welcome to ai decoded. we are going to take you on the road this week. a few weeks ago, stephanie hare said to me after one of our programmes that we focus a lot of our attention on development here in the uk and in the us, but what about artificial intelligence in germany, the powerhouse of european manufacturing? what are they doing on al? so, she has been to find out. what did you discover? so many things that i cannot wait to show our audience. without people in government, talked to doctors, scientists, researchers, artists. we have a beautiful show ahead it's really going to take you behind the scenes of what is going on in germany with al.— in germany with al. let's take a look.

music. grey skies over berlin, a fitting backdrop for germany as a whole. only two weeks ago, its coalition government collapsed. its economy is stuttering. core industries like car manufacturing are under threat from competition abroad. could artificial intelligence get europe's most powerful economy back on track? willkommen, bienvenue, welcome to the first episode of artificial intelligence decoded, set in berlin, germany. it's the 35th anniversary of the fall of the berlin wall, and we're here at the falling walls science summit, looking at a gathering of some of the world's greatest experts on artificial intelligence. the summit has attracted europe's brightest minds in science and technology. founder sebastian turner

hopes they will build bridges and break barriers. this weekend...actually, november 9th is the day the berlin wall came down. and this historic date raised the question in particular for those who never expected the berlin wall to come down which are the next walls to come down? one of those people breaking barriers is richard socher, whose pioneering research into prompt engineering led to the explosion of generative ai. if you could make germany one of the best countries

in the world to do ai, what needs to change? yeah, i've actually been asked that exact question by the german chancellor and the previous chancellor. you have to start early on in high school. after english, which is hard to avoid at this point, you got to know how to speak english. after that, the second foreign language you should study should be python. because if you don't understand programming, it's very hard to participate constructively and actively in the future. ai is also ultimately a programme. second, it's very hard and painfulto fail, period. even in silicon valley, it sucks if your company doesn't work, right? it's very, very stressful. but at least in silicon valley, if you do fail, a lot of people say, "well, you tried something really ambitious, and we're 0k helping you succeed and try to succeed another time. " it's culturally very painful and legally very, very painful to fail in germany. then there'sjust not enough

of a vc ecosystem venture of a vc ecosystem, venture capital ecosystem in germany. it's a little bit better in the uk, but also like overall in europe compared tojust silicon valley, it's very underdeveloped. then there's a "not invented here" mentality in large german companies where they say, "oh, this little start—up, we could build this ourselves. and of course, in theory, google, amazon and facebook meta could build any start—up, right? like, but in practice, even those big us companies are buying a lot of smaller companies, and it helps them and helps them to innovate faster. it helps them to get go to market faster. it helps them to bring young new ideas into the company and overall be more competitive in the market. and so that also isn't very strong in europe. the bureaucracy

of starting companies is also just too much. i think it needs to be continuously reduced. as you're a small company, i think there needs to be less regulation. and then once you're really, like...you found your product market fit and you could become profitable, then you can start to regulate maybe more. so while germany may need to make some changes to make the most of ai, it can also play to its strengths. i asked 0nder celik, a berlin—based researcher, what he thinks is the most promising sector. it's definitely going to be | health care, but germany still has a very strong hold on health care systems. i so like all these big research institutes like charite, - and they are investing a lot in r&d in al systems. - so i'm very lucky to be speaking to you today from the heart centre of the charite hospital in this unit. we have some patients who are in a very serious and unstable condition, and the doctors are using a new form of technology based on artificial intelligence to monitor the two biggest risks that can happen after surgery. we're going to go behind the scenes and talk with a clinician, one of the heart doctors here,

who's using artificial intelligence to monitor those risks and try to catch them early so we can maximise the benefits of patient care. we are using ai as an additional surveillance tool to monitor our patients for two very common and significant postoperative complications. if we identify these complications early, that we are able to intervene and then maybe change the course of the disease so that we can have our patients in the best possible way. and i see you've got a big machine behind you. can we go over and take a look? yes, sure. of course. after you. 0k, thank you. so what you can see here is the user interface. and here we see the raw input data. in this case, a vital value such as heart rate or systolic blood pressure or the temperature of the patient. here we see some selected laboratory values. and of course we also see the trends and the dynamic change for these values. right.

and this is all live, real—time data. yes. and now the probability for post—operative bleeding is estimated to be i%, so it's really low. and this also makes sense if we examine the patient. and we can also see here the trend over the last two hours. so we were starting at 25%, which is... getting better. yes. this is the view we get for postoperative renal failure. and what we can see here is that we were starting at 7a%, which is reasonably high. and then we have a declining trend, which is reassuring for us, but the risk is still estimated to be 25%. what we're doing is with this algorithm, we try to identify those patients who are at a very high risk of these complications, but we also try to identify those patients who are very unlikely to experience those complications. and have you any data on how accurate the algorithm is compared to human doctors? there was a big scientific

study that showed that these algorithms performed better than regular doctors. so who invented this ai tool? meet professor alexander meyer.

