welcome everybody my name is jontus Euless I'm the director of The Institute for ethics and Ai and our topic in this colloquium is AI in the law or AI in place of Law and hopefully there'll be an answer at the end um I'm really delighted to welcome our main speaker Gerald J posterman who is the boshima professor of philosophy and professor of law at UNC Chapel Hill he's one of the world's leading philosophers of law and the author of a massive study of Jeremy Bentham and the common law tradition second edition of which came out in 2019 he's also the author of many fine articles in legal and political philosophy and most recently he's the author of a book which is an exciting event in the world of jurisprudence a book called laws rule the nature value and viability of the rule of law there was a quite famous Oxford legal philosopher who recently said people shouldn't bother writing about the rule of law because it's all been said but this book proves that this isn't the case that these fundamental topics have to be Revisited in light of new concerns so one of the things that I was very delighted to see was Jerry's book has a discussion of equity and mercy in that context which were topics that were really ignored in recent years and there's also two topics on the way the digital Revolution influences the rule of law and that will be the focus of today's discussion so Jared's just a real pleasure to have you here we have three fantastic commentators as well first of which is Dr Natalie Byram she's the director of research at the legal Education Foundation and the founder and director of Justice lab which is a policy evidence Center using data and evidence to tackle the most pressing problems facing the justice system Natalie is a very impressive person the kind of person that when you meet her you wonder whether you've wasted your life on trivia she's very passionately committed to using data to enhance the quality of legal justice and she gave a talk once where she mentioned one of the people whose opinions were sought this is someone who was as it were the sharp end of the legal system and she said they think we don't care but they don't ask us if we do and I think if the digital revolution has any kind of redeeming moral aspect to it one aspect of that would be enabling Ordinary People to have a voice about matters such as the justice system that can affect them in quite dramatic ways second commentator is Professor Timothy Endicott um who has the unique distinction of being both a legal a leading legal philosopher and also a leading administrative lawyer Timothy is the holder of the prestigious finerian chair of English law at All Souls College here in Oxford it reminds me a little bit of the famous Oxford philosopher J.L Austin when he heard that Herbert Hart had been elected to the chair of jurisprudence heart at that point was a philosophy tutor he said what's Splendid news another Province falls to the Empire of philosophy but Timothy's gone one further I mean ordinary lawyers don't care about the chair of jurisprudence but boy do they care about the vinearian chair so to see a philosopher Tim occupying that chair it just is the best news imaginable for US legal philosophers here and of course Timothy is the author of vagueness in law which was published in 2000 and administrative law published in 2021 and last but not least we have Professor Sarah Green who is the UK law commissioner for commercial and common law she's held that post since January 2020 kind of pandemic exercise I guess she's previously held academic posts at Bristol Oxford and Birmingham so we're delighted to welcome her back before her academic career she was a software consultant at Accenture and Professor Green has written books and articles on many issues at the intersection of Law and Technology such as virtual currencies smart contracts and blockchain issues regarding intermediated Securities so the way we're going to proceed is we're going to have Jerry talk to us for about 30 minutes and then I'll invite each of the speakers to give a comment of about five minutes which I will strictly time and then we'll throw it open to a general discussion including questions from people who are following us on YouTube so if you are following us please use the chat function to do that but now it's my great pleasure to turn over to Jerry what a pleasure it is to be here um I've not ever been in this particular venue um but I have been around Oxford from time to time I spent I arrived here on Monday and I spent a few days trying to recover memories um this comes as something like experiencing an earthquake but I realized well here that it is actually 50 years ago that I was a graduate student here I hate to actually mention that but um I'm humbled by that but I'm also humbled by the fact that these wonderful colleagues have agreed to to help me think about AI I don't regard myself as as an expert in anything I'm a philosopher so I'm an expert in everything but actually in nothing um and so it's really great to have such accomplished colleagues as well I want to begin my discussion with a disclaimer this talk was not written by chat GPT at least they would like you to think that um I wrote this chapter of my book published now in November but um I wrote this chapter in the finished it in the very early months perhaps the late months of 21 the early months of 22. um in clock time that's at most 18 months maybe 15 months in technology time it's a century um chat Bots were not known at least to me back then um and so I I the AI in this field is moving at warp speed so much so that the um the very creators of um chatbots and the like um Elon Musk amongst others have called for moratorium and now I read yesterday that that the letter calling for that moratorium has acquired 27 000 signatures lots of people are worried their worries tend to be various kinds there's the standard ones we've been thinking about worrying about for a long time the radical epacity of um AI operations and algorithms the um potential bias and partiality that's that that seems to be um plaguing it or could plague it more recently concerns about AI getting into in chat Bots and the like getting into the hands of Bad actors and creating um even worse problems of disinformation than we currently face um perhaps some people are talking also about loss of jobs of various kinds and and at one point even I've heard people talk just which it's getting out of control we have lost control um my concern is a little different at least the concern I want to address today my concern rather is with proposals for what's called artificial legal intelligence using generative AI to do legal work um and the thought here is that AI devices machine learning algorithms and all that um should be introduced to do the work of lawyers officials judges and the like replacing them across the board and the question for us is should we welcome this or resist it as a degradation of our social and political and academic and legal lines I want to set this concern in the context of the idea of the rule of law I wrote this chapter after having I spent a good bit of time trying to articulate a conception of the rule of law and to defend it and then I thought I should raise some serious address some serious challenges to the ideal of rule of law as I propose to understand it and one of them is the claim one of the challenges is the claim that there's no room for the rule of law in the international domain the transnational domain there's another that says there's no place for Equity or Mercy in the rule of law we talked about this yesterday but also the developments in Ai and and Advanced Technologies are such that they present potentially challenges to the rule of law as well and I wanted to pay attention to that what I propose to do today um very very briefly unfortunately we could spend you know weeks on this but I'd like to do very briefly is to First sketch in the briefest terms just enough of my proposed conception of the rule of law that you give some idea where the the where I'm coming from when I ask questions about artificial intelligence and it's potential artificial legal intelligence then I want to sketch what I take to be some characteristics of artificial legal intelligence or what I call computational intelligence and then go on to ask the question what a value would be lost if we were to replace lawyers and judges and the rest of us who do legal reasoning what would belly be lost if we were to do that um and the answer is not obvious I think and I want to spend some time doing that so first of all my conception of the rule of my conception I want propose a conception of the rule of law and it has these features first of all um one needs to get a a handle on of the central concern that animates the rule of law that concern I think has always been historically has been the exercise of power and what my friend Mark Martin Krieger says is the special disease of the exercise of power namely it's um tendency to be exercised arbitrarily and the rule of law proposes law is the cure for that disease or a cure for that disease so here's my slogan um the rule of law is robust in a political Community when it provides protection and recourse against the arbitrary exercise of power through the distinctive instrumentalities of law distinctive tools of law so we have an aim tempering the exercise of power and we have a means law that's the core idea this rule of law then provides a bulwark for protection of the powerless a bridle