Rise of AI Conference 2023 · Berlin

Embodied AI: How will it define our future?

Elisa Czerski · Founder and CEO, N Robotics

At Rise of AI 2023 in Berlin, Elisa Czerski, founder and CEO of N Robotics, argues embodied AI, artificial intelligence integrated into a physical system like a robot, will matter more than software-only AI because it interacts directly with the world and forces developers to confront the alignment problem earlier, since a robot's behavior can be tested in the real environment rather than judged purely on model outputs. She traces the term to combining 'robot', a machine that performs work, with AI, and argues embodied systems raise distinct risks around safety, privacy and autonomous data collection that receive less regulatory attention than software AI, including questions about autonomous weapons. N Robotics builds mobile robots end-to-end, from hardware to software, for use cases like construction and energy data collection, and is launching a fully open-source bipedal robot. Czerski argues true humanoid robots are not the right shape mechanically, and that autonomy and fast production scaling, not appearance, will determine which robotics companies succeed. She describes funding as harder for a female, first-time, deep-tech founder in Germany, and, together with audience members, debates whether Asia or Europe will lead the next wave of robotics.

Key takeaways

  • Elisa Czerski argues embodied AI forces developers to confront the alignment problem earlier, since robot behavior is tested against the real world.
  • She calls for a form of moral status for robots, not because they may have feelings, but because how people treat them may shape how they treat each other.
  • N Robotics is launching a fully open-source bipedal robot rather than a humanoid, arguing humanoid shape is not mechanically efficient.
  • Czerski says autonomy and fast production scaling, not human-like appearance, will determine which robotics companies succeed.
  • She describes funding as harder to raise in Germany as a female, first-time, deep-tech founder, despite reaching steady revenue while bootstrapped.

Detailed notes

What embodied AI means and why it matters

Czerski defines embodied AI as AI integrated into a physical entity that interacts directly with its environment, distinct from software tools like a chatbot. She argues it helps address the alignment problem earlier, because behavior can be tested against the real world during development rather than only judged after training, and that this matters especially for sensitive applications like healthcare and elder care, where human-like physical interaction can help.

Ethical stakes: jobs, safety and autonomous weapons

She raises safety and transparency concerns, describing ongoing talks with TÜV, Germany's technical inspection body, about robot regulation, and says an autonomously moving entity that collects data raises questions the AI Act discussion mostly overlooks. She recounts bringing a robot to a school and finding students more interested in the philosophical question of what robots will do to society than in the technology itself, and argues for a form of moral status for robots, not because they may have feelings, but because how people treat robots, including children, may shape how they treat each other. She also flags autonomous weapons as a central long-term risk.

N Robotics: full-stack, open, and building a bipedal robot

N Robotics develops mobile robots across hardware, software, electronics and mechanics, targeting use cases like data collection in construction and energy where conditions are too expensive or dangerous for people. The company is preparing to launch a fully open-source bipedal robot, choosing two legs without a full humanoid body because, in Czerski's view, humanoid shape is not the most efficient mechanical design, and open hardware and software let universities, researchers and mid-sized companies actually build on the technology instead of only seeing it in marketing videos.

Why autonomy and production speed, not humanoid looks, will decide winners

Czerski argues the deciding factor for robotics companies is autonomy, not appearance: many robots on the market, including a Boston Dynamics unit she references, still require a human operator with a tablet, which she says provides little value. She also points to production scaling as a major bottleneck, since some robots take one to two years to deliver, and says N Robotics is built from day one to manufacture quickly, which she expects will determine which robotics companies succeed.

Funding gaps and Germany's position

She describes it as harder to raise funding in Germany as a female, first-time deep-tech founder, and says N Robotics has been self-funded and bootstrapped with ten products and steady revenue so far. She argues Germany's strength in manufacturing should make it well positioned in robotics, but sees a gap in government and private risk appetite, with many German companies treating robotics as a ten-year-out bet rather than acting now, and describes regulatory compliance, including the EU AI Act, as an added burden for a small founder.

Q&A highlights: Asia vs. Europe

Audience members raise the earlier embodied-cognition research tradition, citing Rodney Brooks, and companies like 1X in Norway, and ask whether Asia will lead the next robotics wave given cheaper hardware, such as a roughly 3,000-euro quadruped robot from Unitree versus a 130,000-euro Boston Dynamics Spot. Czerski agrees Asia may move faster short-term but argues Europe's stricter data and security regulation, if implemented early rather than as an afterthought, could become a long-term advantage, alongside Germany's manufacturing talent, provided funders take more risk on hardware.

