Rise of AI Virtual Chat 2020 · Berlin

Create an AI Roadmap for Business

Christian Guttmann · Chief AI Officer at TietoEVRY; founder of the Nordic AI Institute

In this Rise of AI Virtual Chat episode 17, recorded in September 2020, host Fabian Westerheide talks with Christian Guttmann, Chief AI Officer at TietoEVRY, founder of the Nordic AI Institute and professor of AI, about how companies build a realistic AI roadmap. Guttmann, who has a psychology background and a PhD in distributed multi-agent systems, reflects on a decade of AI hype and reality: funding and political attention have grown sharply, but adoption inside established companies remains slow outside fast movers like Google, Amazon and TikTok. He explains that not every AI approach needs large datasets, pointing to few-shot learning, deep reinforcement learning and knowledge graphs, and argues that common sense and causal reasoning, not just pattern matching, are the field's next real challenge. He then lays out a step-by-step roadmap: assess the company's identity and ambition, distinguish real autonomous AI from analytics or automation, start with meaningful iterative use cases, centralize standards and research while decentralizing customer work, and size internal AI teams to company scale. The conversation closes on automation versus augmentation, emotion AI and Guttmann's optimism about AI's role in healthcare and drug discovery over the next decade.

Key takeaways

  • Christian Guttmann says AI integration into established corporations has been much slower than expected, outside fast movers like Google, Amazon and TikTok.
  • Not all AI needs big data: deep reinforcement learning can improve from single feedback signals, and few-shot learning aims to generalize from one example.
  • Guttmann argues common sense requires causal reasoning, citing Judea Pearl's point that machine learning fits curves but does not explain why events happen.
  • For structuring AI teams, he recommends centralizing standards and research while decentralizing customer relationships and daily project work.
  • Asked what matters more for successful AI adoption, Guttmann says clearly it is people and organizational acceptance, not the technology itself.

Detailed notes

Guest background and a decade of AI hype versus reality

Christian Guttmann is introduced as an "ecosystem builder": originally from Germany, raised in Australia, now based in Sweden, with a psychology background and a PhD in distributed multi-agent systems completed about 22 years earlier. He is a professor of AI in Australia, founder of the Nordic AI Institute, and Chief AI Officer at TietoEVRY, a roughly 24,000-person infrastructure and consulting company, while also advising and investing in AI startups. Asked how the field has changed over 10 years, he says the biggest surprise was how fast companies realized the value of pattern matching once large e-commerce firms in the US and Asia used it to personalize ads and feeds; before that, similar technology was often relabeled "intelligent systems" or "analytics" because "AI" sounded too esoteric. Westerheide adds that Asgard had four or five AI portfolio companies in Berlin in 2014 and over 200 by 2020, and that Angela Merkel once called the internet "Neuland" but now discusses AI.

What worked, what stalled: ethics, integration and hype cycles

Guttmann says he did not expect how much debate around bias, moral machines, robust AI and explainability would emerge, calling it a topic absent from his roadmap six years earlier and now a major field. What surprised him more, though, was how slowly AI has been integrated into established businesses: Google, Amazon and TikTok move fast, Siemens surprised him by moving well, but the rest of the corporate world, especially small and mid-sized companies, remains very slow. Both agree the field is not in an "AI winter" but past peak hype, calling it "autumn" and heading toward more realistic, proof-driven expectations rather than science-fiction comparisons. Guttmann also cautions against over-regulating speculative future risks, arguing that AI regulation, a reflection of a society's ethics and values, should stay innovation-friendly and be tested in sandboxes before being written into law.

AI without big data: few-shot learning, common sense and knowledge graphs

Guttmann pushes back on the idea that all AI needs massive datasets: deep reinforcement learning can improve from a single right-or-wrong feedback signal rather than millions of labeled examples, and one-shot or few-shot learning approaches aim to recognize a new object after seeing it only once, closer to how humans learn. He points to knowledge representation and knowledge graphs, used by companies like LinkedIn and Facebook to connect people and by Google Search, as an underrated area that retrieves and connects information faster than databases. On the field's next challenge, common sense and transfer knowledge, he cites Judea Pearl's work on causal and probabilistic networks: machine learning is good at fitting curves but does not explain why something happens, and building genuine common sense requires understanding causal relationships, not just correlations. He also describes an Asgard portfolio company that built swarm software controlling hundreds of drones with a single command, about ten years ahead of its market since industry and defense did not yet operate enough drones to need it.

Building a company's AI roadmap

Guttmann lays out a four-part approach for making a business "AI ready." First, assess identity and ambition: is the company product or service based, B2B or B2C, and what it wants to become in two to five years. Second, get clarity on what AI actually is, distinguishing autonomous decision-making systems, such as self-driving cars or automated ad placement, from dashboards, analytics or RPA that only display information without acting. Third, choose meaningful, iterative use cases that create visible value and win over skeptical stakeholders rather than trivial pilots or a single high-risk "big leap"; he references a Dutch company that agreed with its board and investors, years earlier, to accept lower dividends while pivoting its business around AI. Fourth, structure the organization: full centralization fails because a small central team lacks the knowledge held at the edges of the company, while full decentralization causes teams to reinvent the same solutions; Guttmann recommends centralizing standards and research, the approach he says Siemens and SAP also use with a central research lab plus local machine learning teams, while decentralizing customer relationships and daily project work.

Q&A highlights

Asked whether companies should build an internal AI consulting team, Guttmann says yes, especially to keep skills in-house, and gives a rough sizing rule: around 10 employees, buy AI capability through APIs; around 250, consider building some capability; at 5,000 or more, maintain an internal AI team. Asked about the relative importance of technology versus people for successful AI adoption, he answers clearly that it is the people, citing organizational acceptance, changing workflows and trust, such as a salesperson trusting daily call recommendations, as more decisive than the technology, and frames the whole shift as digitalization, of which AI is one term. On automation versus AI, he says automation and RPA execute simple, repetitive scripts that create value quickly but cannot learn or self-improve, while true AI systems have some awareness of their environment and can adapt from feedback; he prefers the word "augmentation" to "automation" for human-AI collaboration, since automation sounds like it simply removes jobs. On emotion AI, prompted by an audience question, he distinguishes needing emotions to be intelligent, wanting AI to have emotions, and being able to build them, and says recognizing emotion, for example to flag hate speech, differs from a system having emotions itself; he sees an underestimated market for emotionally aware systems such as virtual companions, more visible during the coronavirus period. Closing on the next decade, he expects growth in natural language processing, machine translation, virtual companions, healthcare robots and AI-assisted drug discovery, mentioning recent discussions on COVID-related drug research at the Royal Society of Medicine, and hopes AI will help address problems such as cancer treatment and transport safety.

