Rise of AI Summit 2020 · Berlin · Society & Strategy Day | Hybrid Area

AI Strategy Beyond the AI Race: Winners, Losers and the Countervailing Move

Jesse Lehrke · Researcher at Freie Universität Berlin

At the Rise of AI Summit 2020, Dr. Jesse Lehrke, a researcher at Freie Universität Berlin who describes spending recent funding breaks learning Python and neural networks, argues that most current AI strategies are really just strategies for winning the AI race, not for what to do once AI, or a technological singularity, actually arrives. He frames strategy as the moves and countervailing moves opposing wills use to get more out of a situation than the underlying balance of power suggests, particularly useful for weaker parties, and argues academics should not fear "futurology" given the stakes. Because the singularity is by definition unknowable, Lehrke proposes thinking about the limits physics imposes on any actor, human or AI, including the speed of light and quantum-level constraints on secrecy, as tools that can be used to shape an AI's behavior through embedded uncertainty rather than eliminated uncertainty. Drawing on Clausewitz's trinity of passion, reason and chance, and on nuclear deterrence as a model of control through non-use, he argues design decisions made today, not decisions made after the AI race is won or lost, determine how controllable future AI actors will be.

Key takeaways

  • Lehrke argues most published AI strategies are really only strategies for winning the AI race, not for what to do once it ends.
  • He argues weaker parties can gain leverage by manipulating a stronger actor's, including an AI's, physical limits rather than trying to close a capability gap.
  • He reframes Clausewitz's trinity of passion, reason and chance as decision, information and uncertainty for the AI era.
  • He argues academics' fear of being called "futurologists" is itself a risk, since a topic's importance, not its ease of study, should determine whether it is studied.
  • In Q&A he argues supervised learning on labeled data is a dead end compared with semi-supervised, unsupervised and reinforcement learning.

Detailed notes

Strategy, the singularity and why most AI strategies fail

Lehrke defines strategy as the moves and countervailing moves opposing wills use to extract more from a situation than the initial balance of power suggests, a logic especially useful to weaker parties. He frames the singularity, the point where technological capability exceeds human comprehension, as his second focal point, arguing it is more useful to discuss than incremental AI steps because it is an extreme, unqualified case, even though by definition no data exists about what lies beyond it. He argues that most published AI strategies, despite their titles, are only strategies for winning a zero-sum "AI race" up to that point, not strategies for how to actually use or control AI once the race is over, and that current security planning suffers from a "lack of imagination."

Why planning for the unknowable is still necessary

Anticipating the objection that planning for an unknowable singularity asks too much, Lehrke argues it is both possible and already happening: he cites Henry Kissinger writing on AI despite admitting he knows little about it, and historical military strategists who envisioned air power and naval power as dominant tools before the technology existed. He says academics' fear of being dismissed as "futurologists" is itself a danger, because the value of studying a topic lies in its importance, not its ease. He argues the singularity's dual religious-and-scientific character requires an interdisciplinary approach, and that even though pursuing AI is partly a belief-driven, self-fulfilling process, some form of advanced AI or singularity technology is effectively unavoidable, since any country's restraint only lets others advance along a different path toward it.

How the nature of conflict changes: interaction, violence and latent actors

Using Clausewitz's concept of the nature of war as interactive, violent and political, Lehrke argues all three elements will change as AI advances. On interaction, he says today's framing of an "AI race" between known human rivals will give way to facing AI-enabled adversaries whose capabilities are unknown, and eventually to AI itself as a "latent actor" that political science's usual models, built for human rivals, are not equipped to handle. On violence, he argues speed will matter not just as faster hardware but as faster thought, production, via real-time additive manufacturing updates, and faster outcomes, potentially producing conflicts resolved without visible fighting at all, which complicates the traditional strategic goal of defining "victory" and "political purpose."

Managing uncertainty through the limits of physics

Lehrke reframes Clausewitz's trinity of passion, reason and chance as, respectively, decision, information and uncertainty, and argues that in the singularity era, strategy should not always aim to reduce uncertainty, since much of it cannot be reduced, but instead to manage or redirect it, including by increasing an adversary's uncertainty in ways that expand rather than threaten their options. He proposes an axiom that any agent with limits behaves differently because of those limits, and that understanding an actor's limits, including physical limits such as the speed of light and quantum-level constraints on what can be observed or kept secret, lets a weaker party shape a stronger or AI actor's behavior. He illustrates this with a thought exercise: an AI with unlimited time but a processing speed capped at the speed of light faces so many possible futures to evaluate that maximizing its choices becomes a form of deterrence, an argument he ties to design decisions available today, such as deliberately embedding uncertainty into an AI system using existing tools like Bayesian statistics and chaos theory.