Quotes

I'm really not too convinced about that we need human-like kind of robots, because I don't really think that's the best shape mechanically.

Elisa Czerski

No one cares about what kind of technology would you have when it's not accessible.

Elisa Czerski

Questions answered in this talk

What is embodied AI?

According to Elisa Czerski at Rise of AI 2023, embodied AI is artificial intelligence integrated into a physical entity, such as a robot, that interacts directly with its environment, as opposed to software-only AI like a chatbot.

Why did N Robotics build a bipedal robot instead of a humanoid robot?

Czerski says she is not convinced human-like robots are the best mechanical shape, so N Robotics is launching a fully open-source bipedal robot with two legs and no full humanoid body.

What determines which robotics companies succeed, according to Czerski?

Czerski argues autonomy, reliable operation without a human operator, and fast production scaling determine success, more than a robot's appearance or whether it looks humanoid.

Mentioned

embodied AI, alignment problem, autonomous weapons, open-source hardware, bipedal robot, EU AI Act

Keep exploring

More from Rise of AI Conference 2023

More on Ethics & Trust

Transcript

Auto-generated transcript, not edited. · 4,775 words

Show the full transcript

foreign delighted to welcome Elisa and the founder and CEO of n robotics who are really focusing on transforming the ways in which we collect data and using cutting-edge mobile devices to do so so please without further Ado Alicia please welcome to the stage thank you so yeah just everyone come in make yourself comfortable and I'm really really honored to stay here to be here today actually I visited the conference a few years ago as a student as a um and got a sponsorship to be able to listen All to these amazing talks and right now I'm standing here thought it's a really really cool thing for me so thank you for listening and also take the time to listen what I have to say so my topic is on body day eye and I spoke to some of the people here today and and asked them if they plan to come to this talk and everybody was like okay I don't really understand what means embodied AI so I think that's something I will start

with um embodied AI my background is AI research I combined philosophy with animatic and Neuroscience to think about what AI actually means from a technical standpoint but also from a societal standpoint so at one point I was like okay I don't want to write about it anymore to talk about it anymore but actually apply it actually do something that can help us and I think technology is the biggest possibility of actually changing something so um the first slide is okay of course okay how can robots or embodied AI can actually um help us solve the big challenges that we have like climate change Healthcare where everyone um yeah got this in the last month's environmental disasters so we get many many customers asking us okay we have a problem and and we want to train amazing models we have cool software but we don't have the hardware to collect it it's sometimes too expensive it's too dangerous for people to collect high quality um data or high volumes so that's why what we do we

put mobile robots into application and that's the application we've thought of the first it makes um sense it's it's used in different kind of verticals like construction um or energy so um what is embody that I here's a brief history and so I actually talked with one of my colleagues she's doing a PhD in economics and she was researching the effects robots have on the um workers Market at the moment and she compared it a little bit with the Industrial Revolution and when you think of it that our whole um yeah everyone everyone everything what we have currently in our economy is based on industry Revolution that people are able to be more effective that we are able to produce more um with less efforts so um yeah and so the machine was invented actually robots from the words is actually a check word for work so it's actually a machine that performs work and what combines when you combine robots with AI it's going to be embodied AI so what does it mean someone

asked me it's a is it a speaker with an AI tool also embody that I I was like no it has to intact in the real world and that's also where all the potential comes in but also the risks and when you have an entity that autonomously moves in our society that is able to collect any kind of data and not really many people are noticing it we are speaking about the AI Act but we I think that the robotic sector is not um so much spoken about yet um so yeah that's what I just just spoke uh just spoke about um why why does it matter why do we need embody there and so this leads to more better adaptable and context aware systems my city always had when I when I created this presentation it helps identify the alignment problem right so you can test your software much more early in the process and really make sure that it does what it has to do instead of just training the models more and

more and not really knowing what's what's the end result and but also when it comes to sensitive implications like healthcare and elderly care and we think like this kind of human-like interaction and possibilities can really help in this kind of areas um yeah ethical consideration that's of course one of my favorite topics um I also already mentioned Safety and Security levels and that's something we for example are in constant um talk with the turf the regulating organization here in Germany to talk about what kind of regulations can we think of when it comes comes to robots but also transparent explainability like why does the robot water does and it scares many people like our office is in a pretty public space so everyone can can come by and come by and when people go by who are like workers and they are really afraid that it takes their jobs and when I talk to CEOs of big companies they say okay of course the half of the people we have in production right now we're

