot 2012 conference. I presented and discussed Patrick Hubbard’s (University of South Carolina Law School) Regulation of Liability for Risks of Physical Injury From “Sophisticated Robots”. Presenting someone else’s work presents a difficult challenge; thanks to being an academic I have perfected the dark arts of bullshitting about my own work but doing so about someone else’s work is far more difficult.  I tried my best to present Patrick’s work as comprehensively and fairly as possible and to raise some questions that could spur on some discussion. (I will place the slides online very soon so you can see what I got up to.)

One of the points I raised in response to Patrick’s claim that robots that displayed ‘emergent behavior’ would occasion changes in tort doctrine was: How should we understand such emergence? Might we need to see if robots, for instance, displayed  stability, homeostasis and evolvability–all often held to be features of living systems, paradigmatic examples of entities that display emergent behavior. Would robots be judged to display emergent behavior if it was not just a function of its parts but also of the holistic and relational properties of the system. I also asked Patrick how the law should understand autonomy given that some philosophical definitions of autonomy–like Kant’s for instance–would rule out some humans as being autonomous. (Earlier in the morning during discussions in another talk, I suggested another related benchmark that could be useful: Draw upon the suggestion made in Daniel Dennett’s The Case for Rorts that robots  could be viewed as intentional agents when we trust robots as authorities in reporting on their inner states, when its programmers and designers  lose epistemic hegemony.) An interesting section of the discussion that followed my presentation centered on how useful analogizing robots to animals or children or other kinds of entities was likely to be, and if useful, which analogies could work best. (This kind of analogizing was done in Chapter 4 of A Legal Theory of Autonomous Artificial Agents.)

Earlier in the day in discussing automated law enforcement–perhaps done by fleets of Robocops–I was glad to note that one of its positive outcomes was highlighted: that such automation could bring about a reduction of bias in law enforcement. In my comment following the talk, I noted that a fleet of Robocops aware of the Fourth Amendment might be be very welcome news for all those who were the targets of the almost seven hundred thousand Stop-n-Frisk searches in New York City.

As was noted in discussions in the morning, some common threads have already emerged: the suggestion that robots are ‘just tools,’ (which I continue to find bizarre), the not-so-clear distinction–and reliance on–true and apparent autonomy, the concerns about the need to avoid ‘projecting’ human will and agency onto robots and treating them like people (i.e., that we need to avoid the so-called ‘android fallacy.’) I personally don’t think warnings about the android fallacy are very useful; contemporary robots are not sophisticated enough to be people and there is no impossibility proof against them being sophisticated enough to be persons in the future.

Hopefully, I will have another–much more detailed–report from this very interesting and wonderfully well-organized conference tomorrow. (I really haven’t done justice to the rich discussions and presentations yet; for that I need a little more time.)

Kraftwerk Makes Us Tell The Truth: We Are The Robots?

Kraftwerk’s The Robots has been an electro-pop classic ever since its release–on Kraftwerk’s classic seventh album, The Man-Machine–in 1978. My brother and I discovered Kraftwerk at roughly the same time, and, like many other schoolboys, quickly became entranced by its revolutionary blend of synthesizers, vocoders, and electronic percussion.  Some thirty years on, I still get a kick out of strapping on the earphones for The Robots (and turning up the volume to eleven); I don’t dance to it but the temptation never quite goes away. (I’ve only seen The Robots  performed live once, when I saw Kraftwerk at Sydney’s Enmore Theater in January 2003.)

Besides triggering the urge to flop around in slightly demented fashion, there are two juvenile fantasies of mine that The Robots gives comfort and succor to: One, as as part of a grand book tour for A Legal Theory for Autonomous Artificial Agents, I would make a presentation centered on the book that would feature The Robots playing in the background as I walked on stage (I don’t need smoke machines or lasers). And, wouldn’t it pretty nifty if I could get an e-book version that would play The Robots when the book file was first opened? Trust me, I spend time thinking about this stuff.