Quotes

AI has always been around and has developed all the time.

Christian Guttmann

AI is not just machine learning.

Christian Guttmann

It's the people. The people are a major factor in this.

Christian Guttmann

Questions answered in this talk

Does every AI solution need a lot of training data?

According to Christian Guttmann at this Rise of AI Virtual Chat, no: deep reinforcement learning can improve from a single feedback signal and few-shot learning aims to recognize something after seeing it only once, so large datasets are mainly needed for pattern-matching approaches like deep neural networks.

Should a company centralize or decentralize its AI team?

Christian Guttmann recommends, in this Rise of AI Virtual Chat, centralizing shared standards, practices and research while decentralizing customer relationships and daily project work, since full centralization loses edge knowledge and full decentralization leads teams to reinvent the same solutions.

What matters more for successful AI adoption, technology or people?

Christian Guttmann says clearly, in this Rise of AI Virtual Chat, that people matter more than technology: organizational acceptance, changing workflows and trust in AI recommendations determine whether adoption succeeds.

Mentioned

multi-agent systems, deep reinforcement learning, knowledge graphs, common sense reasoning, causal inference, AI roadmap, centralization vs decentralization, automation vs augmentation, emotion AI

Keep exploring

More from Christian Guttmann

More from Rise of AI Virtual Chat 2020

More on Enterprise AI & Adoption

Transcript

Auto-generated transcript, not edited. · 12,175 words

Show the full transcript

welcome we are back uh my name is fabian westard i'm an entrepreneur and investor with a passion for artificial intelligence um this is our Rise of AI virtual chat this is a series of conversations with ai thought leaders and entrepreneurs and investors and people you know on the policy side or in the you know execution side people who understand ai who help us to teach ai and help us to adopt ai the overall mission is to build a european ecosystem that's why we do this that's why invite very interesting people every week here that's why you join us here today this is episode 17. this is live if you have questions you can use the questions function in swap card you can use the polls and answer them please introduce yourself tell us who you are what you do about ai if you're here new or if you're returning guests this is also will be a podcast at a certain time and you can watch the replay anytime on youtube when we have finished this

today so what do we talk about this episode episode 17 well i will talk with my guest about expectations and reality you know where's ai coming from where we today will be what is like reality and truth and for this as usual i point out to my t-shirt which is a good indication for you know what we expect and what really works um he called it our guest end to end so we will talk about integration and use cases about regulation and standards and my guest for this is dr christian goodman he is you can't put him in a single word like you can't describe me with a single word i call him an ecosystem builder he is one of us involved in many many different worlds and he's in the ai for years he is original from germany he grew up in australia he lives in sweden he started psychology but did his phd in distributed multi-agent systems already 22 years ago i think and that means he has a very diverse background he

comes from the brain side plus ai side i think this is highly interesting he's a professor of ai still in in australia he is the founder of the nordic ai institute which is more an ecosystem involvement but he's also recently the chief ai officer i tried to find a shortened for this i like its chief ai officer at tier every which is a software infrastructure consulting company in the in this space so these are are enough words i just you can talk with christian about everything and you will like him we like him that's why he's here today welcome christian hey hello great to be with you let me have a look at a chat uh wow people introduce themselves hello to everyone uh wonderful to have you all here um christian give us a brief and not a detailed description how what are all your activities in the ai ecosystem a brief prescription about all the absolutely okay and oxymoron but hey um look i mean i'm super excited that the ai is really

picking up now to such a such a large extent so you see activities across the board so i would say the activities that i'm involved in mostly is a lot around business so how do you actually make uh the different companies that we have in europe and the nordics become ai first companies you know how do they become aware of what ai can offer to them so there's a lot happening there's with my team here tier to every you know we're looking at a lot of those types of opportunities so that's one big thing you mentioned ecosystem builder so it's a good good i think it's a good description in many ways because well before we start really building out and making many companies ai ready i mean you also need to build the right let's say frame right right infrastructure and so that leads me often to discuss these topics with government officials uh travel to brussels before kovit a couple of times talk to commissioners and a number of people as one example

right and and how do you lay out what's the framework and i think uh at least you're aware and many others perhaps are way in the in the call you know there's a different ai regulation proposals data strategy proposals product and liability that all shapes the ecosystem right and then of course i'm very active look there's so many new startups in this space there's so many new initiatives i'm more than happy i'm i'm i'm a startup advisor and angel investors so it's great to see these new companies popping up and pushing pushing the boundaries so i guess that's sort of in short and then of course maybe one more is of course the uh being in the academic world so i'm as you mentioned before i'm still in the academic world so pushing that and being at the cutting edge it's quite exciting so what i like about you you have this holistic view i mean you are in a large comp corporation 24 000 people but it's a tea company i mean you have

startups you are teaching you're lobbying you're a public speaker i mean you have been at Rise of AI conference before you will be at our summit this year um in an even larger role you know we're trying to involve you building this um and you've seen the whole ecosystem for quite a while tell us you're you're feeling like where has it been like 10 years ago was it today like what surprises you did you see it coming and what came where you say oh i didn't see this 10 years ago and where do you say 10 years ago we were expecting this and we are still not there yet yeah i would say personally yeah i'm i have i always expected it i guess it was my you know dna somehow because i believe in that topic so much so it's a very exciting topic but if you look back 10 20 30 years ago right um if you start with 10 years ago i think you saw already the application of ai tools you

know particularly machine learning i myself built a fair bit of those types of technologies but of course this major uptake and it's probably more a realization of technology that has been around for some time the realization of many companies that you can create a huge amount of value uh in a short amount of time you know like uh the big companies the big e-commerce companies uh from the us particularly and also from asia they just have recognized you know this pattern matching uh it's just so valuable uh to start personalizing ads and and and feeds and all those types of things so so i think um that what i'm a bit maybe what i didn't foresee is that there's such a big realization and and understanding of what this technology can do and um but you know ai has been around and i think another point to mention is that in some sense ai has always been around right but when i was in business i don't mention the companies but we couldn't quite call