on the exercise of Power by those who are powerful and importantly a bond that brings us and holds us together in a political community is very important to see because um power those who wield and exercise power are not required by the rule of law simply to submit to the law they must acknowledge that others are in a position to assess their activities with respect to that so accountability plays The crucial role in the idea of the rule of law it's not enough to have root laws there to have principles to have standards there must be framework within which those who exercise power are held accountable to the world the bond is important because the rule of law makes possible and serves I think a compelling vision of what it is and can be for us to live together in a political community so this ideal can be articulated in terms of Three core principles I'll call the first sovereignty of law second equality the third Fidelity sovereignty of law the main idea here is law alone rules law rules that means that the exercise of ruling power governing power is legitimate just insofar as it's authorized by ordained by the law but also those who exercise power must do so through the law by means of the law and thirdly it means that those who govern by or through law must themselves be governed by it I call that reflexivity that's what I mean by the sovereignty of law equality very quickly we won't spend any time on this today equality here doesn't mean equality of things but it means a quality of standing you might call I think of equality here as a kind of um position Visa the others and so um the the notion at the core of the rule of law of equality is equality in the eyes of the law by which I think we should mean those who are bound by the law must equally enjoy his protection and recourse thirdly and this is a part of my conception that I think is unusual um I insist that um the the rule of law be underwritten by a certain kind of ethos the problem is this law sovereignty of law requires that law rule but laws don't rule people do Plato's told us that and so how do we understand this idea of Law's ruling being the only ruler and the answer in in my view is that the law rules just in so far as there is in a political Community a robust ethos a set of responsibilities and relationships rooted in core convictions and commitments According to which each of us must take responsibility for holding each other and especially those who wield legal power wheel governing power accountable to the law it's a it's a community-wide responsibility Fidelity then is I think the animating animating Spirit of the rule of law the rule of law is a important moral political ideal and it has no native Force or something like claim but what you might want to ask on what basis on what ground I mean it makes important demands on those who exercise power it makes important demands on those of us who are to hold them accountable by what right and so I do spend a chapter in which I try to explain what it is that the rule of law this principle serves they're more fundamental moral value or principle and here what I propose is that the rule of law makes possible and serves a compelling vision of the community and of individuals place in it it brings under one normative root actually four different related closely related moral values one being um equality now understood as peer Hood having the status of being able to see oneself and others as bound together as peers secondly individual dignity third freedom from domination or subordination and fourth a sense of community that binds us together those are not independent um values they are brought all together into a single conception of of a of a community and I call that conception of value membership on the model of friendship being a kind of important value for people membership binds together in in interesting ways these four different but closely related values law when it meets the demands of the rule of law law when it meets the demands of the rule of law plays a key role in securing and sustaining and nurturing communities political communities that aspire to realize this value of membership so much so far well the law is supposed to provide some kind of resources for holding those who exercise power accountable um resources for tempering um the the exercise of ruling power what resources does the law provide what tools are in laws toolbox um I want to mention six although I spent a whole lot more time articulating them I'm doing really quickly one these are Salient features of law that I think are in the law in the Lost toolbox and help to serve the aim of the rule of law one law is normative which I mean law orders social behavior structure social behavior um according to Norms IT addresses Norms to persons who take those them up into their practical deliberation so they are its law is normative first of all it's not it order it it orders um Behavior through these norms and not in the way you might say that the Snake Charmer orders the behavior of the snake um it doesn't goad it doesn't simply cause IT addresses Norms to those who are capable of grasping them and um inter in integrating them into their own practical deliberation secondly important too law is positive not in the sense that it's you know bright and sunny but in the sense that it is the product of of human activity embedded in historical processes and the like thirdly law must be public in a couple of Senses one most important for our purposes um it must be public in the sense that it's addressed in the public to a public that is two people in a public in a community not individually to each party but it's addressed to a public a public that's composed of um that uh many individuals that are engaged in social interactions at various levels it's addressed to the public fourth law puts at the disposal of us all a disciplined practice of deliberative reasoning Law's rule is the rule of deliberative reasoning um that is to say law doesn't govern by reason in an abstract sense law governs through the process of reasoning and that reasoning process is modeled in institutions from with which we're familiar mainly courts and the like but those are models of an activity that goes on outside of those forums as well um and even in the public domain this reasoning for this reasoning it's not merely a matter of output that Accords with rules rather this is reasoning that follows rules and Norms intelligently appreciating that rules and Norms are reasons and can operate his reasons and as potential justifications for actions fifth law essentially involves judgment to recognize utterances first of all we need judgment in order to recognize utterances as reasons as rules and the rules as providing reasons for Action we need judgment to bridge the gap between General Norms or rules and particular applications thirdly judgment engages the world the world around us the world and its and the affordances it provides that it um helps those who reason legally reason with the law and in the law allows them to follow norms intelligently the law involves judgment um in order to assess the relevance of evidence we need judgment to assess the relevant evidence and the and the relative credibility of those who give testimony um and finally it needs judgment it requires judgment in order to decide what to do we end we end up our process of legal reasoning with requirements on what we should do the question though is should we do it what should we do judgment is always involved in making that move from what the law requires or it permits us to do to what we should do and finally law has an essentially moral Dimension I don't mean by that that law necessarily makes appeal to moral principles rather what I mean is that reasoning persons are implicated in the conclusions they draw and the actions they take on the basis of the collusion conclusions they draw they must bear responsibility for those conclusions and for their actions and they are liable to others who will hold them accountable for it so the moral Dimension is a dimension of responsibility tied to the overall concern of accountability so now that says a little bit about laws tools and maybe a little bit about what it is to reason like a lawyer um let's contrast let me compare that with we won't contrast yet let me compare that with what I call computational intelligence AI devices algorithms machine learning large language um models and the like take information that we have and generate information that we don't have it's sort of like Espionage intelligence right you you get new information from this Espionage intelligence and this information can be useful for decision making but it's not a decision it's information now it's often said that these devices make predictions but that's a little bit misleading because what in fact these devices do on the whole is uncover correlations based on statistical analysis so chat Bots are in a sense if I can be really crude about it forgive me they are very sophisticated forms of autocomplete third um the output from such um computational intelligence um to put it really crudely Bears little relationship to truth or to put it a little bit more um kindly it has little relationship to the world and all its complexity it works on data on various kinds that can be found on the internet and elsewhere um it doesn't appreciate values and norms they don't appreciate reasons as reasons um the problem isn't that so there's a lot of talk lately about the problem of alignment that these