not going to have them in five or ten years so we now have to make sure that the development of robots and robotic systems and applications is inclusive and also takes care of these kinds of people um yeah to also design to maximize the benefits of course everything one everyone thinks of um robots in the army or in war and of course autonomous weapon systems are really really scary so we all have to make sure that this doesn't happen but also because considering the long-term societal impact actually was um bringing one of our robots to school and showing them what it can do and in the end I was thinking that everyone is interested in robots but actually all the students have been interested in philosophical questions what will it make with our society to have robots and that's something I what really fascinated about that the Next Generation cares about that so much and then the business world right now I don't really see that kind of interest and I think that we have

to have some kind of mobile status for for robots not because they may have feelings or may have Consciousness no but because it will change the way we interact with each other when for example be a violent against robots how how do we then treat people around us especially when it comes to education and children um so where we'll be and um actually uh at the speaker's talk yesterday was like okay you know where we will be in 10 20 years of course I don't um but we clearly see see a way here so AI will be even bigger and I totally agree with this whole AI Act um consideration that we have to regulate it but I believe we have to regulate it in size um and of course what kind of companies are there doing that and but it was it will be a more General versatile system but in the long term we're gonna see what we see right now with neuralink and the combination of humans Ai and robots in the

long term and everyone has to think about for personally where they want to stand and if they want to be part of their Journey or they want to be part of the regulation for example so um yeah that's also the the transforming the way we live and work and that's actually a short term impact and that's why I actually got into robotics because for me it's actually an enabler for a self um returning life you can for example I have a cleaning robot at home I have a morning robot at home it really enables me to have a life whatever I want to have without sacrificing for example a clean home and it's so weird but for me it really means a lot having this kind of machines also washing machine a computer everything that enables us now we have to think what will enable our children and what kind of products this will be and yeah so we made sure to to have a European female-led company to try to transform this industry long

term so what we can do um to ensure everything goes well I think the the thing I'm talking about the mouth is investing also for governmental perspective but also from a private perspective as a female founder it can be sometimes difficult to raise money that's just the truth when you call go to the statistics um but I think we have to make sure that we keep Innovation here in Germany and also help creating Innovation here in Germany and I think we're still lacking that so I'm really happy to read so many different stakeholders here and challenging them and also together see what we can do to ensure that and legal framework of course and now a little bit to my company so what we do and we are a full stack robotics mobile robotics company um we are creating developing and Manufacturing mobile robots from a hardware software Electronics mechanics standpoints and ensuring that we are um have the power of the full ecosystem from the tiniest component to the whole system solution in the

end really ensures that we know what the customer is doing with it that we know how we approach approaches this kind of technology and um yeah we can also ensure application I kind of compare that with AI 10 years ago so everyone was like oh cool there's now now this technology out there what we should do with it I I want to try to use it but I don't see a clear application my company and that's a little bit the same thing with robots we get calls from different kind of companies saying okay we got a robot it makes sense for marketing we've got great pictures but it doesn't really solve a problem but at first you have to make sure that the customer understands what is the problem problem for example when it comes to data collection when it comes to ensuring the safety of workers doing difficult over your dangerous tasks as well as yeah just to make sure that they really understand what they want to have as well as um being

able to access it and that's also a different point right we the robotics Market is pretty um yeah they're not that many out there um I think pretty big ones in the last who have been around the last decades there's not been Innovation around and there's so much space for companies going in but also bigger companies investing in robotics and AI to ensure they're ahead of time and they're not missing the next big Trend that's coming up here for us and I think especially as I said the combination of AI and Robotics really will be the next Trend so that's a little bit of the robots I blurred out the one it's going to be launching in uh two months our biped a robot and it's got us quite excited because everyone's talking about humanoid robots and what you can do with them and that's still the application problem so right now our vision to really go into to different applications to then see what does the world really need and what is this nice

to have for example the German Red Cross approaches and said okay it would be really great to have something for disaster control when for example earthquakes happens we can search for for survivors and that's really something close to my heart um enabling these kinds of things through technology we have a crowd-paid robot and what's what's the biggest thing about our company is that our software and Hardware is completely open and I think that's part of our new generation saying okay technology no one cares about what kind of Technology would you have when it's not accessible and not not accessible to our middle stand companies it's not accessible to our universities researched everyone working with it and you can only um yeah put really used to robotics when we ensure accessibility um yeah our values responsive entrepreneurship diversity and and our kind of bubble it's normal but when I talk to Big corporates that's not the normality and I'm I think that's that's quite sad but um yeah I think it's great that we're changing that