it ai so it was unsexy people ran away from it it's too esoteric you know and we called it often intelligent systems uh smart cities uh uh you know analytics you gave it all sorts of words which were let's say more uh in germans along faith you know so you would be able to talk about them and don't scare people away let's say so it has always been around and has developed right all the time um i i understand this feeling that ai emerges a new term i mean ellen turing spoke about ai et cetera and then you know it's science fiction i came from the science fiction era i mean i always like to tell the first Rise of AI meeting was called berliner futuristan it was 10 people meeting in a hacker's space which was a spaceship and we talked about the singularity might be close honestly it's not close i don't know if we will make it by 2045 then my first ai conference was agi conf so agi is a very

old format where ben Goertzel introduced the first time sofia which later became sofia the robot but the first prototype i saw it like four years before it went live in tv for me this experiences i saw this really nerdy topic where everyone was applauding because he built a robot who acted in real time and then four years later it's a tv and you see pictures with jimmy kimmel and you see angela merkel talking to this and it hit mainstream but it did not improve in technology it just became awareness mainstream um so where where do you see like integrations there i would say technology is only solving one percent of the problems 1999 is still execution and we will talk about this 99 soon but when you say like is the death illness movement like where are we heavily missing and where did we over achieve it like what would you be proud of as an ecosystem yeah yeah i was just looking at the background change that you had so does that does that

mean it's enlightenment no it's like it's yes it's hope it's hope that it will you know but we have many pathways to go yeah yeah no good good um yeah look i mean i can just say i mean i'm not the biggest fan of uh you know and and i think you and it's in part related to the second part of your question you know the disillusionment or you know getting a bit of a different sense of what reality is but let me say this i mean science fiction is a very powerful uh tool right so because when you we humans when you can imagine it you can build it right so you have and and when you look back um a lot of those science fiction ideas they have been driving the community towards building it right i mean and it's also like if you want these other early use cases you know when you look at jean-luc picard and the way he talks to the star you know yeah so you know you

can imagine it or i don't know knight rider for the or lack of a better example here but you know but yeah you know you get a bit of a sense and then then actually it might be worth i might be deviating a little bit but i think um the good thing about science fiction and robert heinlein he was one of those big science fiction writers you know azimov heinlein they were all in the same category and he was very very uh clear that there is science fiction and fantasy fiction and science fiction from his point of view is always based on scientific thinking and theory and fantasy fiction are things that have no basis in science it's magical because it's magic and it's me yeah it's along those lines exactly when you start having fairies and so on you know so but i mean that's just word i i just uh uh you know uh read a bit about the um so about when i get to the question yes and i will point

out questions yes what can we be proud of like everyone here is in the ai space more or less but we always like don't have always the timeline what is an achievement what we can say good for my view funding it exploded like coming from very small level exploded when i started 2014 with asgard i had four or five ai companies in berlin today we have over 200 so from four companies to 200 companies yes and politics improved a lot i mean for angela merkel our chancellor internet was neuland today she talks about ai i mean that was another jump too i mean we have an ai strategy it's the first time that we have something what are other things where you say hey this worked out it's good it's a uptick yeah no that's a good point and i would say um it's increased from my perspective now it's an increased there's a certain level of maturity the types of discussions that i'm having the interests in the sector the let's say the cxo

levels and the board of companies there's much more happening there in terms of thinking and really feeling pushed the fear of missing out you know so it becomes a much more tangible topic right and um i had a very good uh discussion with the chairman of nokia who i think is quite known for actually having taken an action rather than talking theoretically about the topic he went to the us and actually learned about it right so you start having this maturity of understanding in the in the in the business suite and that of course generates more testing environments so i think the number of proof of values proof of concepts you know is has gone up and i think it is important is a very important part you need to really iteratively uh get into this topic and i think we see a big uptake there um and of course funding you know as you said i'm looking a bit from the perspective of what type of startups and you know companies and and the

public sector what type of funding they are receiving these days right because that's so important for them to start testing and so we see a big uptake there and i mean another another measure you know it's now the number of phd students for example that are in this space it's like explosion right so everyone wants to learn about it but yeah it's great and um and so there's a lot also approaching me which is great so keep doing that you know and and that's that's a good thing many in our company in our our customers environment many of them have a very high ambition right so they really want to get into they see it and it often comes um it often comes from the technical engineers right certain technical the technical folks they really start getting their teeth into it and they really wanted they asked me a lot what should i learn what courses should i do what books should i read do you see a lot there and and i think i

mean maybe we come later to it but it's of course all about breaking it up down to what value it creates right what sort of business does it create no matter your enthusiasm it always needs to be turning into so more technical people go into it more researchers more funding more startups more integration on the other hand what doesn't work and as i said we we still have problems like this yes the machines are coming and um we still have challenges from not enough data not labeled not quality we can have improvements in usability yes uh i mean this is still missing a lot it's not so i would say the largest challenge is one first expectations by mass meteor oh it's like ex machina it's like terminator no it's not it's honestly very very very very far away from this it's very dull it's very boring if you really think about ai it's technically exciting but it's it's storytelling boring yes and the most exciting thing is a twitter bot which gets racist like

attention-wise i mean that was human fault uh i think we have learned a lot about ethics this i didn't have on my roadmap six years ago that we will discuss so much about bias system moral machines robust ai explainability that wasn't on my roadmap i think that's very huge and will be a huge topic and what i didn't expect it will take so long is integration into companies and businesses so google is fast amazon is fast you know uh tick tock is brilliant fast i mean they want to buy just for the algorithm but then if i look at i at siemens i was very surprised but if i look at the rest of corporate world middleston very very slow very slow and i think that's why i think it's good that you are here and that you do what you do and then we will talk a little bit about this how can we bring ai more into on the street but this is like where i say hey we are not in the

winter but it's auto mich and the time it's getting colder it's less hype more realistic now prove what you can do exactly now you brought up a lot of very interesting points here right so uh where to start which one to bring up let's start with like like yeah what's missing what's not optimal um if you think it's a getting winter in autumn or if just being spring all the time yeah i think we might be in spring for a little while longer and i think the reason is that there's so much application so many applications yeah right i mean when you compare winters of ai in the past in the 70s and the 80 it was in the 80s and in the 60s right it was more that people discovered some technological limitations they didn't understand at the time but now it's really like a slow i would say it's a reasonably slow uptake right um uh and so we are in screen i think the excitement is there i see when i'm talking