(In the summer of 2006, I played The Robots for Brooklyn high-school students at the conclusion of a summer ‘camp’ that had introduced them to, among other things, robotics and cryptography. I had  taught the cryptography track but thought the young folks that had worked on robotics kits would appreciate both the track and the fact that computers and music were connected in ways other than downloading. I’m not sure it went down all that well; most of the students in attendance found the sound perplexing, so at least for that generation, or that demographic, the track had not aged well nor come across as relevant.)

There are many good versions of The Robots out there; this soundboard-recording from a Birmingham (UK) concert on 15 July 1991, from the Dynamo Deutschland CDs is particularly good. The Russian lines “Я твой слуга” (Ya tvoi sluga, I’m your servant) and “Я твой работник” (Ya tvoi rabotnik, I’m your worker) come across particularly clearly; the lyrics in this live version are also slightly, interestingly, different from the standard lyrics. (Since this is a soundboard recording there are also some irritating sections where concert-goers can be heard talking!)

The most memorable part of this live track are the sing-along chants, between 1:55 and 2:15, for the chorus “We are the Robots”. In performing the sing-along so vigorously, the Birmingham concert-goers perhaps make two kinds of statements: one, an acknowledgement, in this hyper-corporatized and industrialized age, of the enduring relevance of the two lines in Russian quoted above; and second, a vocalized bridging of the gap between the robots and themselves, perhaps even a joining of communities. The former is appropriately disturbing, but the latter at least can be optimistically read as a denial of difference. (As I often sought to remind my interlocutors during the recent online symposium on my book, we are often more like robots than we might imagine.)

Report on Concurring Opinions Symposium on Artificial Agents – II

Today, I’m continuing my wrap-up of the Concurring Opinions online symposium on A Legal Theory for Autonomous Artificial Agents. I’ll be noting below the various responses to the book and point to my responses to them (Part I of this wrap-up was posted yesterday).

While almost all respondents seem to have seriously engaged with the book’s analysis, Ryan Calo wrote a disappointingly unengaged, and at times, patronizing post that ostensibly focused on the book’s methodological adoption of the intentional stance; it seemed to suggest that all we were doing was primitive anthropomorphizing. This was a pretty comprehensive misread of the book’s argument, so I struggled to find anything to say in response. Calo also said he didn’t know whether an autonomous robot was like a hammer or not; this was a bizarre admission coming from someone that is concerned with the legal implications of robotics. I noted in one of my responses that figuring out the answer to that question can be aided by some intuition-tickling questions (Like: Would NASA send a hammer to explore Mars? Can hammers drive?). Calo’s follow-up post to my comment on his post was roughly along the lines of “We don’t know what to do with artificial agents.” Well, yes, but I thought the point was to evaluate the attempt currently mounted in our book? I didn’t quite understand the point of Calo’s responses: that we don’t have a comprehensive theory for artificial agents i.e., the book’s title is misleading? I could be persuaded into mounting a guilty plea for that. But the point of the book was to indicate how existing doctrines could be so suitably modified to start accommodating artificial agents- that is how a legal theory will be built up in a common law system.

Deborah DeMott (Duke) whose writings on the common law doctrines of agency were very useful in our analysis in the book offered a very good analysis of our attempts to apply that doctrine to artificial agents. While DeMott disagreed with the exactness of the fit, she seemed not to think that it was completely off-base (she certainly found our attempt “lively and ingenious”!); in my response I attempted to clarify and defend some of our reasons for why we thought agency doctrine would work with artificial agents.

Ken Anderson (American University, Volokh Conspiracy) then discussed our treatment of intentionality and deployment of the intentional stance, and queried whether we intended to use the intentional stance merely as a heuristic device or whether we were, in fact, making a broader claim for intentionality in general. In my response I noted that we wanted to do both: use it as a methodological stance, and in doing so, also point an investigative lens at our understanding of intentionality in general. Ken’s reaction was very positive; he thought the book had hit a “sweet spot” in not being excessively pie-in-the-sky while offering serious doctrinal recommendations.