Q&A highlights

Asked for practical advice for startups and investors, Lehrke says giving AI systems a genuine conception of time and space, building on existing mechanisms like LSTMs and attention, is an important next step, and gives high-frequency trading's physical placement between exchanges as an example of space and time already mattering commercially. Asked whether investing in high-quality labeled data provides durable advantage, he says he believes, citing Google's own direction, that supervised learning on labeled data is a dead end compared with semi-supervised, unsupervised and reinforcement learning, and warns that Europe's strengths in physics and robotics research could be undermined if its AI initiatives are bought out by companies like Facebook or Huawei before their longer-term payoffs materialize.

Quotes

Strategy is actually most useful when you are weak. It's needed to counter weakness and again get more out of the situation than the balance of power suggests.

Jesse Lehrke

Everything has limits, and an agent with limits thinks and acts in a different way. They take those limits into account in their behavior.

Jesse Lehrke

Questions answered in this talk

Why does Jesse Lehrke think most current AI strategies are not real AI strategies?

According to Jesse Lehrke at the Rise of AI Summit 2020, most published AI strategies only aim at winning a zero-sum AI race and end the day that race ends, rather than planning for how AI will actually be used and controlled afterward.

How can a weaker actor gain leverage over a more powerful AI, according to Jesse Lehrke?

According to Jesse Lehrke at the Rise of AI Summit 2020, by understanding and manipulating the physical limits any actor faces, such as the speed of light or quantum-level constraints on information, since an agent's behavior can be shaped through those limits even without matching its overall capability.

Is labeled training data a lasting competitive advantage in AI, according to Jesse Lehrke?

According to Jesse Lehrke at the Rise of AI Summit 2020, no: he argues supervised learning on labeled data is a dead-end path, and that semi-supervised, unsupervised and reinforcement learning represent the more durable direction.

Mentioned

singularity, countervailing move, game theory, deterrence, Clausewitz's trinity, latent actor, reinforcement learning, unsupervised learning, labeled data

Keep exploring

More from Rise of AI Summit 2020

More on AGI & Future of Intelligence

Transcript

Auto-generated transcript, not edited. · 5,038 words

Show the full transcript

and that's around strategy and i have a very very special guest here we already spoke a lot it's highly entertaining to speak to him i'm very happy that he's here in person coming the long way from berlin more beat i actually haven't asked him we spoke so much we i haven't asked him how to pronounce his name but i think it's Jesse Lehrke a researcher at the frey university berlin with a phd but he will explain you more and he will talk with us about strategy and especially the countervailing move which i have no clue what it is so i'm very glad to have you here thank you let me position myself well here all right uh thank you rise of ai thank you fabian and veronica for organizing this event and all the others who assisted uh thank you also to everybody out there who has chosen to spend their time and their quarantine with us today my name is jessie paulierka i am most recently a researcher at the frey universitat berlin however

i've taken the little coveted break in our funding to immerse myself in the technical sides of artificial intelligence coding learning python learning neural networks which as a social scientist was something new for me and i bring up this little anecdote because it's something i'm going to want to return to at the end of my talk today i'll be talking to you today about ai strategy beyond the ai race winners losers and especially the countervailing move the move you make in order to offset if somebody else wins the ai race so to structure my talk i want to give you kind of two focal points rather than a road map about what i'm going to talk about this will allow me some flexibility as well when i'm discussing my uh my points focal point number one is strategy for those of you unfamiliar with strategy strategy is the move and countervailing moves by opposing wills as they seek to get more out of a situation than the initial balance of power would suggest this is particularly

useful for the weaker party who seeks to use the paradoxical logic in order to leverage things like time and space to gain an advantage much like the map in the background illustrates when canada was trying to gain a strategic advantage against the united states in the 1920s to conduct a preemptive invasion if necessary read about it if you like it's quite interesting uh to see how a small power thinks they can possibly win focal point number two is the singularity most of you don't need a definition here of what the singularity is i've put up my own kind of a collection of thoughts on it here for you it is the exponential explosion of technology when capabilities exceed our abilities to comprehend them that we cannot see beyond this point in time this singular event because the technology becomes so advanced we cannot imagine it it renders all of our current approaches to strategizing almost useless because these are based on capabilities we have to understand the capabilities of our adversaries in order to compete