to customers now across the board whether it's manufacturing or retail or uh you know health care and so on i mean you you see the um the ideas around use cases and application cases with values bubbling right there's more awareness people have have more uh you have more more awareness right and so this is also to the point of what type of culture and mindset do you want to want to build um and then you mentioned what was it talking again like what is what is not working it's like like okay let's integration is lacking and that's slow what else is not optimal like where you thought we would be further down or where we got stuck for example for me the ethical debate was way too high and it was not enough debate about how do we get it on the street how do we build an ecosystem how do we make it real yeah yeah i mean look i mean what i'm seeing now is that we have to be always careful uh

with um how do you want to call i mean too much yeah i try to say it but i think that is of course value i'm thinking of regulation right and um and i'm not saying that there should be any regulation so i don't want to be misunderstood i'm trying to choose my words very carefully here right but it should be very innovation-friendly uh regulation which is coming in and a regulation and a law is a reflection of our ethics and our values on the morale that we exhibit right so it's actually quite a complex uh scenario and uh you mentioned psychology before that i studied i mean it's not really not a trivial thing to do um we have to be of course a little bit careful uh that we don't put in lots of regulation about theoretical topics right i mean i compare that a lot i think you also popped over to silicon valley and other places but usually when you solve problems you know you're not starting with well what are

the problems and what should we be regulating right i mean what should we not be doing you should be thinking about what could we be doing what are the opportunities and if this is so i i feel sort of if there's and you need test grounds right and if this is going in the in the wrong direction uh and you see that in those sandboxes in which you can experiment well then you should really think about you know what type of uh uh what type of regulation you might want to bring in for example yesterday germany upgraded its monopoly law uh to regulate google amazon et cetera digital platforms more for example if google is advertising themselves higher than their customers not allowed to it's interesting they also say for example amazon the clients who sell stuff want to have contacts for their own clients you know who pushes something amazon is keeping them and then it's conflict gdpr says no you can't forward this but ai terms wise you want to have the data

and you want to have the benefits of it so it will be very interesting to reload i agree it has to be done on the case so and you mentioned something else i remember you brought up so many very important topics but for example this how do you make a data accessible data is one factor right not all and this is something that some people uh misunderstand also not all ai needs a lot of data right so there's one particular area which is machine learning and that area when it comes to pattern matching and so on so that needs that needs a lot of data as of today with the technology we have like cnns and rnns and the neural networks right so that's that's one thing and of course that type of data there's a lot of value having access to it you mentioned the smaller medium-sized businesses you know and there's of course there is an issue that is bubbling up which is how do you make these smaller companies whether it's manufacturing

or elsewhere how do you allow them to have access to this type of data to stay competitive because if you have these very large companies that can technically collect their own data about the manufacturing process they might be sitting over there on their on their billions of data rows right but then you have a new newcomer right a new startup that comes in or a smaller smaller business and they want to improve their products or their sales they are heavily disadvantaged and i think there we need to be careful that we don't build these mega companies uh that have access to all this data and this is in fact reflected in the discussions that i'm i'm a big supporter of an open data system that you have like a gnostic shaft like a union you know where you pull your data and you analyze it the thing is b2b data is often not personal so it's not about personal privacy uh b2c something else so we can and b2b share way more data and have

to do it a couple of questions came from your statement first one you say not every ai solution needs a lot of data tell us what ai doesn't need a lot of data oh i mean there's plenty of other topics when you look at ai right it's divided into many different uh many different areas like you have for example um what do we have you know robotics or i was thinking actually even within machine learning there's now a lot of research happening where you have this one-term data or one you know one or several uh turn data learning so instead of having the system learn over millions and hundreds of millions of iterations right and having huge data sets it would learn based on a type of common sense basis just like you and i would learn something so if i show you a new item like a pen or something right you i show it to you once and you will recognize it next time right you don't need to see it like a

million times in order for you to recognize so there's a lot of research there another one is deep reinforcement learning right so even there you need essentially one piece of feedback you tell the system well that is wrong whatever you're doing it's wrong and so it learns based on that type of feedback and this doesn't require a lot of data it requires essentially one one um you did it well or you did it bad tight feedback right so that's just within machine learning and then you have you know natural language processing you have a big area called knowledge representation you know it's often underestimated but if you for example it's just a matter of how do you represent the data effectively efficiently not in databases but for example like linkedin facebook these social media sites they have topologies so they connect people you know and you have the connection then the data is connected yeah exactly so so they have there's many many multi-agent systems the area that i worked in it's a lot about

um how do you make multiple ai systems work together right and so there you need to think about okay social choice theory for example uh multi-agent decision making group decision making which was part of my thesis uh you know so it's a very different and i say your your idea will return in a couple of years so i had a portfolio company with asgard which was called excellent dynamics and they had a swarm intelligence so at the end they were able to control hundreds of drones with one software you know one order this like do abc protect this area and they did the pathway and everything together and um i love this great idea but they were ten years ahead of time because industry and defense didn't have enough ai machines out there they had like one drone and one robot they didn't have hundreds so they saw something which was too early so your idea was even earlier but it will return one day i'm sure um i also felt that in research what

next step is to solve is common sense and transfer knowledge like both is missing how do you want to teach the ai common sense i mean this area has also been around for a long time in the 80s it started right so it's not something that just recently popped up how do you teach an ai system common sense it's a big question but of course one one some of the attempts you know is essentially to start building a corpus of knowledge on which you can and build right so what that's expert system uh what type of abstractions can you can you create from that right so what type of let's say stereotypes about the world uh can you extract how can you i mean the whole idea behind common sense is a bit like how can you see the connections in the background you know how do things causally related for example right so that you can think ahead and uh just to name one a good colleague of mine is julia pearl you might

have heard of him he's he's extremely well known on heaven has made major contributions to uncertainty probability networks and when you hear him right um he would he would say that all this machine learning is good and fine you do sort of you fit curves good exercise but you don't know the reasons why something happens right and he has been extremely known he got the turing award and so on um and i mean you need to you you gain a lot of benefit from uh understanding the causes uh between different actions and about the world and that gives you a basis to build some type of common sense you know uh and that allows you then to essentially be much quicker so you can throw you can throw ais in the world that will understand the world my personal thesis is that all these pieces have to come together for example one portfolio company is uh called max industries and they're teaching robots to learn so it's very similar they see a human movement and