AI devices are not tracking our values in some way think the problem that's the problem I'm concerned with here is not alignment it's appreciation they don't see reasons as providing justifications so large language models can reproduce arguments parrot them but they can't recognize them as articulations of values and norms or weigh them as reasons we might give that have some moral significance um they're curators of opinions they don't engage them as opinions now it's also true that a lot of the correlations uncover important patterns patterns of similarity and that often leads to further kinds of applications and uncovering of the kinds of information um and one it might be inclined to think of this as a kind of analogical thinking thinking reasoning by analogy but I think that's not quite right because to determine the relevance of similarity of one group of things ideas models with the like examples and another the comparables and the the the Target and the source um the way in which they connect them is by way of counting how many similarities but analogical reasoning is a reasoning by which we see the connection between the source and the target as having some connection that makes practical or legal or maybe moral sense so analogical reasoning evaluates these similarities and relevance is then determined not by numbers um but by what makes sense um but so to sum this up Computing doesn't perform legal reasoning just faster and more efficiently in my view it simulates it by doing something altogether different there's no Lex X machina so simulation replaces argumentation calculation replaces deliberative reasoning I would also mentioned very briefly the um the approach that um those who defend AI in the AI artificial legal intelligence have speak of social ordering by artificial intelligence in two different ways perhaps more than that um I'll just mention two one is by way of hyper particularized personal directives they're tailored with Precision to individual actions and circumstances in light of the peculiar particular features of each individual the model here is uh as it were a one to one um the the director Direction giver and the direction receiver I'm brought to mind of hobbs's model of um the activity of the Sovereign The Sovereign was one who gives commands to each individual um notice it bypasses deliberation and goes directly to action the other um I'm drawing here on the work of Lawrence lessig he distinguishes between them the mode of social ordering I'm calling normative with what he calls architecture that's his term of Art um architecture shapes and channels Behavior through non-deliberative means so unlike scripts and recipes well let me give you an example the example he gives is speed bumps on a neighborhood Road they they get you to go slower but they don't do it by saying speed limit 20 miles per hour and they don't even say things like drive here as if your kids lived here right what they do is they put obstacles in the way if you're doing so when in in my town in Durham North Carolina there are streets that have just enough room for one traffic lane and two parking Lanes so what they do they engineered it so that um the traffic lane snakes around and cars park in the other places and so what you get is um you cannot go very fast through there just because you've got cars parked every once in a while and you have to snap through it um that's architecture they order social behavior but through non-deliberative means sometimes um theorists and the like talk about these as if they were scripts or recipes but scripts and recipes work through our deliberation through our judgment um and some of us don't have enough grasp of the world around us to be able to follow those scripts very well and we do for those recipes we do them very badly if my wife were here she was earlier lecture of mine she would tell you I'm very bad at it why because I can't follow a recipe I don't have the appropriate experience over time for that thank goodness she's back home um so what AI does in this mode of social ordering is it goads rather than guides it offers guard rails rather than guidance regimentation rather than regulation now if now if air description of what some people want to call artificial legal intelligence we might say okay that's different so what it's a difference so what what's lost that's what I want to turn to now keep in mind that the objective of the rule of law was to temper the exercise of power to protect and provide recourse against the arbitrary exercise of power through the law the rule of law deal recognizes that there might be other ways of tempering power so you might think this is just an alternative an alternative way of doing the same thing maybe even more efficiency and more efficiently I think it would be unfortunate in one respect anyway that way of thinking of it is unfortunate in one respect because um the CH the rule of law ideals choice of law was not just a matter of what seems ready to hand what seems may be efficient for the time being 100 years 200 years thousand years ago right there's a more there's a closer I might think even principle reason for going for the law given what its aim is and the underlying the value it means to serve so law is chosen for principal reasons or reasons internal to the very kind of aim it has um law provides a mode of governance and a mode of Association we've seen bits of that in the in my opening comments and I want to look at what's lost by looking at it under those two heads looking first at law as a mode of governance and then law is a motive Association so first of all what's lost if AI were to replace law as a mode of governance and the interface between those who exercise power and Ordinary People now if I'm right AI leaves no room for judgment or deliberative reasoning interpersonal engagement or decision making responsibility and there are five I think unfortunate consequences of this first this disengagement from the world AI mimics but doesn't inhabit our world the world in which we as human beings live interact shape our plans and projects and actions and build our relationships there's no room for judgment enhancing World engaged elements like common sense um empathy compassion can't thought that in order to make a move from a concept which he thought of as a rule in its application we needed sort of the common sense you learn at your mother's knee is um AI doesn't provide that so it can't understand the meaning of certain actions of individuals and of the state or officials or their impact on our lives for example the experience of subordination the limit on one's Freedom the challenge to one's dignity doesn't have the capacity to grasp that to understand that to appreciate that so Dimensions through which those who are affected by this mode of social orientation ordering can't understand um or appreciate those interactions with them those impact on them secondly participation AI dominated dominated adjudication for example of civil disputes or AI dominated dominating the activity of criminal process provides no opportunity for Meaningful participation by those who are affected or by the public in those activities adjudication by persons of persons is replaced by Administration by and of things I hesitate to Echo Ingles but it does sound a little bit like it AI has leaves no room furthermore so there's no room for recognizing or respecting dignity in these processes third there's an impact on the law itself law not only is an insta is not only an instrument of governing power it also affords resources forums for contesting power and for correcting law on which power often relies law provides means of self-correction AI lives leaves no room for such contesting or self-correction of course revision is possible but revision puts all the actions of AI Beyond Ordinary People and puts them in the hands of experts fourth accountability of course opacity of AI algorithms and the like puts all of the actions of AI Beyond persons but worse um accountability is of of AI instruments is radically different from the accountability we talk about within the context of the rule of law it's not a matter of assessing actions relative to values or Norms rather um computational accountability is often talked about more accurately as explainability an assurance of the reliability of the functioning of the device so accounting is done rather than accountability counting and calculation rather than offering reasons and justification the problem here is not that it's ineffective but it's the wrong kind of accountability moreover because it's necessarily always in the hands of experts it can't function in a network of mutual accountability that helps us build a political community that is structured by and sunk deep into the ethos of fidelity with respect finally to the mode of social ordering this is this while this regulation is in a sense impersonal um and so may be able to allow us to transcend typical human failings of partiality although it hasn't quite done that yet um it transcends the personal dimension in problematic ways I believe it removes regulation from deliberative domain of those whose behavior is regulated rather than addressing it it removes it also from our awareness and replaces government of persons with the administration of things this I think is not emancipation and it's even more problematic if it escapes our attention and or our awareness finally let me conclude with of just a few brief thoughts about the impact of replacing law in the