ah, thanks much for watching. i'm clear. richardson in berlin from the in the team. we really appreciate you taking with ah, will you become a criminal? mm. franklin mayo already know with hackers,

paralyzing the tire societies computers than elsewhere. you and governments that go crazy for your data. we explain how these technologies work, how they can go in for, and that's how they can also go terribly. watch it. now on youtube, you enter a bit of text and artificial intelligence. software will paint any picture you like . this idea has gone viral and a i. art has even want to waltz. ah, text to image, eyes creative or simply good at copying. and what does it mean for artists? our topic on ship. ah.

a sat koala reading, a book, a hologram of grandma, or the most beautiful place on earth. text to image generators like mid journey dolly and stable, diffuse and create every image you can think of super fast. all they need is a few key words. these a eyes are trained with a huge amount of online images and use these as templates for their own creations. how can you use image eyes and how creative are the results? this is how the end of the world looks, based on an ai muddles interpretation. turning text into images is all the rage right now. i. so how does this new ai artwork? well, it generates photo, realistic or painting like images from a small text prompt. it all started which dolly created by open a, i open a i fet, it's machine learning model with tons of images screen from the internet. the user

types in the text prompt and out pops a unique image. very good at doing this sort of things. you might get a graphic designer to do and the guy good at doing some fun things. the things that put together, you know, a koala on top of about all things that you might have never seen before. let's try that out using another ai arch model. mid journey. we type in or prompt and out. fairs or koala on a boat. you can also check it up a bit. koala on a boat, under a starry sky. so i could ask for a cat in the style of the gas and i get it right cubic like cat or i could ask for a cat in the style of an golf and i might get a very starry night like cat. we really don't know what's going to work about isn't gonna work, said isn't a lot of playful discovery going on. there's no way of knowing precisely had a machine like muddles will determine the makeup of each image, typing your prompt

a 1000 times, and you'll get a 1000 different outputs. ah, to use me journey, you need a discord account to use tally. you need to sign up at its mother company, open a i a for people who just want to have a quick bit of fun. there's crown, the model is less advanced, but with the right prompts, you can get some interesting results. is this art not really if you ask me, in order for the e, i to create good images, you'll have to experiment with it for quite some time. it's called prompt engineering. but once you've develop a sense of how the i works, the results can be really impressive. you could create a whole film universe without ever having studied op, like fabi unsteady. ah,

the images for this scene, but generated by artificial intelligence, no active sense woocommerce. hello. this is in quantico quantum leap that opens up new possibilities in many different areas in field field . but i've been long 5 young states from berlin has produced an entire film well with image a eyes like mid johnny. he came up with the idea by chance. kept his bic is in funnies. i saw this picture of mine that pulled me into another world. so i kept prompting until i had several images of that world, which i used to make a movie trailer, dallas, mac, and trail often. fin salt tells the story of space travelers you discover a strange planet covered in a mysterious salt layer. it's just that the molecules are inverted. what the heck, do you mean by averted? there's no profit spent is fabia wants to create a scene, say,

equality technology salesman, combust. he just jumps rice in many asked us that this item of the 1st step would be to explore the surroundings of this hardware, sharp in mid journey or stable diffusion on. i do this by writing promise, want to see what the a i generates on the ma fiber. this process basically means that a get started by putting together the beginning of the story without knowing what the ending will be like that near the right. now i'm imagining there's a device that keeps bleeping and blinking, that may be, it wakes pop up from an afternoon nap wash, and he swears at crumply, after that, i don't know wrecked floats on this, him the stuff i think it was in the near. when the story is done, hobby, i'm starts voicing what, what, what the hell was that? and another angel changes his voice where there was that hope that most of it. finally, the scene is finished off with synthesizer sounds. the completed scenes, so called story seeds are then posted on twitter. after each episode,

the audience can bode on what will happen next. this way, fabulous doses newly created universe is explored collectively. that is definitely a. salt is an experiment to test, to possibilities and limits of imagery. eyes honest, that is glover does, is in the can. i think a lot of people in the community who have better ideas for this film universe than me for him when he was. and that's what i find really cool about these new technological possibilities who has finished and isn't and isn't taking lucas david's hub. at some point javion states, i would want his followers to create their own episodes using ai and the my was eames ish, really creating film content is usually really difficult and sort of a thing as new technologies that is do it in a fairly easy way. and, and it also bridges the gap between creator, in this case, me and the community, or audience record