then they copy it but the objects are each time different so they they just learn the idea of that's a ball i grab it they have to learn this that they exist and then they can do it with every ball and they teach it and improve itself so this is the direction where it's going but it's as you say it's not data intensive they don't need a lot of external data it's a lot of machine learning a lot of machine learning building this world model building this adapters to the world giving this common sense and so forth so it's very challenging to solve yeah exactly and i talked to talked to a guy in the um he was also one of those early veterans you know the early pioneers uh and i remember very deep discussions quite a while back but you know he was he was really demanding that when one claims they're doing ai or they publish a paper in ai they should always state in the beginning how this work they are

doing contributes to ai and so ai is not just machine learning ai is about building machines that you can that are smart more aware conscious they have common sense they have a whole bunch of uh let's say what you want to call it your characteristics which are which we would associate with uh with people with humans right and so there was a he was quite quite clear that that is something you would expect in papers but today it becomes a bit watered down let's say right people throw themselves i honestly ai doesn't fit because we don't speak about artificial human intelligence i prefer for example machine intelligence because at the end it's always a machine and it's intelligent in itself so we have the first audience question for you and it is from asking is there a place for graphs in particular knowledge graphs in the current deep learning dominated ai landscape and if so what benefits does it add so first of all i believe so very much so and in fact there are

a number of companies that are using them these knowledge graphs so now i'm not entirely sure which type of knowledge graphs you are meaning if it's sort of semantic models or topologies about people let's say but uh they're certainly used in as i mentioned before social media like linkedin and facebook and so on the way you're representing information and knowledge i think google search too google search is a very interesting knowledge graph system yes yes i think the knowledge graph and the knowledge representation is a big that's a topic both in psychology the way how we you know memorize things and structure things it's not a database it's not just facts you know that you that you're putting together it's a lot about well how do you connect concepts yeah so so i think it's very important in terms of value i mean if you go a bit deeper from a technical perspective it's it's simply that uh you can retrieve so many examples but you can retrieve information much quicker right so if you

were to want to connect take linkedin as an example i don't want to make any particular advertising here but if you want to search for someone right you see these degrees of which people are distant from you and and of course that is something you can do significantly quicker by um by you know connecting the uh connecting or having a representation around the topology rather than having a database so there's a lot of savings in terms of search and and then of course you can detect uh patterns right in the way how people are connected with each other and so on so i i agree i would say it's even an old ai field and it's strong there it's not so popular like deep learning sure but let's be true deep learning is from me the reason why we are in this uptick again why this became real because it was rediscovered and it was the right tool for abundance of data and a lot of patterns and objection requisition i think it pushed the

whole domain a lot forward and everyone out there nlp and so forth saying like oh we understand it you get more funding and attention too deep learning is not the only solution we see it here but it is a reason we should be thankful for deep learning and now we continue with a whole bandwidth of technologies yeah fair enough enough and exactly as you say i think it's a realization of these huge uh benefits and values that you created once it became known how the big e-commerce companies did this pattern matching right and that you make essentially billions of dollars from that and you can really map that and clearly see that uh i became a no-brainer i think that's really you needed this catalyst you needed to have this understanding in the system and realized whoa whoa what's happening and i i honestly i enjoyed tick tock very good algorithm pinterest very good yes amazon brilliant yes facebook you mentioned the facebook honestly sucks but uh linkedin was really good you know recommendations all

this i mean it works really fine i mean happy with this it can be more you know my theory is still like it doesn't sound me right like each time i sell put in milk it says something very random and it's like it's not optimal but it is convenient i enjoy going upstairs and saying like alexa light on alexa tv on alexa shut it off it's like cool it feels a little bit like star trek on the ship like computer do this to that uh i enjoy it so we said we want to speak a little bit about ai roadmap business so let's take a little bit of time for this um how do you start like like when do you say in companies ai ready means they should consider thinking about ai what are the needs you have to have first before you say no you have a lot of other homework to do or yes you should consider ai as a feature product strategy element yeah yeah so let's break it down a

little bit uh with a few questions right and then what i can do is sort of provide a bit of a vanilla approach you know how could it be uh and then we can and then in every of those four steps you know there will be so many questions and issues coming up but let's try to do it that way and i think one of the first questions you need to be aware of uh nice background you would like to be aware of is um is okay which company are you i mean you have to ask yourself which company you are today uh and actually i would be interested to know that to make some type of roadmap assessment so are you for example a product or service company are you a b2b or a b2c company right so there are some very basic questions you need to review first i would say because that determines where you're going and of course also which industry are you and is it private is it public which

type of private are you in are you manufacturing you know and let me also say that i'm now what i'm explaining now is really from the perspective of many perhaps 89 of the companies that we have in europe i would give a very different type of advice if it was about startups let's stay with b2b european company and it can be a manufacturing but let's let's assume b2b so but i just want to say that on the outset right so then of course you need to do a little bit of the homework i think the first step is to really look at what are your ambition levels i think about basic questions on uh your maturity uh what company are you today what identity do you have today and when it comes to ambition levels it's like with any good strategy strategic question uh well what type of company do you want to be in two or five years time right so that these are questions that you might want to ask yourself if you

look at it from a leadership perspective from the cx cxo perspective on the board right if you have a bigger bigger ambition then of course um you know you want to see again you know where you are what success state you are defining yourself to be in uh looking at the competition looking at what others are doing what value are you offering to the system and so on and so on and then of course what type of capabilities do you need so these are some strategic considerations that i think you need to ask yourself in the beginning then the next step second step is a lot about okay let's assume step one strategy and management phase efficient levels yes maturity what's the outside where are you starting right and that can be any i mean this is a lot of discussions i'm having with many companies right i talk to the ceo i talk to the board members the owners of companies that are anything from i mean large companies they have big revenue streams

right but they are also to be clear they are not data they are not data natives right they are obviously let me jump in here when you talk and you have a feeling for the market like how real like startups they are all about ai but how what would you say like how much is really currently on the your clients and non-clients i mean if they're client of yours they want to do something but there's a huge dark market not content in you what do you say like well i mean if you look at the let's let's call them what do you want to call them established companies that have been around for decades or 100 years or something it's it's still very limited right i mean if you and then you talk about real ai and there's another aspect in the first phase that you want to consider which is clarity what is actually ai versus business intelligence or analytics or rpa they throw sometimes it's all thrown in the same pocket and it's