place of AI in the place of law as it affects our motive Association so law makes possible have argued or at least suggested here um a vital and invaluable form of social interaction and political recognition of a vital form of community it provides a vernacular if you will um that members can use to navigate their social lives and it provides a public means by which members can regard respect and deal with each other as co-members and see themselves as a Unity despite the differences the Deep differences that divide them and it gives resources for them to negotiate and renegotiate the commitments of their common life together all of that's lost if we replace law with AI as mode of social organization social ordering AI regulation carves up the polity into small bits atomized bits um and doesn't help establish or sustain Community among the bids no resources are available for constructing or sustaining personal or communal lives over time its vocabulary is radically private or singular contest so members can't really conceive or articulate or contest or revise their understandings of their common life together finally without law we are deprived members are deprived of a key means of of expressing and acting out Fidelity to each other in a way it puts the value of membership beyond our capability so let me conclude it strikes me that this is all a very high price a high price for efficient and effective regulation even if we Grant and could be assured that a noble end is being served and its distortions are mitigated and its benefits are distributed equally or equitably um whatever aggregate benefits AI technology in the place of law might bring high-valued public goods are lost so let me just say I've said nothing about the role of AI in in the law I've just addressed AI replacing the law there's a whole lot that must be said about the role that we might still find for AI though we have to find ways in which we feel we are control in control and not Ai and so with that okay cool [Applause] so let me so I think thank you so much that's a really wonderful talk and it raises both abstract questions about this ideal you've articulated about the rule of law and how it might not be satisfied if we replace law with automated systems but also there's a further question which is well but to what extent is that ideal actually being realized in the world as we know it so if it's not actually being realized then it looks rather different we want to employ these systems so there are I don't think we've settled on who what the ordering is did you guys have any views brilliant so can I ask that you go to the lectern and we can begin with you Natalie thank you thank you good evening everyone Professor Christina thank you so much for that presentation and um for your book which I read in preparation for this evening and now having read your book and listened to that talk and there is strong potential for my remarks here to be quite short because I find it nearly impossible to disagree with anything that you've said yeah your description of the Law's distinctive mode of reasoning reminds me why I was attracted to the study of law in the first place and your account of the value of law as a mode of governance as a framework that creates the space for compassion for Meaningful participation for the contestation of the exercise of power for accountability for reparation and Community speaks to the reasons why I made the choice eight years ago to work for a foundation that exists to extend access to the legal system to those who need it and are without it now I want to preface what I'm about to say by noting that you were very careful to state that in this discussion we should seek to compare like with like not digital technology at its best with Lord It's saddest or worst or vice versa but truthfully I'm not sure that conducting this discussion at the level of abstract equips us to address the challenge we are facing because the prospect of governance by AI as the replacement or augmentation of large parts of the legal system with AI is not some distant Prospect it is here it is now the project of AI and the place of law has powerful Advocates both within and outside the legal system in England and Wales and globally on the train here I was reading an article by Dr sat Lazar a visiting fellow at The Institute who stated that we have to design systems and ethics for people as we are now not as we should be to put a different way context matters so in terms of access to Justice where are we now both globally and closer to home the picture is to be frank deeply concerning as Professor tassieulas points out in his paper on AI and the rule of law oecd research estimates that less than half of the world's population live under the protection of the rule of law in England and Wales Court backlogs are at record levels legal need is widespread surveys have shown that over the past four years over half of all adults experienced a legal issue individuals with problems relating to immigration enforcement welfare benefits and discrimination life-changing problems were most likely to try and to fail to seek help Recent research from the resolution Foundation found that in 2022 one-third of workers at or around the wage floor roughly 400 000 people were underpaid the minimum wage in violation of Labor Market rates the same research found that low-paid workers were the least likely to enforce their rights through the employment tribunal due to a combination of high fees lack of support delays and strain associated with the process even when they do and are successful the awards are far too rarely enforced the study reported that almost 50 percent of awards made in favor of employees are unpaid Behind These statistics are children stuck in unsafe housing dying from mold exposure parents struggling to feed their families people who are made ill ignored and exhausted by the very systems and processes that are meant to protect them in this context where the legal system that we have too often fails to deliver on its promise and in a policy climate that has at least recently been dominated by a combination of techno-utopianism and demonstrable unwillingness to spend resources on the justice system it is unsurprising that governance by AI with its promise of cheap and Speedy access to decisions looks appealing to those in charge of the system the choice is framed as a decision by AI or the status quo I fear that appeals to the value of law as a mode of governance however valid are the rhetorical equivalent of bringing a knife to a gunfight so how then do we move forward and how might we use the arguments that Professor Pristina has advanced in doing so I have two thoughts so firstly we need to end the culture of paternalism in our approach to instituting reforms to Legal processes and democratize discussions around Justice Reform Arguments for and against AI are often framed as a debate between people who conceive of the law as a tool for Distributing outcomes versus those who believe that laws virtue lies in Fair process in advance in these arguments it's striking how often those on both sides of the debate make appeals to what people want to justify their claims which is interesting when as a discipline we have made such comparatively little effort to ask them often our failure to ask the public what they think and want is a function of deeply ingrained legal exceptionalism simply put we often assume that the justice system broadly and Court processes in particular are too complicated for the public to have legitimate views about our Collective failure to invest in programs to promote public understanding of the law have in fact made the system extremely vulnerable in health and education public concern about the expansion of AI has forced governments to row back from its expansion to counter this last year we at the foundation commissioned the first ever study combining polling and public deliberation to explore the public attitudes to third-party use of data in court records the research explored public attitudes to the ReUse of Court data for purposes including designing ai-based alternatives to litigation for those on low incomes we found that a lower proportion of participants on low incomes who in theory stood to benefit from this development were comfortable with this than their higher income counterparts in the deliberation exercise participants made appeals to a number of the values that Professor pestima has articulated particularly around participation and the desire for a process capable of holding power to account and deep concerns about the ability of AI to deliver this engaging the public asking them what they want and need from the justice system is vital to build the case for governance by law my second thought was prompted by Professor steamer himself in his defense of the value of adjudication acknowledging that courtrooms and legal processes are often not paradigms of respect for the Dignity of individuals who are subjected to them he contends that this fact is not reason enough to supplant them with decisions made by AI he argues that AI Technologies eliminate opportunities for valuable participation by their nature and deep ambition whereas the obstacle standing in the way of realizing Law's ambition are practical in nature we can recognize and value this ambition to Accord parties the dignity they are due and measure actual practice against this ambition and seek to bring practice closer to profession and this is surely the key to resisting