right now um I'm not have it full time yet but I'm really curious about any discussion questions um also critique feedback um I'm happy to yeah hear anything from you thank you thank you Elisa and I think we have our very first question here at the gentleman and stripy shirt hey thank you it's it's great to see the topic at the conference because I think this is certainly one of the big big areas which has been around for decades you know and I was just um interested maybe two things like one is there's so much research done Rodney Brooks and and these guys right in the philosophical question of relay I cannot actually merge without having a body but these were these would have been great topics to also bring up I think uh because I think it will it's one of the big things that we will see in the next five to ten years right so um would be interesting to hear more about it if you have insights in or like your

opinion about that because I yeah you know as you may remember or know in the 80s 90s was a huge topic right with subsumption architectures and so on um and the second one is of course there's a lot of companies that are using robotics already right and real AI particularly in the healthcare sector like I'm thinking of One X in Norway or Catalia Health in California with Corey Kidd and for ahead of Robotics and Sweden so a lot of beautiful companies that are using it right so I think it's one of those major puzzle pieces uh that in the next maybe five to ten years will bring another wave of change and um maybe last comment you made these projections of maybe 200 years but I I'm actually a bit more I'm thinking we can have these types of robots when I now see One X for example in Norway these robots look very close to you know something quite tangible and you you now enter chechi BT and these new technologies into it it

will be interesting to see how they evolve in the next five to ten years actually so I was just now with a couple of points you yeah you want to comment on those yeah thank you so much um I really appreciate any um input so um yeah maybe I will I will start with the last question um yeah there are um many robotics companies out there and what we have seen is a real problem um I think what's crucial about our product is autonomy it doesn't for example we got a spot here today I think most of you saw it from a different kind of company and it requires someone a colleague of mine having a tablet um to to actually um yeah say what it has to do and and I think that doesn't really bring any value and it's the same thing with drones we really have to ensure that the um that the system is safe enough reliable enough powerful enough but also careful enough when it sees for example people or

animals or whatever to operate autonomous and that's why the customers come to us they say okay there are robots out there but they do not work in the web point one they don't work in the way we wanted to work reliable for example and the second thing is also um yeah we we don't want we wanted to have a system that's complicated for example the the Boston mix what we brought today it's really happy to carry around you have to um to change the whole body to change the to change the battery and having this power over the ecosystem to really say okay we're gonna have a battery system that's um going to be changed autonomously by the robot and I think that's going to make the difference and that's gonna long term determine what kind of Roads company going to be successful and who not and the third thing is I think also when we talked about other companies I think like for example Elon Musk do really well is production scaling production and

hardware and that's a really big Challenge and that's something we approach from day one to ensure that everything is can be produced really really quickly and right now when you try to buy a robot one of those you you said they're either not available or you can wait one or two years um and that's not how you innovate and my it's my opinion and but of course all of these are really really great robots and and it's a different kind of saying okay we want to have humanoid robots or I'm really not too convinced about that we need human-like kind of robots because I don't really think that's the best shape mechanically um so we're starting with biped robots so in it's it's going to be like two legs without a body actually and it's making the same job but it's much more easy and technically and also um yeah from a from the price standpoint of course we kind of have still the best robots cost like 500k or whatever but then we know

what can buy it what does it make any sense so and the other thing um the other question yeah there's a really really great research out there of course like in the 80s I wasn't born yet um but yeah we're working with different kind of Universities at the moment and the most of them say we have great phds we have great software but we have no Hardware to actually um test what we have done and that's so sad um so we really try to keep our software and Hardware so open and everyone cannot work with it even our bipedal robot could be will be complete open source to give back to the community because these are the kinds of people that were actually realize all these Visions we have today is not going to be companies probably but the community of really smart people and but they don't it's like the Raspberry Pi right it was a revolution that everyone now could code on Hardware that's accessible and I think that same kind of resolution

has to happen in robotics and not like fancy kind of robots that are in marketing videos do you think we could lead uh we could lead in this area in in Germany like globally yeah I I hope so that's like of course um our plan to build a really cool um good company and we have customers and we have Revenue we have a really great team that some of the people are here today already we're really passionate working about it but I think the environment is quite tough building that here and for example for me as a first time founder dealing with all the texts with um working safety it's and now with EU act AI Act it's really tough and not no one's really helping you but only throwing stones in front of you all the time and I still believe I wanted to build here because I I was born here I lived here I want to give back to the society that helped me to raise me kind of it's kind of

my fault but um yeah we're gonna see I heard that some kind of companies taking their IP outside of Europe to ensure that's better handable for that them the compliance part maybe we have to do that in the future but um right now I'm still really hoping for funding in Germany we're uh with with many VCS and and talks right now but especially like from the governmental side I see a big gap in funding in a way men present females present and the way deep tech companies are perceived when you're not the B2B SAS model that's normally funded and I think that's sad that people don't really want to take risks and I think Germany has to want to take risks and that we can succeed and be a leader in that area is interesting because I was sitting there thinking about why has robotics not had that should we say speed of adoption as we've seen with AI in the first and second and third wave and I wonder whether it's because this construct