to be able to work with data is like a superpower. so we're here in berlin, walking along the east side gallery. this is one of the remnants of the berlin wall. we are witnessing a very exciting time in the ai space. we came from image processing using convolutional neural networks, which is a special type of neural network or artificial intelligence to catch patterns within images or videos. and we all know this and use it daily when we use our smartphone, or we also use it in the hospital for radiology. and now we just witnessed two years ago the advent

of generative ai. so now we see ai is producing something meaningful. and we are just starting to to use this in the clinical work. ai will augment medicine and will increase the operational efficiency and will increase patient safety and will enable better outcomes. and in the end, there will always be physicians. ai is an instrument such as the ct is an instrument. so i will enable us to have better, better precision in our decisions and to be more more focussed on prevention and the correct therapies. health care is a strength for germany, but what about the rest of its economy? we asked the new minister for education and research. the car was invented in germany - the bike was invented in germany. _ many more products. but we're losing ground because when you're i economically successful, you get arrogant. - you think, you know, - nothing can happen to us. and we see now how vulnerable we are. l so therefore, it is extremely. important that we have a more science—friendly environment, that we have a more - business—friendly environment, that innovations take place - in this country. many start—ups had their starting point here, - but when it comes to money to grow, we're not, - unfortunately, a very| risk—friendly country. and that has to change. and then children need to have access to technology, - need to have access to an ipad, need to know something - about artificial intelligence. they need to know something about the algorithms. -

about where to get the news from, how to critically read who is behind that, - what is the intention, where is the source? | and if you don't know this, you haven't finished - school, in my opinion. you had some extraordinary access. people always asked me how does this relate to our everyday lives and when you look at the filming in the hospital, that's where is really making a difference. but here's a thing. we have talked about it as a competition and in fact tony blair, the former prime minister, said we need to get going because as a global competition and we need to be first. how do we compare from what you have seen because he said to me initially i don't know what germany is doing but how do you think having been there now we compare to our biggest european competitor? i think it's a cultural question. some countries want to be first. some countries want to be the best and i think germany probably falls into the latter category. trust is really important in germany, so they don't want to roll something out particularly in a hospital setting unless it is trustworthy. so it's not about taking a cautious approach what

about a trust first... i5 taking a cautious approach what about a trust first. . ._ about a trust first... is also about a trust first... is also about feeling _ about a trust first... is also about feeling a _ about a trust first... is also about feeling a failure - about a trust first... is also about feeling a failure like l about feeling a failure like what your contributors said. i think in the vc investment ecosystem and culturally, yes, there is that and that is probably linked with will imprisoning outcome i wanted to be the gold standard. it’s be the gold standard. it's amazing _ be the gold standard. it's amazing how _ be the gold standard. it's amazing how culture actually applies to what we're doing here in the futuristic environment and culture really does play a big part of it. a short break and the other side of it we will talk about how the germans are applying ai of it we will talk about how the germans are applying al to the germans are applying al to the visual arts. stay with us.

so often we hear that al isjust "statistics on steriods". in other words, it'sjust really intense mathematics. so how or why are artists using it?

how are the germans applying it to music or art or visual things?— to music or art or visual thins? �* ,. , things? berlin is always been this made — things? berlin is always been this made it _ things? berlin is always been this made it for _ things? berlin is always been this made it for creative - things? berlin is always been this made it for creative so i this made it for creative so what i loved about this and what i loved about this and what you are about to seep is people are using ai is a tool, as a collaborator and it raises the question of when does ai become an artist? so let's take a look. so we're here in berlin, walking along the east side gallery. this is one of the remnants of the berlin wall. we just came from the falling walls science summit, where we were looking at how researchers and scientists are using al to break down walls in the scientific process. now we're going to go and visit two berlin—based creatives, matt dryhurst and phil langer, to see how they're using ai to break down the walls of their creative process. music.

generative ai is transforming images and video, creating a new art of the possible. for some, these images provoke curiosity and wonder, but for others they can feel unsettling and weird. phil langer is a creative working at the cutting edge of what ai can achieve visually. so talk us through how you would at least start. so i would need, in the beginning, i would need like several images of that person. the more images, the better basically. ideally also, like, in different light and different angles. put this in this... i'm using midjourney here, for example. and then you have these this prompt that kind of pushes of course the ai in a direction that you want. so a prompt is like a phrase or a set of instructions.