real life into actual thanks to artificial intelligence. exciting, fabulous. as a uses a i to create still images. he then animates them using other programs. soon. this process will become even easier takes to video generators are already being developed by google and meta, for instance. an example from a you, as art fair shows, just how well image a eyes work to day. in the category digital art, the 1st prize was awarded to this image created by the mid journey i. but what do artists make of the new competition? i now i has been using computer programs for his art, for 20 years. the artist from just sort of these, the machine as part of the creative process. but like many other artists, he has mixed feelings about new image. a i is awful. do you have these moments to

where my 1st reaction is called wow. what's this one is about yet, but then you reach a point where you think, okay, what kind of quality does this new technology have? and how can i use it to him and the moments. it's hard to say where this journey will end it as a hidden gay from my perspective. this technology is very limited with because i don't see it as creative in its current state, aster east. as nice as could be our teeth. the eyes can only create images with the help of humans. the input comes from the prompts, but whether they're creative or not, the results are often indistinguishable from human knowledge. and that could cause artists that job with. these magazine covers, for example, were created by image a eyes still, diana eyes does not believe that a eyes will replace artist as they lack creative impulse. ah, if they don't listen to your father says i do see a danger in

a i is replacing artist temporarily earned by art. would all eventually end up looking the same. that thought was, and denise think, another threat and we might not have any human illustrators left in the next generation. so we need to keep this profession an expertise alive. i thought in this in their pocket, if toby was professor of ai at the university of south wales in sydney also thinks we need human authors. i don't think it's going to take away the job to vote is i don't think it's connected in any way. replace out is just like photography did remove painting what these programs produce may be useful for illustrating articles like, but they're not going to speak to us. the way the great out speaks was they're not going to come up with some profoundly new way of seeing the world. but how do these a i models actually become so good at generating images? it's down to the data that they learn from, which also includes artworks. this has led to many complaints by artists. some of

them have him suit. they argue that their art has been used to train a eyes without their consent. but how does art even get into these datasets? among others, a non profit organization from hamburg is responsible for this lie on, has trawled through the internet to compile different data sets with a total of 6000000000 images. apart from stock photos, they also include private images and up lyons. data sets are free under open source licensing. they've been used to train big ai programs like mid journey open source data sets are supposed to enable more people to set up a i program. companies that set up business models based on data sets of other people's images is something that i know ice is critical of that when these of him not cain and even of these companies are not generating profits yet. they should be held responsible for using other people's images and

should pay up. duffy and paul creators are dependent on being paid for the work of i to leave him when it comes to imagery, eyes, copyright issues are still unresolved. this also becomes a problem when artists use the technology for their own art, his, my kind and then vanished with no saying that an image i created using a i as someone else's property doesn't make sense to me. here at o e. good year, tarver. human van would make a mockery of my life as an artist and my job as a professional artist who leads to live office art inflicted upon limby at a fraud on few. who owns the images created by a i? and what role does the art used for inspiration play? difficult questions? text to image eyes have other issues too. they often reproduce sexist and racist stereotypes. if the data says used for training on neutral, neither are the images, the lens, the app create stylish portrait photos from a few selfies. lenses was also trained using ly on data sets,

a colleague and i tested the app. i'm a white european males. why? the self is i uploaded were no beauty shots. the results i got were mainly very manly versions of astronauts and superheroes. most pictures i could recognize myself and there wasn't much skin on shoulder. my colleague arianna comes from columbia. she was confused that the ap presented her in a very european look in most pictures with white a skin light brown hair, and even blue eyes. and many resolves showed a lot of skin or tied jeans, even though none of the selfish she had uploaded to did that. unfortunately, every eye is only as good as the data. it's based on what are your thoughts on the eyes that generate images from text upon toy or the future of art. let us know by for now, ah,

and 2000 kilometers in an electric camper. a check in special with alison and eric b lasted the d. w. travel recorders and youtube ers drive across europe on electric power tourist highlight and insider, tip and elect. you're buying road credit check in o d w india o. brimming with ambition and inequality. 75 years ago, mahatma gandhi peacefully led the country to independence and what is left of his vision with gandhi's legacy.

in 30 minutes on d. w. or what secrets lie behind these walls? discover new adventures in 360 degrees. and explore fascinating world heritage sites. d w world heritage 360. get the app now ah ah ah ah ah, and welcome to the special edition of check in. to date we're back to accompanying d. w, traveling reporters and you tubers.