governance by IAI it is incumbent on all of us to ensure that the legal system lives up to its promise investing in research to understand where issues occur taking action to root out the problems that we know exist bias racial bias gender bias and ensuring that the public have the tie type of experience when they encounter the courts that makes them want to defend governance by the law perhaps after all the benefit of the threat of governance by AI is that it forces us to look inward to confront our legal system as it is and to redouble our efforts to ensure that governance by law fulfills its promise for everyone thank you very much Sarah Jonica okay um thank you for inviting me and thank you very much um for uh everything that has been said so far my uh five minutes is probably going to be um very apparently an exercise in how to say you're not a legal philosopher without saying you're not a legal philosopher and probably also how to say that you don't know what five minutes is uh without saying you don't know what five minutes is feel free to throw things at me if I overrun um I did have quite a strong idea of what I was going to say this evening before I turned up and I've just just jettisoned all that now that given um what I've heard and what I've been thinking about while I've been listening so I'm a commercial lawyer mostly and by training I suppose is the best way of putting it and now as a law commissioner I have lost the luxury that I had uh when I was a full-time academic of being able to grasp onto ideas and and pursue them because I thought that they were the they would give the ideal outcome I say that because I I think I might be disagreeing on on some issues here although I'm disagreeing in such a subset of law that it we might actually come all the way around and and say that what we are doing actually is agreeing the reason I said the thing about that being a law commissioner and and being faced with a reality that I have to deal with so AI is happening as as Natalie has said we are already faced with it parties are already interacting on the basis of algorithms on a huge scale and so one of the questions that I've been faced with is how should the law respond to that now I think that the subset that I've referred to is commercial law commercial parties and maybe a little bit broader contract law of course the essential element of contract law is agreement it's consent it is a mutual um meeting it's a Meeting of Minds it's a mutual um uh identification of the way in which you want your dispute to be regulated and to be taken forward and so my question um as we were hearing about or uh party integrity and party autonomy and being governed in in quite a passive way of course all those things don't apply in quite the same way when we're talking about is a commercial contract situation if party so we've already seen algorithmic dispute resolution being signed up to on quite a large scale so of course that's parties saying if x happens then we want y to happen as a result if you've both done it if you've both been informed if needs a review is under duress of any kind then why should the law not say okay when that happens we will step back or it will step back and the parties can use the algorithm that they have fastened on to resolve their disputes I guess that's a question and what we're really talking about here it's the it's the other end of the scale really I think there are ways in which artificial intelligence I know that's quite a generic terminal and I'll give a couple of examples can actually democratize law in a way that is not possible without it so what it does allow parties to do in certain situations is to set up their own dispute resolution model and of course as long as they're both happy with it and of course what we're not talking about here is any harm inflicted on anyone that the state would be interested in that is a separate question so I am um you know constraining my remarks to the commercial Arena but it's a very important Arena not least because the more parties that set up their own dispute resolution systems and then adirin are quite happy with the outcomes the more resources the more legal resources there are for those who aren't in that position and who have to resort and we would want them to Resort given what we've heard about discretion and equity and mercy there are many questions for which those are the obvious tools um to use in relation to now one of the things that really interested me when I first started reading about AI systems um was in the context of chess and of course artificial intelligence uh is is very important to chess players increasingly so and in fact Magnus Carlson who for those of you who don't know is one of the best chess players in the world will will leave the recent um controversy aside uh well that was directly him but anyway Magnus Carlson fantastic chess player and he was at least by his own account one of the first players who excelled in the artificial intelligence era so the era in which chess players practice against artificial intelligence and of course artificial intelligence has this amazing sort of to me incomprehensible store of data so how did he do it the way that he did it was he started making deliberately non-optimal moves because that of course confused the AI and that's how he managed to beat it so what does that tell us that tells us if you're a party who wants to I mean I think all of this also makes this question as commercial lawyers we are always told and we always say in fact what we're looking for is certainty what commercial parties want is certainty now ai gives you certainty um of course it changes over time because it's only as good as the data you feed it but it does give you I think more certainty than human discretion it may learn guile over time but it's certainly um doesn't have it in any instant way so what does the Magnus Carlson experience tell us well if parties do want that predictability without the possibility of Mercy or capriciousness or the sort of Equitable response of a human commercial parties don't necessarily want that for obvious reasons it's not usually their livelihood or their welfare at stake it is simply their financial resources and actually what they do want is certainty what they do want to be able to do is say well if I behave in this way then the outcomes will be within a much smaller subset than they would be if we weren't using an artificial intelligence because of course commercial parties just want to be able to price their behavior and to factor that in and so if they can do that then my question is why why should they not be allowed to and then leave that pie that sort of jurisdictional and Court resource pie alone for those who really do need it um how long have I got left okay so I've got one final thought then um and it's and it's again it's it's in support of AI and as I said it might not really be a disagreement I don't think because I certainly do not think that you know robots should be taking over the world and that everything should be left to AI but I do think it has um a lot to bring to um the law party now the other example the concrete example that I wanted to use um was uh one of the the projects that we have at the law commission at the moment is how to reform the law um to accommodate decentralized autonomous organizations or doubts now these are effectively again for anyone who doesn't know apologies if you're completely familiar with it but a decentralized autonomous organization is essentially a company that's run by virtue of smart contract so it's a coded company and of course what you can do with a company like that because it is all based all the decisions all the governance decisions are made on the basis of algorithmically determinable solutions effects then why and then the code Works through and everything happens what of course you can do is you can ensure in the code that governance happens as it should and that the law is adhered to and that the company's acts are you know that they are doing what they are supposed to do because there is no room for human error negligence criminality now of course there might be at the coding level but it's certainly reduces that and code can be checked before it is run I'm pretty sure that was my last minute um so I will uh end there thank you [Applause] thank you so much to John and to the hall Institute for bringing us together thank you Jerry it's a pleasure to be with you and I I thought I should bring you a a Horror Story um and I've got it here this morning and I'm ready The Cutting Edge Jerry this morning at my kitchen table I gave my son who is innocent of any legal training the question should the court in judicial review defer to an administrative authorities interpretation of the law should the court defer to an administrative 30s interpretation of its own policies he has chat GTP and in about four minutes he came up with an answer that I graded at 58. which is a high 2-2 of a second class result um and he thinks that I was biased against it because I knew that it was coming from this dodgy Source um and and the Striking thing about it is that although he's not a lawyer when he got the output he realized that there weren't any sources in it so he put in another prompt saying please can you cite appropriate cases that justify the points made in this essay here's what it said I apologize for the oversight I have now included relevant case law to support the arguments made in the essay now um it was still banal and