against them why do i talk about the singularity rather than all of the miscellaneous ai steps that we have on the route to the singularity well one as an academic it's much easier to illustrate an extreme case i don't have to qualify everything i say with uh with some other fact and two it's just more fun to talk about the singularity than it is to talk about alexa so if we're going to look at the singularity though we have a problem the problem is that it is an event that we cannot see beyond we don't have any data we don't have any information but we know it's probably coming not necessarily exactly the date so 100 years 50 years is very reasonable this is what i like to call the dark force deterrence theory dilemma draws on the works of season lou a chinese science fiction author in his novels a protagonist must develop a theory neither based on nor testable by data to prepare for something that is distant but certain that something that

is coming uh in a book that was all about aliens and alternate dimensions i found as a scientist this the most fantastical unbelievable part of the books that we as scientists with no data could develop a theory to actually resolve a problem uh that we were facing seemed unrealistic nonetheless this is the problem the task that lies before us today because of this dilemma though in most security planning today we see a lack of imagination as the 911 commission put it actually we see strategies that aren't so much strategies for an ai era even if they have that title they are strategies for the run-up for before the ai era there are strategies to win the ai race not to actually use ai in a manner that obtains an objective so these strategies tend to only focus on winning because the ai race is seen as a zero-sum game you either win or you lose and you might as well just you know hang it up and become a farmer uh nonetheless though strategy is

actually that's most useful when you are weak it's needed to counter weakness and again get more out of the situation than the balance of power suggests maybe i'm asking too much though maybe this is a bridge too far uh the singularity is by definition unknowable asking a planner today to sit down and say hey imagine using ai asking dr franca in 50 years what are we going to use ai for would be like asking a general from the 1890s an era of colonial and indian wars to both imagine and then put into operation the tools of world war ii so maybe we're asking too much i say no though for two reasons first some people are already doing this there are plenty of policy experts there are respected in their own fields who have begun offering their own insights and their opinions on what we should do about ai since he's a big man and he can take it i'll point out henry kissinger in this regard says that he knows nothing about ai nonetheless

he's written several articles on it labeled transhumanism is the most dangerous ideology that mankind is facing he and others like him are making policy even as the ai experts or security experts are sitting aside going no we can't do that we don't have the data but it's done been done before within the military field there are several scholars in the air force an italian scholar and in the naval realm american scholar who envisioned air power and naval power long before they became the dominant strategic tools of major countries so we can do it and indeed we must however we have a fear of futurology especially as academics we're terrified of being called a futurologist if you want to see a futurologist there's one speaking uh concurrently to me on the other stage very good follow him on twitter i recommend it because we shouldn't fear futurology because this fear is a danger to our own future because the merit of study for any sort of thing we're going to look at isn't whether it's really

easy to study it's how important it is so how do we study ai well this is a throw out to my academic friends you always have to have the slide that mentions the janus face nature of something so uh the genus nature of the singularity is that it's at the intersection of religious belief and scientific knowledge so this requires a very interdisciplinary perspective to understand we need we're going to need a lot of different people to get together on this issue but before i dive into an approach first we need to step back and cover one qualification if this is partially based on belief if we are the ones driving artificial intelligence of a bunch of nerds believing in the singularity make it a self-fulfilling prophecy then why don't we just avoid it why don't we just skip the whole process other scholars have dealt in depth with this on the multiple paths there are two various forms of artificial intelligence so there certainly is no fate except the one we make uh nonetheless artificial

intelligence in some form or another or singularity technologies in some form or another do lie at the end of that path uh and this path shouldn't be seen negatively this path opens up opportunities for different technologies and for different countries to leverage their comparative advantages in different areas i have my own views on the european ai landscape that's been off discussed today and i actually do see a little bit of hope if the european tech sector can survive the next 10 10 20 years we can go into that in the question so perhaps so in other words no you can't stop it you can't stop the signal as people who even preceded ray kurzweil making the singularity famous have said the singularity will come because every move a country makes will simply encourage somebody else to follow another path to it so how do we tackle the singularity we need another way of thinking and i'm going to rather ironically take nuclear weapons as a sign of hope here nuclear weapons signaled for many strategic