exactly. so the prompt kind of tells the ai what what you want to have the recipe for, basically the result. 0k. and it is also a little bit might be like also the recipe of like the creatives of the future, i guess. because of course it's very easy to recreate these images somewhat, but if you want to be very specific, you need also very specific prompts. it's the art of the prompt. the art of prompt might be the future. exactly. and i would just say i kind of describe what we want to see. i want like a hybrid image of an elegant woman and a cat in front of a grey background. and i would adjust it to like, brunette hair and like, brown eyes. and i would just adjust certain parts of the same prompt for different animals. ok, so you're already telling it. this is what i want the output to look like. exactly. yeah.

and then it's almost like a child that you're having to correct it. it'll give you versions of it, and you'll be like more this, less this. getting closer. n ot exa ctly. too far. you have to kind of steer these results with i said before, with curating, you get like somewhat random results. and then some of them are more human—like, some are more like animal—like. and you kind of steer the direction you say, like, let's continue with this image. and you can also adjust the prompt if you see something doesn't work quite right with the eyes or with the hair, you adjust the prompt in a way that it works better. do people feel emotion, do you think, when they look at an ai—generated image? i think it's a mix because they are very beautiful images and they're nice to look at, but they have this uncanniness. you kind of don't know, is this real or not? and i think this mix of like beautiful image to, like, kind of like disturbing images when they come together. it's kind of like an impactful image and that makes them kind of complicated, but also like impactful. i think young people also

now are like children, not children, but people of the younger generation. i think they're already questioning the reality of these images that are out there of this content. right. i think it goes like even in politics or culture or whatever it is, i think our generation is different in that respect that we we always assume stuff is real. like you said, we grew up in an era where we saw a photo or video, we assumed it would be undoctored or in the news. but i think nowadays young people are always kind of thinking this might be already generated or manipulated in a way, and that's going to be even more in the future. every day it feels like a new question i don't have the answer to. but maybe life isn't about having answers. maybe it'sjust about being curious enough. to keep asking. i always thought by now life would make more sense. i do you see there being any danger of losing the skills and the training that people like you had when you were coming up in your career? if somebody can just leapfrog all of that and go straight into using these. yeah, yeah. i think it's just important to work with these tools. i think it's they're kind of like when you have creative thoughts. the time is now that

between the creative thought and executing this idea is, like, gone to zero now with these tools. and that's kind of what makes us so powerful. like you don't have to have a long skill set or long experience with very complicated software tools. you can just basically technicallyjust type it in and then you have a result that you want. but because everybody can do this now, i think there's going to be a flood of content out there of ai—generated creatives. right? so i think to stand out creativity is even more important than it was before. so are you an optimist or a pessimist when it comes to ai? i don't know. i'm neither. i'm a hybrid. you're a creative. yeah. either way, you're going to have fun with it. yeah, i guess. hello. hey. matt dryhurst and holly herndon are stars of the art world. their new show at the serpentine gallery in london brings together choirs from across the united kingdom to create a groundbreaking art form, a meeting of humans and ai through music. we're visiting their berlin studio to find out how it works. the actual sound of al to us is the sound of fans. our studio often sounds

like it's about to take offjust cos all the cooling necessary. and so actually, someone, andrew in the studio came up with the idea of figuring out how to alter the fans. so the speed that the fans turn in order to have them play music acoustically. jury—rigged a bunch of smallerfans. exactly. now you're going to be able to show us how you got them to make musical notes. exactly. and for some reason, only two of them are working right now. but i can play... exactly. we wrote a song book explicitly for al training, the idea being that if you sing the book from start to finish, you will have covered the full range of the english phonetic language. in order to be able to contribute your voice to an ai model. so the song book itself

was inspired by old, kind of like renaissance music and sacred heart music from the american south, which actually has this kind of diasporic relationship with english folk music. we took that book throughout the uk, recorded 15 choirs who volunteered their voices toward a large dataset that we then use to train the ai model. and so when you go to the show, you'll be able to see every stage of the creation of this model, hear its inputs and its outputs, and some of the music that led to the songbook itself. and so the 15 choirs own their voices. they own their recordings. we actually had to negotiate with them to use them for our show, which is consistent with the principle. yeah, and so then they basically have a protocol that's put in place for them to be able to expand the trust and maybe value that data. and the reason we did it is because there's not really many examples right now of people proactively finding ways for people to co—own and govern their data. there's this big open question of, like, what to do with people scraping the internet, what to do with your likeness in these ai models. and so we're like, ok, this is an opportunity for us to set some precedent of saying you can actually do this slightly differently.