interesting interesting that the cases it cited important cases were not well used were not it didn't show the ability to do something with a particular judicial decision to make an argument so it's never you're never going to get a 2-1 or a first using shot Jeep TP unless they develop it and they'll do that by the week after next uh so there there there's my Horror Story the implications are are Stark for the Integrity of our students um and for the future of legal education which I'll come on briefly at the end but I did want to say something in answer to Jerry's question what if value would be lost if we replaced the role of legal reasoning in Social ordering by by that by by uh by the capacity of machine learning to come up with outputs that um simulate argumentation Jerry says simulation replaces argumentation AI Technologies replace accountability with accounting um and I I broadly agree if we and yet I think I agree with Sarah Green as well I I agree with Jerry if we think of of a of a general policy of replacing uh rule by law of a community political Community like this with ordering of society by Big Data um and by artificial intelligence solutions to problems of Big Data here's what we would lose we would lose intentionality in in Social ordering which means we would lose personal responsibility for decisions as to matters of private concern we would lose public responsibility responsibility of our community for decision in matters of concern to the community and on that broad agenda if that agenda would pursue that's a catastrophic loss we would lose Pro we would here's another slogan like like simulation replaces our argumentation we would replace process with processing legal process came up comes up in in what Jerry said today in his book um and it's so fundamentally important legal process is a systematic array of techniques enabling intentional Decisions by human beings to have effect in The Ordering of society and we would lose that the the general sort of move that I'm imagining well on an intestacy the test eater's Goods would be distributed by Big Data analysis of their preferences this is something that's actually been recommended in law journals speed bumps we would wake up tomorrow and the speed bumps would be there Jeep chat GPT or another another algorithm could determine where they are to be and and in those cases private and public what would we lose the intelligibility of accountability for the distribution of a person's property the intelligibility of accountability for the public decision about just how bad it's going to be to get home on my street or just how bad it's going to be when everybody else is trying to get home on down my street and for those decisions we ought to have responsibility I agree with Sarah because if you think for a minute about the speed bump situation I think the decision makers I think they ought to be responsible and that the decision to put a speed bump ought to be made by a public agency through intentional decision making for which it's conceivable that they could be accountable and for which they are responsible but it might be so good if those decision makers had Chachi PT or another facility to do the modeling modeling accounting even Jerry says that it's good to have um to it's it that that AI would be useful for accounting so wherever accounting would be useful in public policy making Bring It On modeling potential massive democratization of information and assessment of policy these are these are these are great goods and we shouldn't lose track of them just because it would be a a terrible loss to give over the the governance of a political Community to chat GPT um uh so I just want to close by asking the implications of chat GPT for what US academics do do you know what I'm not going to use it in my research because I find the ingratiating tone just weirds me out when I'm talking to it when I'm talking to it listen to me when I'm reading the output from an algorithm and here I am trying to think what's who designed this and how do they did they program it to apologize or has it done some machine learning magic that makes it produce an output that is the same as if it were apologizing that that freaks me out but there could well be other similar techniques for research that would be brilliant um and and brilliantly useful and and that's a matter for the future what about teaching why what about our students they could get from chat GPT a lot of what they get to us and the week after next it'll be better I'm still ahead of it I think I can outluster chat vpt even now and it's not just out blustering I have an intellectual gift that all of you have that that it demonstrably lacks of focusing on the particular illustration or the particular turning point in the law or the particular ground for decision so I don't know how we'll be far long will be ahead of them in that respect but but here's why I don't think we should replace our teaching by giving students chat GPT there's the potential that I could understand him or her um and and they're not there yet um in in uh in AI thanks [Music] [Applause] I'm tempted to say that a truly ingratiating chat GPT would have gotten the first because it would have cited to Professor Endicott okay so I think what we're going to do is maybe allow Jerry a chance not to sort of um feel the need to respond to everything that's come up but maybe to pick a few highlights and respond to that and then we will throw it open to General discussion from people present and online Jerry so I um I'm oh I am here I I'm I'm pleased to have these comments and um that brings completely to mind the need I have to interact with people closer to the ground than I am at Google Earth I have a wife who is a um a serious practiced lawyer for a long time and she read a good bit of this book and it's so much better because um even at one point where I was citing a case and describing it she said that's wrong how do you know I said she said because I was the lawyer of record in that case and it didn't go like that at all so the bringing things together is really important and I value these kind of comments a lot um what I hear for the most part are suggestions about how there are there are things that um algorithms and AI can do and maybe do well and do better and I'm I'm perfectly willing to accept that as a possibility I think we would have to argue um particular kinds of cases like for example the ones you've mentioned to see whether they they actually um perform on their promise um I should say Timothy that um I gave up teaching I'm I know Emeritus now for four years in part because I got the feeling that I didn't know where my students were anymore and that was four years ago right who knows I mean they didn't have chat Bots available to them then but there are two things that I wanted to mention here that I think are of uh a serious concern and it came up in part with Natalie's comments um I really like the idea of getting over this paternalism business um and and Sarah's um and that's the thought maybe the hope that at least for important parts of the law especially work Computing or um getting reports and doing that kind of research can be really valuable I agree that there are such um one of the things that might be made more available to us is just access to Justice um we are very bad at that I can talk about my own legal system and not yours but we are very bad at that um and um we if we could find some way of being I mean having the law actually more accept accessible to people who are not already wealthy and educated um so much the better um I'm I really believe that so but here's my worry um I remember now is it maybe 20 years ago when we all thought that the internet was going to be the great democratizing force and what's happened just the opposite and My worry is I mean maybe we learn um that's what we're not in the mode anymore at at um at the institutes that develop these things um in the mode that of of um Rush ahead and break things were much more careful perhaps but we need to be more careful and the problem is that um the technology is a lot further along than our understanding of how these things work so it it gets away from us and I worry not in particular cases but in because of our terrible experience with the internet's failing on its promise to be democratizing that this will fail too a good friend of mine Paul Gauder nailed it now at um Northwestern has already argued that the promise of the the sort of egalitarian promise of of AI is likely to work just against it because what it'll do is it made these really efficient processes even more available to wealth and to those who can get the wealth um the the the the algorithm manipulators do their work and if that's so um democratization um will fail that worries me um maybe I should just stop there because we have more questions I I really wanted to talk about private dispute wrestling let me just say one more thing about private dispute resolution um yeah and and but this is this is not really a new thing um there have been arrangements for um commercial relationships that try to get around the law because law tends to be far too costly and so in mediation or arbitration mandatory arbitration like we've had those devices they're not AI driven but they're those devices for a long time and the one concern that sometimes comes up is that what it does is that not only it might serve the interests of the two parties but there is a there's a public good involved in there being a legal process