of buy or build as a corporation if you think about wow I believe Germany should be the people that can build it we have some of the best engineering mechanical talent in the world and so arguably the hardware should be actually relatively simple for us to produce and so I wonder whether it's because of this mindset that actually is it too expensive to bother to build or is it that actually the buy-in solution isn't quite there yet like where's that missing Catalyst that we need in the robotics sector to get it to be used by the corporates because the use cases are evident but the actual manifestation of that hasn't quite tipped and maybe what is that Tipping Point that we're looking for it's quite interesting I spoke to one like German middle stand company who's also investing in kind of innovative companies they call it and I was like yeah it's really cool and they were like yeah Roberts we probably believe it's going to be relevant in 10 years and you're like okay

that's people told already also about Ai and then it was there and everyone's talking about so I think um some of this kind of first movers are missing saying okay we are taking the risk um but also everyone's saying Hardware is hard um I think the perception and that is it's really difficult and really pricey it's still there even though I don't understand how big software companies can spend a lot of money on the same product all over again and I think we we got um yeah self-funded bootstrap so far we've got 10 different products um a good Revenue Stream So for us it wasn't hard but yeah I think more people just have to let go of that perception that's refreshing to hear are there any other questions or discussion items yes please we're clearly an Enthusiast yeah I mean the um I'm just thinking I mean my prediction would be I'm happy to hear your comment that the robotics AI component in these waves of AI that will come in is likely going

to come from Asia Japan maybe maybe China and I believe one of the reasons is that there is a less of a distance like there is in Europe in general in Germany in particular a bit of a fear because that's a manifestation of an AI that you can actually see whether it's a quadruplet robot or a bipedal robot and essentially this is like the manifestation of what some people fear already today right so that makes it overall hard but otherwise if he were to overcome that and possibly also regulation I agree with you this is one of the strength that Germany and and the Europe has which is manufacturing right we manufacture a lot of cars and and and trucks and you know whose Quantum has chainsaws you name it right so we're good at that we should be able to build that um and I absolutely convinced that this will be one of at least four or five big waves just as we see now and it's just amazing that we basically just focus

on the heat now which is chat gbt the whole world is very excited but I give it another few years um and then we have things like distributed AI which is an area that I worked in for a long time embodied AI uh and you know causal AI which is another big area so all these things will come and be basically right now even at this conference if I may say that it's a very low focus and possible awareness of these areas and I guarantee they will come in the next over the next 10 years so maybe a further comment on that but I the guess as to where in the world it will come from is an interesting one I think yeah Asia of course like um you can it's I think like just from a price point perspective you can buy a boss Dynamic spot for 130k at the moment um a quite a bit robot from Asia from unitary is I think a 3 000 Euros it's like it that's like we

have been shocked uh we have been quite happy that we haven't put a quadruped robot out there for like um as a business case because we actually saw that coming and of course you saw that scary videos that you had this kind of robots going around in Corona time starting at people you're not allowed to be outside right it's it's also um it's scary but also very interesting at the same time but of course um I think especially in robotics and also Ai and the security is really important um like the security data and and also um yeah the whole regulation and there's like China of course not not the company or the the the um yeah the area to go and that's a really great benefit of us producing here in Europe and especially Germany and to be able to address these concerns in a really early level and Implement them in our development what I think will be long-term a benefit short term of course they're gonna with everything there have a higher

speed of developing things putting them into the market but I think long term we can still succeed enough of course Japan is um really strong with robots a long time that had robot dogs when I was little I saw that in television so um yeah I think the society and how the perception of robots and the openness to Innovation is really really different and I think when we a little bit adapt to that in Germany we have also a big chance as you said there's a lot of talent here and you should be able to do that but as you said like an AI conference there's a little little representation of these kind of topics I'm really happy to really read here but of course I'm now asked to speak about AI what I really wanted to do 10 years ago where no one cares about that and everyone wants now to me speak about Ai and I'm like yeah I can do that but I really want to speak about robots I'm somehow not

always on the trend um yeah but I'm trying to just sensibilize the people for this kind of connection and as you said there are many kind of Trends in this area and to be really um to be prepared for that and also be part of that because um WT is now really interesting but maybe next year was going to be next year fantastic well that's brought us straight on time so thank you very much for the wonderful discussion Elisa thank you very much thank you so that wraps up our session

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.