i will still desire to go into a room and hear someone sing beautifully. nne of this changes any of that. and if someone wants to reject all these tools and just sing beautifully, then i love that they will do that, you know? so why are these two artists based in berlin and not silicon valley? i think that silicon valley generally is its best pr engine. a lot of this work is happening there, and a lot of the work is not. for instance, on sound—based works, a lot of the researchers we've worked with have been like in barcelona, some of the best researchers for sound are based in europe. when it comes to the image model space, what characterises ai image models is they were pretty much all built in germany on datasets curated in germany. if you are training kind of like world—class scientists or remarkable artists or whoever it might mean, and ultimately they have to leave to the united states in order to realise their ambitions, that's a problem. you want clarity in order to be

able to assume risk, right? because you can go, and i've known of people who do this and they hire 50 people, and then they have to shut shop because it turns out that they were out of bounds on certain on certain things. so regulation is good. there will always be a desire for somebody who's really good at telling a story or who's in person with people, right? i mean, actually, the value of this source of truth or whatever is becomes even more important, arguably, in a time where it's really difficult to discern the provenance of information because there's a bot somewhere in the middle. you know, gathering different people to co—ordinate to accomplish something, because fundamentally, particularly when you're dealing with large ai models, a flux of chatgpt, you're looking at kind of the aggregate input of all of us together like it's us, in aggregate, coordinated by algebra. so what's holding people back from really engaging with artificial intelligence? it's become somewhat fashionable in more urbane kind of circles to show incuriosity. which is really odd.

that's like new for me coming from the arts. but the last five, ten years, i think there's been a lot of backlash about tech. a lot of kind of like caricatures, some of which are qualified, but like a lot of caricatures of tech. and i think that that's the only thing i worry about, is that people might be encouraged to be incurious about this stuff. they're turned off rather than turned on. yeah, exactly. a social stigma starts to be generated around it, right? like the tech bros and this kind of language. ultimately, ijust don't think it helps anybody. not to say that people who are, you know...there who are, you know...there are many people who are critical of tech who i think are very well qualified in their in their criticisms. but it's more just you don't want to see a curiosity gap emerge. i'm so glad that you went. i've learned so much tonight but most of all that you look great as a cat. in terms ofjust taking two elements of that. one is feel, and who you are creating cameras with. he says

he is a creative, so it's not just putting a code and it produces it for you. it seems that he is working with that and it evolves.— that he is working with that and it evolves. yeah, there is and it evolves. yeah, there is an iterative _ and it evolves. yeah, there is an iterative process _ and it evolves. yeah, there is an iterative process because l an iterative process because matt and holly's art programme is called the call, like prompt engineering. you type it, write a call and the machine responds just like in call and response choirs. forthe just like in call and response choirs. for the parallels between human activity and human machine activity are clear. ., human machine activity are clear, ., ., , human machine activity are clear. ., , ., clear. so long as people engage because we _ clear. so long as people engage because we have _ clear. so long as people engage because we have got _ clear. so long as people engage because we have got now- clear. so long as people engage because we have got now in - clear. so long as people engage because we have got now in the| because we have got now in the visual art or the people who really are engaging people who are incurious or turned off by it. i are incurious or turned off by it. ., �* ~ ., are incurious or turned off by it. i don't even know if that's limited to — it. i don't even know if that's limited to the _ it. i don't even know if that's limited to the art _ it. i don't even know if that's limited to the art think - it. i don't even know if that'sj limited to the art think that's kind of kind of all of his looking just at silicon valley or the us— china tech rate. we look let all countries in the world and all of how they are engaging with al and that's with the curiosity gap can be closed but he's doing to me different with her in it so inspiring. different with her in it so

inapiring-_ different with her in it so ins-cirin.~ . ., inspiring. what about your curiosity — inspiring. what about your curiosity because - inspiring. what about your curiosity because you - inspiring. what about your| curiosity because you were desperate to go to germany and now you've come back, your views? �* , ., , views? i've never been as inspired _ views? i've never been as inspired by _ views? i've never been as inspired by uses - views? i've never been as inspired by uses of- views? i've never been as inspired by uses of ais i l views? i've never been as - inspired by uses of ais i have been seeing it used in germany so that you to all of our friends in germany for that. it made me want to go into this for other country so look out, france, and everybody else, we are coming for you.— are coming for you. taking the camera. looking _ are coming for you. taking the camera. looking forward - are coming for you. taking the camera. looking forward to . camera. looking forward to that. that's it for this week. our thanks to stephanie. justa reminder, if you are enjoying the ai decoded series, then all the previous editions are on our youtube site, so do have a look at that. let's do it again same time next week. thanks for watching.

hello, i'm christian fraser. you're watching the context on bbc news. translation: today there was a new russian missile. j it is obvious that putin is using ukraine as a testing ground. translation: combat testing | for the oreshnik missile system is being conducted by us in response to aggressive actions of nato countries against russia. he said that russia believes it has the right to use its weapons against military installations of those countries that allow ukraine to use their weapons against russian facilities. the uk's support for ukraine is always for self—defence. it is proportionate, coordinated and agile. joining me tonight are former british army officer and military analystjustin crump and ian bremmer — political scientist and author and the founder and president of the eurasia group.