available to those two parties and that public good of um managing the law and manipulating it and correcting it and like we'll be lost now if I'm afraid that that might be the case also there but we need to talk about the details thank you just to pick up on that little bit this issue about democracy I mean there's already as part of the um breakup of our Democratic ethos a way in which the rich have departed from the rest of us and live in a different world one of the ways they've done that is they no longer care about our broken legal system because they resort to arbitration and some of our great judges and legal academics get paid huge sums of money in order to operate this parallel system so as it were that's already a kind of attrition that's happened to our Democratic culture and potentiality and that takes Sarah's point that it is attractive to think you could divert some resources from commercial disputes and then have more resources but it also risks further entrenching the idea basically the Richer no longer our fellow citizens but they're people operating at a different level kind of shadowy level doing their own thing a non-legal Black Market yeah to that as well though when um when I was referring to commercial parties in in the most part I completely agree but there are some examples I think where one side is commercial but the the the dispute the cheap dispute resolution does actually favor the the small person as it were so um with parametric insurance contracts so the obvious one that was used was if you um your flight is canceled you know you you as whoever you are you might not have any resources at all but all you have to do is get the algorithm to confirm that your flight was in fact canceled and then the money that you are then due is automatically assigned to you so in the main absolutely but it isn't only the big commercial parties that can this is one of the interesting things about Jerry's paper that in a way Jerry you made things a bit easier for yourself but in a sense talking about wholesale replacement of light right but someone might say well look um there's an element that's just pure drudge work there's an element of minor forms of issues where your procedural value is important though they are could be legitimately trumped by other sorts of concerns yeah so I think in a way that's the kind of practical end of this problem when that's appropriate right one one thing that I I take that point exactly um completely um the rule of law is one important value amongst others and it has limitations the limitations are sometimes on the value itself and sometimes on what law can do and if we could find other ways for law doing that kind of thing then it may not be inconsistent with the rule of law before we open up so I think there's another type of threat here so um which is partly driven by the economics of how these things work and these tools are primarily developed by private companies and in order the real attraction is in those kind of low low value high volume areas of the justice system that's where which is where the majority of people interact with our legal system and if we are to say that actually for reasons of Economics we're going to lift all of those people all of that volume of disputes out of the system it creates this where are people's interactions with the law going to come from part of what is trying to get across in my commentary is that you know one of the reasons why the law is so vulnerable to take over by these particular Technologies is because there just isn't the good experience that people have with the justice system to advocate for it and I worry exactly as John's expressed in shorter and more eloquent terms that we end up creating kind of two-tier just system and undermine our democratic values in the rule of law further because people don't realize the Practical value of it unless they're interacting with it now I really do want to go into the questions from the audience so who wants to kick us off yes there's a question here at the front and why can you just wait for them thank you Christina said that architecture is um the rule of social behavior the guidance of social behavior in non-deliberative ways or much better than that but we're never free from architecture and I think when we're discussing these questions artificial intelligence the way it affects our lives the way it affects our culture and our legal system depends on the architectures in which it's implemented so I've noticed this discussion is largely about do we use artificial intelligence or not and I think it should instead be about how do we use artificial intelligence how do we architecture our use of artificial intelligence so that it supports what we value and the objectives we want to achieve I think that an approach of let's not use artificial intelligence is likely to succeed but I think one in which we carefully architecture it could very well so to collect a few more questions I mean assuming they exist right at the back my name is Colorado I'm one of the D field students here so my question is about something you said Professor pestima that one of the problems with using AI as a legal decision maker is that AI doesn't have the tools for its own self-correction but I wonder if the case might be even worse because we can involve AI as a decision maker and it increasingly adds into our body of Precedence And if every precedent changes the rule by applying it in a new context then over time are we not inherently and inescapably assigning control to AI over actually what the law is and will become there's one more spot if anyone wants to take it Linda thank you very much this is a quick one um I was curious to hear a little bit more about democracy so you spoke about the rule of law but it seemed like Democratic Values were sort of like the elephant in the room I was wondering if you could that make that a little more explicit what the relationship is there okay um let me reply in that order um yes indeed you know I mentioned this um arrangement we have in in my hometown where things are that's a pretty good way of doing it that's architecture of in in an effective way keeps people from going too fast through through um residential streets I like the idea and the question is not architecture or Norms it's where we deploy them how we deploy them how responsibly we are in deploying them and when it's architecture um are there points where we must introduce some kind of accountability and what form would that take absolutely um and so John was quite right to say I've made the case easy for my vice Myself by putting it in terms of yes that's right yes and that's right because there was just one kind of problem I wanted to address here um that opens the door for just these kinds of questions um and each of my commentators have suggested various parts where one could really put much more in attention into it and I confess that it's they who will need to do this because I don't know enough about it to be able to say anything intelligent I can just say some things about what values are at stake what concerns are at stake and then we have to do the really hard work of design design is what we need to do but we have to be motivated by and see the background of the values um but tools of self-correction um I'm I'm puzzled about about the self-correction idea because in some sense you know um at least as I understand the way the algorithms now work so that that gets Timothy's um paper on on a legal issue is that it's it's working with all of this stuff and it does some kind of statistical analysis on it and then it issues an outcome now suppose that gets plugged into the data that's a new one and then you get plugged that into the data um and it might be self-correction it might be self-corruption right and we need to have some kinds of controls on that kind of thing um when I talked about Correctional revision I really meant well it looks like these outputs are not doing what they're supposed to be doing they're not meeting the design and we've got to go back and change the design I mean the the thermostat is not actually measuring the the temperature in the room and shutting down the furnace when it has to um but this is a little bit different right it's sort of internal to it so the potential for self-corruption it seems to me is pretty great on Democracy last I um I just did a a talk in a couple of different places on the rule of law in its relationship to democracy and it's a long story but my my particular view is this very quickly um rule of law is one thing democracy one kind of important value democracy is another they are distinct values it is conceivable that we could have a community governed by a relatively robust conceivable robust rule of law which doesn't meet best standards of democracy however um they are so intertwined so interdependent that it's really that conceivability is not a practical possibility they are um as I want to put it at one point the um the Democracy can work only if there's rule in law rule of law in place to protect all those features of democracy think of election structures and the like um that have to be maintained in order for democracy to thrive at the same time a democracy that um um the rule of law that doesn't issue into something like democracy is like a promise that's unfulfilled there's an ambition behind both of them and it's an unfulfilled promise that we're still working on to the commentators for any of their concluding thoughts but I just wanted to ask