thinkers the death of traditional strategy you couldn't ever use a nuclear weapon to obtain a strategic objective rather it became an exercise in utilizing uncertainties in a creative manner in order to develop a doctrine a belief about how we can ensure these weapons are never actually utilized it was control leading to inaction that ensured our survival even while conflict in general continue to exist so i take the non-use of nuclear weapons as a sign of hope countries that were previously 10 years ago non-use being very qualified with the two uses there were uh countries that were previously leveling sidilis and fighting a long war didn't use these weapons for decades now i'm going to segue into stuff that uh my uh my uh predecessor dr franco would uh find some familiar i'm going to try to take this a little bit from another way though i'm going to start from the broad and work down to how maybe you also as business leaders or also security professionals could take a security perspective and derive operable

actions you can take today from a larger doctrine or belief to guide your thought very briefly i'm going to talk about some of the ideas of close fitz because this is the way that we can just ground our thinking in something people understand and because we're in germany so just to cover this klaus wits had two main concepts one on the nature of war one on the trinity the nature of war was interactive violent and political and the trinity was passion reason and chance talking about the first i'm going to talk about how the nature of conflict and competition will change my goal here is to encourage you to be more imaginative the first point is on interaction do you know your enemy who are we competing against this is going to change as we progress further and further towards an advanced artificial intelligence today the one we see a lot is the ai race perhaps it's because we understand this the most well we can utilize game theory we can talk about competition and

collective action problems mechanisms of diffusion and we can model this and understand it whether this is useful or not whether it contributes to the problem another question but this is only the first interaction we're going to see there are interactions at many other levels once somebody wins that race you've got to deal with an ai enabled adversary you can't use capabilities to understand how this adversary is going to be to act or what they can possibly do because we don't know those capabilities rather you're going to have to develop mechanisms to understand how on a psychological level level or through a game theoretic level an actor's behavior changes when their information changes and when their time horizons change nonetheless these will still be humans and we can still relatively understand how they will be thinking through an anthromorphic lens the same perceptions of time death space and so on however there is a latent actor that we should probably already start thinking about political scientists talk about this in their own game theoretic models how

they are latent actors and our failure to account for them explains a lot of sub-optimal outcomes we see today those actors are unborn generations that we currently can't model in our political dynamics and our political environments we also have to now start thinking about the latent actor that is the artificial intelligence this complicates any sort of game theoretic efforts but it gets more complicated because it might turn out you actually have more in common with alexa than you do have in common with a farmer from wisconsin which i make fun of only because i am originally a farmer from wisconsin so as illustrated in the books by daguerres on the artelec wars the real artificial intelligence conflict will probably be right down the middle of society bouncing around a little bit for the cameraman there i should uh stop doing that so much so this area is not only less considered than some of the other interstate elements it's also got a lot less institutional mechanisms to control it so it's probably one of the

more dangerous elements what about violence how will this change dr franco mentioned that it's getting faster you know that when we look at it we see hyper wars as realio called them speed is important but it's not just faster airplanes hypersonic drones and such it's also speed of thought speed of production and speed of outcomes as everybody knows as the hardware improves the speed of thought the speed of processing power is going to get much faster we have to account for how this impacts time horizons though everybody says the war will be fast but when you have a long time horizon actually you could attrib forever and the strategy might actually be attrition why would you fight a quick for a quick outcome when you can just outlast your opponent speed of production will also be something that changes quite a bit as additive manufacturing 3d printing means that as often as windows updates on your computer today you'll be updating your hardware in real time as well possibly leading to the death of logistics

that dr frank again talked about outcomes will also become very fast it may be that the perfect strategy as sun tzu points out is not actually to have any serious fighting at all if the ai's become so fast the drones fighting out in the upper atmosphere then you may see no violence at all you may just wake up tomorrow in a different country oh we lost the war last night didn't you hear well you know i didn't even know what happened uh outcomes will be different so how are we going to even understand the outcomes because that is the actual point of strategy is to reach a certain outcome the political end as scholars in political science and strategic thought like to call it victory this is an important an element for scientific analysis it's important for strategic analysis as well it's important for artificial intelligence development the target but how do we know what this should be this is probably the part where i'm the most fuzzy myself traditionally they always said it's a