very very different right but it's it's quite limited ai in my world has a lot to do with having autonomous systems that take these decisions not just the dashboard that shows you what could be done which is more analytics but ai is well you have that system that does something i mean the typical example are self-driving cars or when you have on on facebook or google ads that are essentially uh presented automatically right there's no human that clicks uh does something manually but that's that's sort of um you know strategy and goals what's next like okay let's say i have a strategy i want to use ai i have an assessment and analyze the marketing competition so far but i realized hey we don't really do a lot of automated like this we still have a hundred thousands of humans who are stuck at home office yeah yeah yeah exactly right um look i mean and also all the steps i'm mentioning now they don't have to go in a sequential way but the second

step that you need to consider or the second point you need to consider is really well how do you go about it right it's uh and usually one good start thinking uh you know way to think about it is that you are looking at well what's most meaningful what type of use cases are what type of application scenarios are the most meaningful for the company and i use the word meaningful because they don't have to be the most impactful they don't have to be the ones that create the most value they should also not be trivial trivial small three-month sort of projects right but they should be they should be projects which are showing some value which convince many stakeholders in the organization because you will have a lot of people that are not opening their arms to these types of technologies in fact the majority of people will not want to have that you know they don't want to understand it they don't want to embrace it it's very complex it's more on their

daily tasks than what they what they would want to have right so you have that that aspect so so having mean of meaningful iterative uh sort of approaches so and iterative is also important i we recently discussed with the customer that it should be every step every small step we are taking should create some type of value to the organization so that is something that is actually quite important to consider i think that you are not jumping in i mean you can and you this is also proposing and i have seen this execution that you are looking at the other side of the spectrum instead of doing it iteratively you're actually doing it in a um in a big leap but then you have to talk to the the owners of the company uh you have to talk to the investors and you have to tell them guys this year there's no dividend there's going to be 20 less revenue and but we are we are putting ourselves up to be the business anyone who

did this bold leap there is uh they had two companies in uh in the netherlands i can look at the name now they don't come to my mind but they there's um there's one that did that they had a very thorough discussion uh actually now three four years ago and they said hey we really turn this business around and they had an agreement with the board and the investors to go this quickly but i agree with you that is not usually coming i mean usually when you bring up that type of topic as let's say the ceo leadership team with the board and with the investors they start uh looking for a new ceo potentially you know so you have to be a little bit careful how you bring up these types of topics so for now i understand i mean i could exchange it with blockchain and the other technology you need to convince the ceo and then you need to convince the stakeholders within the companies to support your projects um let's talk

a little bit okay let's say i i i have the support by the ceo i have the stake order i have a project with value i mean otherwise i shouldn't do it it's a smaller one it's iterative um how do i get started like do i just hire a bunch of coders or do i have a project like like do i buy a company do i build it like let's assume they say hey ai is cool but i have no clue where to start we just know that we want to do something yeah okay look i mean there's a lot of options then right so so uh so everything you're mentioning is of course uh on the board right as a choice and i should also say everything i said so far is a bit top down right talk to the cxo and the investors that's an ideal situation but of course i strongly encourage the organization also to build things from bottom up so there should be a lot of incentives possible for everyone

anywhere on the ground that uh you know that is uh working their way upwards that show value in the system that way but um so how do you go about it uh it again depends on what strategic decisions you have well i can tell you what doesn't work going back to some other things or let's say what is much less likely to work is for example this idea that you centralize all the ai activities in a company when that has been done and i i don't know any organization has truly succeeded in it uh so you centralize all the efforts or you decentralize it completely and by centralization what i mean is that you have a type of if you want to call it the center of excellence or what it is but they own every single ai initiative they own all the uh all the interactions with the customers and the product and they are essentially responsible for bringing it forward the reason why it doesn't work is that they often don't have all

that knowledge that is at the edges that's in the rest of the organization right so you cannot um you cannot build a small team that really overlooks all the activities right and and also if you go with that approach you need to give that type of team all the power to make these decisions and the changes in the organization which is really tough if you have a company of let's say 50 000 people and you produce whatever it is you produce a car a lawnmower and then you essentially tell everyone else what to do it's very very close to impossible decentralizing isn't a good idea right to just say well all initiatives everything is welcome everyone is doing their own thing because what happens then is that you sort of reinventing the wheel over and over again i mean one clear example now is with the uh with the regulations that are coming in you know the eu regulations and so on uh that's something that i'm for example looking at a lot but there

is no point that the rest of the organization is getting themselves into this topic as deep as we are let me that's there for example siemens oliveira two weeks ago talked about it's very similar process convince the ceo on the business units what they have is like a core team like a research lab but they have machine learning teams in each business unit so they build the products very local but the research is centralized in a certain way and with support they are like task force helping them um i spoke with sap and sap has in research labs too and is similar um so i lost the thought there maybe you wanted to ask so so i mean they i think the i think there's to be very honest when you talk to uli the question is what would you centralize and what would you decentralize like what what what job levels should you have in your local team and what you don't like the researcher for example the coder the ethics people the regulation

people exactly i mean the good a good rule of thumb and that's a bit how we think at uh tier to every also is that the centralized day item is a bit the stem so we're creating standards and practices that can be reutilized across the company so as we are going in and spearheading certain projects uh with customers let's say how does that reapply in other parts of the organization right so who can who else can take it and what parts of that learning that we have you know can be can be useful and can also be easily adopted right so we sort of document and we write down what needs to be done and you know we're also looking at the um the code and the assets that are built and so on so building up a proper knowledge repository let's say so that's a that's a very that's a i think a good start that's what we decided to do then of course uh as i mentioned before you know what what should

be decentralized is essentially the connection with the customer and the existing ongoing projects i mean this is knowledge that is on such a detailed level it should remain on the it should remain in the individual teams it should be decentralized right i think there is no point to take over customer relationships and centralize them he will be very quickly overwhelmed um and then of course you need to think about the different grades of skills that you would want to have centrally it's maybe a little bit along the lines of what will you mention that you have research century but i would go a bit further a bit more expanded i would say that that's how we internally call call my team a bit it's a bit the uh i don't want to talk down anywhere else but we talk about the a team you know like the ones that when we go in and we start looking at the really hairy projects you know the things that are let's say non-standard where customers start to