Ian Kerr (Ottawa), in his response, didn’t feel the book went far enough in suggesting a workable theory for artificial agents; if I understood Ian correctly, his central complaint was that the theory relied too much on older legal categories and doctrines and that artificial agents might need an entirely new set of legal frameworks. But Ian also felt the slow and steady march of the common law was the best way to handle the challenges posed by artificial agents. So, interestingly enough, I agree with Ian; and I think Ian should be less dissatisfied than he is; our book is  merely the first attempt to try and leverage the common law to make these steps to work towards a more comprehensive theory. In fact, given rapid developments in artificial agents, the law is largely going to be playing catchup more than anything else.

Andrew Sutter then wrote a critical, rich response, one that took aim at the book’s rhetoric, its methodology, and its philosophical stance. I greatly enjoyed my jousting with Andrew during this symposium, and my response to his post–and to his subsequent comments–in which I attempted to clarify my philosophical stance and presuppositions, will show that.

Harry Surden (Colorado) wrote a very good post on two understanding of artificial intelligence’s objectives–intelligence as the replication of human cognitive capacities by either replicating human methods of achieving them or via simulations that utilize other techniques–and how these could or would be crucial in the legal response to its achievements. My response to Surden acknowledged the importance of these distinctions and noted that this should also cause us to think about how we often ascribe human cognition a certain standing that arises largely because of a lack of understanding of its principles. (This also provoked an interesting discussion with AJ Sutter.)

Andrea Matwyshyn wrote an excellent, seriously engaged post that took head-on, the fairly detailed and intricate arguments of Chapter 2 (where we offer a solution for the so-called contracting problem by offering an argument that artificial agents be considered legal agents of their users). My response to Matwyshyn acknowledged the force of her various critical points while trying to expand and elaborate the economic incentivizing motivation for our claim that artificial agents should be considered as non-identical with their creators and/or deployers.

Once again, I am grateful to Frank Pasquale and the folks over at Concurring Opinions for staging the symposium and to all the participants for their responses.

Report on Concurring Opinions Symposium on Artificial Agents – I

The Concurring Opinions online symposium on my recently-released book A Legal Theory for Autonomous Artificial Agents (University of Michigan Press, 2011) wrapped up yesterday. The respondents to the book blogged on it from Tuesday till Thursday last week; from Friday till Monday I spent most of my time putting together responses to the excellent responses offered by the participants; I also replied to comments made by blog readers (two of whom, Patrick S. O’Donnell and AJ Sutter, provided very thoughtful and critical commentary).

Frank Pasquale (Seton Hall) organized the symposium and announced it on the blog on February 2nd.  The symposium was kicked off by Sonia Katyal (Fordham) who responded to the book’s argument for legal personhood for artificial agents. While positive in her response, Katyal was curious about whether a strong enough case for legal personhood had been made yet (compared to the historical case for corporations for instance). (This was useful in helping me think about how such a legal-empirical case could be made for artificial agents’ legal personhood, something I alluded to in my response.)

James Grimmelmann (New York Law School) then followed up with a post that addressed the law’s response to complex systems and pointed out that responding to the presence of artificial agents could or would draw upon some of those patterns of response. (Sonia and James had started things a little early so my introductory post on artificial agents showed up after theirs!) James also wrote a follow-up to his first piece, which further elaborated on some of law’s strategies for dealing with complexity, pointing out the grant of personhood was not inevitable. These posts were very useful in illustrating the law’s pragmatic stance towards the presence of complex systems. (Danielle Citron (Maryland), incidentally, wrote a reminder of how automated decision making has been causing a headache for administrative law; in the original version of our book we had begun work on a chapter that addressed this but left it on the cutting floor; it would be good to resurrect that at some point.)

Lawrence Solum (Georgetown and Illinois), who has been writing at the intersection of philosophy and law for many years, then wrote a post suggesting that some dimensions of the problem of artificial agents’ legal personhood could be illustrated by a thought experiment involving zombies.  (I drew upon this thought experiment with another one of my own: how would we respond to extraterrestrials that petitioned for legal personhood?)