political purpose but they only got away with this by defining politics is basically everything and maybe we see the implications of that today in our society where everything is political instead we have to start thinking in another way maybe that outcomes are more existential i'm not going to walk you through this thought exercise because i want to save some time for questions and i'm going to walk you through a different thought exercise right after this anyhow so you can read through my slides when undoubtedly rise of ai provides them instead let's close out with talking about the trinity klaus smith's idea of the trinity was passion reason and chance these were embodied in certain political actors but that's a another issue in my view these ideas in the age of singularity security are best encapsulated in the concept of uncertainty chance of course equates very directly to uncertainty reason equating to the information that you utilize to reduce your uncertainty and passion is about that decision you have to make between those two factors luckily

for dealing with uncertainty this is what strategic theory is all about to take uncertainty and apply it to make yourself more secure traditionally this always meant that we must reduce uncertainty we must be more sure however when we're dealing with singularity success technologies my proposal is that we actually may not always seek to do that because we can't we won't be able to reduce uncertainties instead we have to look at creative ways of managing them of paradoxical ways to if not reduce our uncertainty increase somebody else's uncertainty not in a way that makes them uncomfortable but in a way that possibly just gives them more options but options then increase our probability of being able to work with those actors so let me start this by proposing a starting axiom principle uh this is uh the lines of my own research and there are many other ways we could take it but uh points for thought everything has limits and an agent with limits thinks and acts in a different way they take those limits

into account in their behavior these ways then if an actor has limits can be manipulated if you understand what those limits are and areas of pronounced limits can even be magnified against an area where an actor has no limits perhaps the best illustration of this in the scientific literature is the paradox of immortality that an actor who's immortal who lives forever has much higher costs in the short term because everything reverberates through through centuries through eons their own limitations and i cannot do this i can't think fast enough because i is is reflected through the eons so i propose that we need to actually look further at the limits imposed by physics this my own research looks at ideas like how the speed of light affects time horizons hello theories like the principle of least action will inform the thinking of any actor yes these limits are astronomical the speed of light is terribly fast there are a lot of atoms in the universe yet they are limited and there are other limits limits imposed

by quantum technology uh if quantum encryption ever comes along limits that will be imposed on what can be known and what can be kept secret limits at the atomic level of what can be seen and what cannot be seen current stealth technology is nothing compared to the next generation stealth which will will utilize quantum principles in a way that makes it unbreakable once we understand what these limits are we can then utilize them to push an actor's behavior in our favor whether the actor be a ai enabled adversary of the human nature or whether that actor be an independent ai to take you through a little thought exercise on this let me illustrate my little story for those of you who are really into star trek canon you'll forgive me if i get any details wrong if we look at star trek iv this was set sometime in the future and a ship came to the earth and was sending a signal that was disrupting the entire atmosphere it was looking pretty bad for the

earth they figured out that this signal was trying to communicate with humpback whales however humpback whales had gone extinct 200 years prior so to resolve the dilemma the enterprise had to go back in time and and get some whales and bring them back to the future time travel is a high barrier though may not even be possible nonetheless we can utilize the concept within this that if an artificial intelligence has a long enough time horizon and has enough options within that time horizon that its choices are maximized it still will not have enough information to envision every possible future because it has lots of choices because lots of times there are lots of possible futures that it has to run a scenario analysis of it can only think at the speed of light it's unlimited time horizon is therefore magnified by the fact that it can only process at the speed of light and by maximizing its choices it just has too many things to analyze in this way this principle that we have that

there are limits this idea a doctrine that we can utilize uncertainty we can actually come down to a strategy well deterrence we can deter them by giving them lots of choices but the reason this becomes important today is you can actually derive design principles from it but the long-term vision the strategy that we have to control the singularity has actually designed decisions that need to be made today for instance many people argue we need to program death by default this is the design decision that led to all the problems in blade runner films based on my logic however the longer time horizon the more we will be able to push the decisions of an artificial intelligence actor within directions favorable to our own survival that by maximizing choice by maximizing options by embedding the principle of uncertainty within an artificial intelligence which in my basic artificial intelligence technical knowledge isn't that difficult because there's most of the principles that we would be designing an ai based upon are heavily rooted in ideas of uncertainty bayesian

statistics chaos theory there are mathematical ways of modeling uncertainty and by doing this we can actually create an ai we could live with but how do i know i'm right here how do i know that i am the good man that i have come up with a solution well i can't really this brings me back to my original point then in order to come up with the paradoxical solutions required in the era of ai and a strategy to deal with and utilize those ais we need an interdisciplinary and creative team we need as social scientists to learn to code we need our ai people to go out and read some philosophy we all should sit down and watch blade runner once more and only by doing this can we come up with the solutions that we need for you know the next 50 and 100 years and the era of the singularity with that i'll kind of wrap it up i've got some conclusions i can bring in but we want to leave some time