want to have something which is not not done before right and so we we have our phds that have knowledge in a particular area and they really apply themselves to that so um i mean that these are like two two big topics i mean we also it might be worth mentioning i so that i could sleep nicely at night i actually introduced the um ai ethics principles and certificates in our company and i think the certificates we were the first company in europe that introduced them so everyone that's an ai and data scientist although it's not legally binding uh well we don't we are not legally required to do that as a company but they are going through this certificate and are aware of what it means to build ai systems you know their impact and so on and so yeah that's something that that is done should be done centralized no point doing that like two questions from the audience for this um first is is the idea of almost internal ai consulting company

for enterprise effective so if i understand it should you build your own ai consulting team yes i mean poor share consulting for example is existing probably consulting's existing yeah um i would say uh yes more than before um because you want to have the skill sets internal and you need your own research unit which you didn't have to do about your view on this because you are external sorry you are external you're not the internal guy you mean me i mean you're internal consulting within your own corporation but you go to clients and help them from external so should they build a skill set on their own should they have their own ai consulting business yeah internal and by internal you mean they help other teams inside right so having a bit of uh uh supporting the other teams essentially is that internal other questions is like when do you start to have ai skills within your companies versus when you buy them if you only have 10 people out there and you're you're a

small shop you you just buy something per api if you have 250 people you know small manufacturing i would consider this if you have 5 000 people or more you should have your internal ai team in various degrees but your view okay so i'm still not 100 sure but the the the the the question per se but i mean the internal team if you are a company and you want to support the different areas in your organization uh and you need to build up your business then you have to ask yourself if it contributes i think one again a good rule of thumb is doesn't contribute to the core of your business does it contribute to the core of what you're doing right so that's that's uh that's one question and if you are using that internal consultant as one that is um supporting you know and increasing and improving the core business and the value of that core business and i think it is a good idea um internal ai teams if you are

speaking about something like the ceo then i think one thing you just need to be aware of or want to be aware of is that they are not uh you know responsible for things so they get the mandate for things which are taking away the speed you know of of other teams uh that you know so you need to be quite careful how you're slicing the responsibilities and how you're giving the internal team um what what type of tasks you give the internal team versus the rest of you i'm not sure if i got it yeah i would say it's a tricky question like when do you start to have in buy or build like that's a little bit money right that's so yeah right pontoon's second question is from pontos he asked what would you say is the relative importance of technology versus people slash organization when it comes to successful implementation or usage of ai and business right okay so there i can say quite clearly it's the it's the people i mean

the people are a major factor in this and i can tell you uh many of the discussions i have with customers or when i go to conferences i highlight very much that it is a lot about the people about the the humans about the acceptance of the technology uh and the the technology while it's sort of what you want to call it the core of the machine right the nucleus the engine where all the where all the difference comes from you know um but you need to be aware of the business uh you know the people the workflows the responsibilities that they have today what tasks are changing what type of uh what type of skills new skills do you need so it has a lot to do with how it changes the organization and the people's responsibilities then trust building trust into the ai technology so if an ai system tells you every morning whom you should call as a salesperson well you better trust the system because by the end of the year

your commission depends on that type of recommendation right so you have that part and then of course i mentioned earlier and it's a lot about strategy and strategy is also in part based on the culture right the culture and the values so and values are something that the company has to define some do it together with their employees and some do it you know depending on how the market is but all those types of things so my theory is that whatever every company has to do out there has nothing to do about ai but digitalization like every business model needs to get digital it needs to be adapted to 21st century and ai is just a term a personification a tool set but also making it a little bit more more real and abstract for a ceo it's for so easier to say okay we want to have ai whatever it means then we have to digitalize and do data management and you know process optimization and so forth i think it's it's just it

open up a little bit way more structural problems which we have for all the time let's look at automotive industry let's look at e-commerce versus stationary you know entertainment media business etc um have a couple of minutes left um and i would like to tap a little bit more your brain about you said something building something intelligent like this was the core when you studied what is for you an intelligent system and like give us a couple of examples like from daily life or business life we say hey that works that's cool and you know maybe something where you said this failed okay so first when i say intelligent system right i just it's a it's really something that i think many in the ai community would say that you know you have a system that has some level of awareness of what happens and how you can interact with it right so you can start predicting what you're thinking it can do tasks that you wouldn't expect to to do right because of the

complexities so that's that's certainly the fascination i think if the company claims to have ai or builds ai whether it's a starter or a big organization they must have that type of vision in their roadmap otherwise i personally always put a bit of a question mark to it as to whether they are an ai company or whether they want to do automation which is absolutely completely legitimate uh but it's not really than completely active let me jump in where do you see the difference between automation and ai oh i mean there's a number of for example i think there can be many um overlaps but automation and ai for example automation rpa would be a lot of automation there's a lot of topics about robotics process uh automation as you know and then you have very very simple scripts you know that just very highly repetitive things and you and i could program that in a day right so that's not that's so simple let's say that you it's it's very old it has a

lot with automation and can create value in a short period of time but you would never call that an intelligent system right you would never never expect that system to learn so much about its environment and the business that it would be able to self-improve itself for example does this conflict because i feel like okay i bought self-executing a script that's automation but at the end i want to have an automated self-learning self-improvement system which then needs feedback from me so therefore it's not 100 automated because it would say in 10 percent the cases hey i have a flag here please give me feedback so i can self-improve so autonomy i mean autonomy is an important part but autonomy doesn't mean that you completely isolate the system from every input around its world right autonomy means that it makes a self-assessment of the knowledge that it has about the world right the ai system so if you give it feedback then and it can evaluate whether that feedback is valuable or not then you start

having a system that is uh starting to be more autonomous let's say but that doesn't mean that you should not as a human have an interaction with the system right so so that's just one one point and i would also like to highlight i mean there's a few words here like automation i would often prefer to talk about augmentation too when you have this uh human ai uh collaboration i think there is so much more happening in this space and automation sounds so sometimes a bit brutal right it says ah ten thousand people company now it will be only a hundred that run the comp it's not like that right you should usually it's more like you have augmentation so you look at opportunities and tasks which can be augmented when you add um ai to so your partner in crime like your night rider your kit you know that helps you every day to uh to do something quicker and better and particularly these mundane tasks right so i think they're there that and

again this is a mindset question you mentioned earlier what don't we do so well um other cultures sometimes a little bit more let's say successful in in creating that bonding between the artificial and uh and what you are doing in your organization like the japanese culture i was thinking pretty much along those lines by the way which is interesting i think it was in the 60s when germany and uh and japan were growing after the second world war both needed labor and the japanese said no we don't import labor we build machines and the germans said no we import them mainly from eastern europe from turkey from you know uh spain from italy etc so a different approach germany went brute force and said we take a lot of humans uneducated and the japanese said no different different approach we have one question one question from charlie asking you with your background in psychology what are your thoughts around emotion ai hmm big topic i think first of all some there's a more basic question