Frank Pasquale then pointed out how bots were being used for political campaigning and could be said to be contributing to political speech; this was really quite a provocative and fascinating post and I regret not having addressed it over at CO in my responses. I will do so soon here.

Ugo Pagallo (Georgetown and Turin), staying with the legal personhood theme, then questioned several aspects of our personhood argument, (while agreeing with our agency analysis in earlier parts of the book). In my response to Ugo, I suggested we were in greater argument than it might have originally seemed. Ramesh Subramanian (Yale ISP and Quinnipiac), meanwhile, took the argument for legal personhood seriously, and wondered more broadly about what some of its futuristic implications could be.

I will have another post tomorrow with summaries and descriptions of the various responses and the discussions that followed. This was an exhausting and invigorating experience in more ways than one.

Artificial Agents, Knowledge Attribution, and Privacy Violations

I am a subscriber to a mailing list dedicated to discussing the many legal, social, and economic issues that arise out of the increasing use of drones. Recently on the list, the discussion turned to the privacy implications of drones. I was asked whether the doctrines developed in my book A Legal Theory of Autonomous Artificial Agents were relevant to the privacy issues raised by drones. I wrote a brief reply on the list indicating  that yes, they are.  I am posting a brief excerpt from the book here to address that question more fully (for the full argument, please see Chapter 3 of the book):

Knowledge Attribution and Privacy Violations

The relationship between knowledge and legal regimes for privacy is straightforward: privacy laws place restrictions, inter alia, on what knowledge may be acquired, and how.  Of course, knowledge acquisition does not exhaust the range of privacy protections  afforded under modern legal systems. EU privacy law, for example, is triggered when mere processing of personal data is involved. Nevertheless acquisition of knowledge of  someone’s affairs, by human or automated means, crosses an important threshold with regards to privacy protection.

Privacy obligations are implicitly relevant to the attribution of knowledge held by agents to their principals in two ways: confidentiality obligations can restrict such attribution and horizontal information barriers such as medical privacy obligations can prevent corporations being fixed with collective knowledge for liability purposes.

Conversely, viewing artificial agents as legally recognized “knowers” of digitized personal information on behalf of their principals brings conceptual clarity in answering the question of when automated access to personal data amounts to a privacy violation.

The problem with devising legal protections against privacy violations by artificial agents is not that current statutory regimes are weak; it is that they have not been interpreted appropriately given the functionality of agents and the nature of modern internet-based communications. The first move in this regard is to regard artificial agents as legal agents
of their principals capable of information and knowledge acquisition.

A crucial disanalogy drawn between artificial and human agents plays a role in the denial that artificial agents’ access to personal data can constitute a privacy violation: the argument that the automated nature of artificial agents provides reassurance sensitive personal data is “untouched by human hands, unseen by human eyes.” The artificial agent becomes a convenient surrogate, one that by its automated nature neatly takes the burden of responsibility off the putative corporate or governmental offender. Here the intuition that “programs don’t know what your email is about” allows the principal to put up an “automation screen” between themselves and the programs deployed by them. For
instance, Google has sought to assuage concerns over possible violations of privacy in connection with scanning of Gmail email messages by pointing to the non-involvement of humans in the scanning process.

Similarly, the U.S. Government, in the 1995 Echelon case, responded to complaints about its monitoring of messages flowing through Harvard University’s computer network by stating no privacy interests had been violated because all the scanning had been carried out by programs.

This putative need for humans to access personal data before a privacy violation can occur underwrites such defenses.

Viewing, as we do, the programs engaged in such monitoring or surveillance as legal agents capable of knowledge acquisition denies the legitimacy of the Google and Echelon defenses. An agent that has acquired user’s personal data acquires functionality that makes possible the processing or onward disclosure of that data in such a way as to constitute privacy violations. (Indeed, the very functionality enabled by the access to such data is what would permit the claim to be made under our knowledge analysis conditions that the agent in question knows a user’s personal data.)