you one question to pick up this idea that the rule of law is one value amongst others so in your account of the rule of law you said it's a corrective to arbitrary exercises of power so I'm wondering are there any bad exercises of power that are not captured by that or do all bad exercises of power count as arbitrary well it's um as arbitrary there could be bad exercises of power that are not arbitrary in the sense I have sort of two two ways of there the number of ways of understanding arbitrary the way I have it in mind is that um an exercise of power is arbitrary if it is um not Guided by um intelligible norms and there is no mechanism for holding the power wielder accountable to them um now um other people will Define I think even Timothy will Define arbitrariness in terms of it's not tracking reasons that apply to one or one's into Philip Pettit would talk about is not tracking um one's interests or the interests of those to whom the law is addressed those or the power is addressed there are various ways of making it out that's the way I understand it there can however then be arbitrary non-arbitrary exercises of power that are bad that are unjust that violate rights right so so that's what I was trying to get at so it sounds like I actually agree with this but I thought you weren't going to embrace this that you could have compliance with the rule of law with quite a lot of Injustice and quite a lot of rights violations happening absolutely and so um the the lecture I was telling you about actually was rule of law human rights and democracy and the argument was um they're distinct it's possible for there to be rule of law and violations of human rights and violations of and and undermining of democracy and back and forth the problem is that the um these quite different but very important um concerns of our political morality are really intertwined so for example some of the most important human rights also are the kind of things that the rule of law also supports on different grounds right um and um to put it a little bit of a different kind of case there are human rights which if they were systematically violated the it would be hard to be consistently committed to the underlying principles of the rule of law and think that that's an appropriate way of dealing with human beings there's a kind of practical contradiction in that not a not a theoretical one not a logical one but a practical contribution how could you be so committed to the principles of the rule of law and yet see that the Wayne in which you're proposing to treat people would be appropriate that's the kind of relationship makes a lot of sense now I'm gonna steal a few more minutes to give commentators a chance to say so Timothy do you want to start us off or something from chat GPT whichever yeah um one thing that I didn't get to in in my talk that I I think is something to learn from the issues that we've been considering it concerns the the purpose of of social ordering in general the I was thinking thinking about process having read what Jerry said the importance of process I was remembering what people call the legal process school that several people in this audience will be familiar with Henry Hart and Albert sacks taught a terrific course at Harvard in the 1950s tremendously influential in in the history of ideas in America and in particular in in legal in legal academics and and legal thinking in general and they started from the principle that the purpose of governance should be to maximize valid preferences and then they did some really good lawyering on how to get there through processes that allocate power in intelligent ways and they were really sensitive about this that get us as close as we can as a community with our different opinions to the maximization of valid preferences and I think some of what Natalie was doing in her talk today would fit right into the legal process School thinking intelligently about how the interests at stake can be represented through intentional action of participants in a in a process in the best legal sense and in process generally they were um lawyers with without a sort of narrow Vision they were thinking not just about adjudication but about Administration and legislation and Constitution making and negotiation and private ordering as they called it now imagine in the 50s they had had AI as we have it today or will have it the week after next um would they have said oh valid preferences can be maximized without all that stuff we don't need process we can we've got processing well I I hope and I trust because I I think they had good instincts that they would not have said that but then they would have said had to say wait a minute the maximization of valid preferences by the way the valid the word valid I love the word valid in in that Manifesto because they didn't say what was valid so they didn't have a theory of of what's right and wrong but they I think they would have had to face up to the fact if they thought about what we've been talking about that the maximization of preferences is not actually the point of the legal system the point of a legal system is to order a community in a real governed way that enables the right actors to intentionally make decisions that will order their private relations and their and the organization of their community in the interests of justice and and and the good of the community so I think this is actually quite significant for the purpose of government it's not to maximize preferences or or we would be better off with the machine it on the assumption that the machine is not just really clever itself at learning but also not biased unless we've had some deeply ingrained preference to make our own decisions I guess but that isn't the reason though that you would say that's right Sarah one wonders whether Bentham would have been happy to welcome um chat GPT well you would know just anyone and and I suspect he would Sarah yeah on efficiency grounds if nothing else um yeah I think the thing I want to round off with is something else I didn't get uh to touch on um in my first five minutes um it is really to to make or at least I mean I did I didn't do this in any sort of crystallized way but make the point that my objection um to the objection about AI only goes so far and really what I was saying about algorithmic dispute resolution is of course all well and good if what you're talking about what we're talking about is um I guess they're sort of horizontal application of law so you know we know what the variables are going to be and we know what the rights are in relation to those variables and so do the parties which is why did they sign up of course because they they in the sense know what they're going to get um but of course the real Beauty particularly of the common law is its ability to develop in a in a vertical way and not just a horizontal way and then there's a strange sort of um existential question for me as a law reformer here that um AI would be really bad at Laura form um obviously it couldn't tell us what to do about itself um because what it what it does do it's it at least in its current um form what it does do very well is I was going to say reason then I'm not going to use that verb it computes by analogy um what it cannot do of course is is cope with any Paradigm changes and that is what the common law occasionally needs to do um and so I agree in an enormous way with what Professor festimo was saying on that point that certainly there are I mean really what I'm saying I suppose is um horses for courses and I think that there are some ways in which artificial intelligence can really help us with legal resolution but it is by no means the answer to everything and most of what I deal with on a day-to-day basis would would not be able and what most um common lawyers deal with on a day-to-day basis would not be resolved or resolvable by an artificial intelligence the interesting stuff really you know as John said it's really um the the grunt work of lawyering can I think be done by artificial intelligence so that's those vertical sorry those horizontal applications and and drafting of you know very standard contracts and things that at the moment parties still pay quite a lot of money for when there's no need to and Natalie thank you so much and so I mean I just want to say this has been such a really interesting exercise and I've been struck by how difficult I've found it to remove myself from the contextual issues because um I think you know my concern and my response was driven by the the sort of value proposition of replacing law and decision making with AI but also the knowledge that the areas of the justice system where these on these Technologies are most likely to be deployed and the under investment in the infrastructure that is needed to make those Technologies fair in that context has driven a lot of the concern but I nevertheless I really commend this book and I really enjoyed having the opportunity to you know cease that out so clearly on articulation of what we should be seeking to defend when we're talking about the value of the justice system so all of which is to say I've really enjoyed the discussion and thank you so much Professor for the book thank you I'm going to seize that positive so thank you very much to Jerry for coming here and giving us this wonderful talk and also for this spectacular panel actually and the really great discussion that we've had in consequence and thank you to everyone for joining us [Applause] thank you guys