for questions or more comments i'd love to hear people's opinions and and feed in and get in touch with me if you like so that we can work together on this my last job i was working with a lot of synologists and with uh roboticists out of munich all these ideas are not born in a vacuum and uh next year in person hope you can all come to rise of ai here in berlin and we can come up with ideas over a coffee thanks uh we give it back to you thank you thanks a lot um i think if you go over there we don't really have questions we only have praise for you which is amazing i get it from the students all the time no no it's it's meant to inspire thought uh not to inspire debate i must say i never trained an ai model but what i what i learned from at least my portfolio companies that what you described as the trinity from klausovitz is very similar because at the

end you need a lot of passion also to finish it but before at least there's something anyways um maybe is there any practical advice to people applying ai like really in practice let's imagine a startup building a product with an ai model or actually scaling the startup creating this superiority is there anything you learned from your python and research experience well i've read from many artificial intelligence scholars i think the camera up there or you who gets more attention hello people i've read from many artificial intelligence scholars that giving ai a conception of time is incredibly important it's an important next step forward they do have some mechanisms for capturing components of time already uh ltsm attention mechanisms and such so i would start focusing on those uh and i believe a lot of people are they're very popular right now so i'm not too concerned it's not getting the attention it does but being able to give your artificial intelligence a feeling for time and space is very important this is also why i

believe that the technology is looking more into the spatial elements of artificial intelligence really important or if you're just not really an ai person but you're investing in it think about time and space when you're thinking about your investments the very simple ideas about this are just for you could see it in long high frequency trading even right now how you position your hub where you're doing all the processing in the ideal location between two major trading centers so that the difference in time due to the speed of light in the pipes are microseconds but you can make a trade in that amount of time to leverage the difference in prices in different locations at almost simultaneous simultaneality but not quite that my couple microseconds makes a difference or space-wise you know buy greenland because it will be uh cooler for you to cool your ais up there right and just being very very boring here maybe with my question but what you always discussing was the strategy of obtaining high qualitative data and then

labeling it and that gives me some kind of data advantage superiority is that a valid strategy that you've seen also in your research where somebody can get the higher ground i don't know or is that just temporary effect and i should aim for more um if this is going to be relatively opinionated though i think google is with me on this i've read some stuff about them so you're talking about label data which would be much more in the supervised learning direction and that is apparently a dead-end path google likes to say they're focusing on semi-supervised learning uh the real cutting-edge people want to do unsupervised or reinforcement learning i know nothing about reinforcement learning i'd love to because i'm into these kind of games and such and that's usually what they uh utilize for you know alpha star and such so my advice would be no maybe don't focus on label data the problem is and i was discussing this with one of our colleagues here today europe i think has a huge advantage

in this long-term technology they've got the human brain project they've got great roboticists they've got physics research and you know cern they can lay the basis for really great singularity technologies uh if they skip over the label data bit the supervised learning bit problem is payoffs are longer for that will the european industries and government initiatives be bought out by facebook in huawei in the next 10 20 years because they're not making the profit immediately uh and they'll get bought up and then we won't actually have the european ai sector that we uh hope to develop right and then looking at christmas uh what do you put under your christmas tree as a wish from everybody that could help making ai more uh more whiteboards that's a three 3000 series nvidia graphics card so that i think every every ai researcher is kind of wishing for one of those under the christmas tree right now but uh one playstation is enough yeah i don't know i need at least two of them okay so um

yeah generally speaking that uh more broadly speaking i would love to just see everybody crack open a book that is not something you would normally find on your shelf or do what i did take a coding boot camp over the summer because you know kovit you can't do anything else anyways bars in berlin are closed do something a little bit to expand your horizons and i would hope that the governments as well would do more to encourage that as well create these ai spaces that some of our colleagues discussed this morning where people can get together and and shoot for the big ideas instead of just always making some more uh more apps for your phone that's a very nice finishing line thanks so much paul for that it's paul right jesse jesse jesse jesse jesse thanks so much for being here i think that was a great finish of the evening my pleasure yes there we go thank you all to everybody else and i give that over to you thank you

Rise of AI 2027 · Berlin · May 2027

Be in the room for the next one.

Ten years of talks started as conversations in one curated room in Berlin. The next edition brings 300 selected leaders together in May 2027.