what what do i emotions required in order to be intelligent right my wife would say police no emotions for ai so there's no emotions no she doesn't want that she doesn't want emotionally i had this discussion before well i think they are important they are definitely an important component of us being exhibiting human intelligence you know i think we make a lot of we have a lot of gut feel decisions we are making you know many of us have read daniel is gut feeling emotions because i have to privately emotions are good because it's human but in business mostly it's not about emotions about honestly pure execution and thinking smartly like a lot of work is like this do i do emotions into a robot who's driving my car do i need an emotional car or do i just need a very rational effect for the system but do you think if you were to have rational thinking that anyone would ever start a startup or invest in a startup no that's exactly the passion

is something else i don't speak a passion self-driving car do i want an emotional self-driving car yes uh or do i like like happy speedy or self-destroying or do i want no emotions i like i don't care if do i need emotions let's first do i need emotions a second do we want them and third can we do yeah yeah yeah so so so i think you do have some of that and when you said passion right when you have passion it is quite irrational i thought i mean when you started totally the you need a leap of faith right so you need some that that's very that's much more emotional because you tell yourself it's possible and the next step will be the you know we will be the winning big next time yeah and it's a mix between passion leap of faith i mean there's a whole bunch of i mean it depends now we can have a whole discussion about what that is do i need emotions into an executing machine an

automated machine do it in a dashboard or something do i need this yeah well i mean again if you define emotion as something that you would call empathy for example and uh prediction i mean these are important aspects that i worked a lot probably for seven years in this human computer interaction right you know how do you build not just chatbots but how do they start detecting what the emotional state is right and how they react to it support i agree i agree if facebook needs to know oh that's a hatred yes the emotion of hatred needs to recognize i agree but do the ai needs to have emotions itself not not recognizing not using it not playing with it but do we all want an emotional ai i mean i could think of a lot of industry areas that where you need that i mean i mean when you for example look at the elderly homes and healthcare right so you know that in the future nurses i mean it's it's recognition this for

example in the psychology literature it's it's practically impossible to be recognizing pure emotions of someone else if you don't live it yourself so it's it it's very very hard for you to recognize something that you have that you have no relationship to so so you know um so this this a lot of discussion in that space and for those purposes i would say there is use in having systems that are emotional but we can again we could expand a lot i don't i know and we have one minute left but my thought was like hey do i need to have a facebook algorithm who understands hate to block hate in comments yeah or does it just need to know okay this is hate i don't care if it's hate or not but i was taught to flag everything which is hate speech and i block it and some human has to to ex give access to it or not again super complicated because i mean for example what do you categorize as hate right so

would it be to the majority of people or would it be who is most uh inflicted by it is it even hate would it be tolerable there are different levels of hate and you know this is going to open up a whole different chapter and that's then you have a knowledge graph rate in certain degrees and then you have self-learning systems and improvements and so forth but i can say if i just have a last comment i mean when it comes to emotional ai and emotions and ai this is a long discussion right i mean you find workshops and conferences and i was in a lot of workshops uh quite some time ago about creativity emotions all these sort of let's say software topics so it's extremely interesting i'm more than happy to is it just thinking about it is it actually a real challenge currently like i don't feel that this is real it's very abstract at the moment it's certainly more on the on the on the l on the side where it's

not the core of sort of business systems that's for sure but i think you would be surprised it's just like who would have known that uh these pattern recognition algorithms would have made 100 million euros you know to to build ads and have this marketing industry popping up right so i i can see for example also in the corona times and the kobe times you know there is already um systems which are replicating the way how you interact so you have your virtual friend right and in there i think there is a market for that which we have underestimated uh there will be more coming in here and we will have more conversations like this because we just touched the surface last question last statement yeah the next decade of ei what do you see like where will we be in 10 years and what do we need to achieve like today to be there but yeah give us a positive vision of a height oh absolutely i think from my perspective the 10 the

next 10 years build up a smergos board of many many different different technologies that will become very valuable you look at naturally i think i still think natural language processing and machine translation and these types of virtual friends that will be a big topic you will start seeing more and more technology pushed into self-driving cars uh you know you have robots that will be introduced to the healthcare space so i see so many applications now where you have you know robots being used in the healthcare space then for example medical uh drug discovery uh with corvitas i was at the royal uh society of medicine a couple of months ago my last trip you know overseas before and there's there's a lot of discussion on how you can use ai or drug discovery and drug testing and so on i mean uh i think my in my mind the best of worlds i would see ai playing a major role in topics that matter society like for example you know and starting to seriously treat

uh diseases like cancer you know where it contributes to solving these big nuts uh the big big problems that we have in the world you know when it comes to transport and safety and so on so i i i'm extremely optimistic and i think every year will bring us more and more exciting applications so i'm i'm just this is like fantastic time to live in you know every year there will be more i'm without doubt i think this is a very good closing statement so everyone in the chat that was christian goodman and i thank you for this we will touch again for the summit latest you can see him live on the summit again he will have one or two more gigs and i thank you for being here today yeah baby i'm fantastic thanks great questions good chat wonderful talk to you again bye so um i like this because it feels for me a little bit for me to be a little bit like autumn um that was our virtual chat with

christian the next one is in two weeks and it will be with auto yawn and otto jonas the former you call can call him chief innovation office at the bundeswehr he was running the the the um cyber innovation hub and he was there for four years and he built from nothing to something so he will talk about hey how do you build in a large corporation a public corporation which is not modern which is very old school how do you bring in innovation but how do you also fail because he's not there anymore he left for cultural reasons you know where is the stop where can't you go any further with you need a different skill set also so i think it will be interesting our partner for this episode will be kai park i will introduce them a little bit more in detail but you should check them out kai park they are very supportive and i'm thankful for them for all the support especially during corona and then if you haven't checked out

we have a hybrid summit you will meet so if you want to meet christian again in person you can buy yourself a ticket to be in person if you want to have more of him online time you can get a free ticket and stream everything online and christian will join us from sweden to contribute to the good cause to make you know aix accessible to everyone and we have to apply the i day where we dig deeper into topics we touch today so i thank you for joining today it was a pleasure as usual and i hopefully see you again in two weeks and laters for the summit bye 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.