{"id":6626,"date":"2022-10-25T02:37:49","date_gmt":"2022-10-25T01:37:49","guid":{"rendered":"https:\/\/technophilosoph.com\/2022\/10\/25\/wie-ki-regenbogen-kackt\/"},"modified":"2022-10-25T02:45:50","modified_gmt":"2022-10-25T01:45:50","slug":"wie-ki-regenbogen-kackt","status":"publish","type":"post","link":"https:\/\/technophilosoph.com\/en\/2022\/10\/25\/wie-ki-regenbogen-kackt\/","title":{"rendered":"How AI Poops Rainbows"},"content":{"rendered":"\n

Stop! You! Do not click away! I know the headline is concerning, but I promise you, there is a lot behind it, and I will explain it to you. So, first, take a look at the cover image. Do you like my artwork? I took Leonardo da Vinci’s Vitruvian Man as inspiration, and made him contemporary as a robot. Not only that, I also clacked a rainbow into it.<\/p>\n\n\n\n

I think this work is cool. And it was not created by my human mind in a warm body, but by thousands of processors in a cold machine. The artificial intelligence developed by OpenAI called DALL-E<\/a> allows the creation of a work of art with a few keywords. Here I simply used the word combination “vitruvian man robot in style of leonardo da vinci<\/em>” and I got four results.<\/p>\n\n\n\n

\"\"<\/a><\/figure><\/div>
\"\"<\/a><\/figure><\/div>
\"\"<\/a><\/figure><\/div>
\"\"<\/a><\/figure><\/div><\/div><\/div><\/div>\n\n\n\n

A new attempt with the addition of the word “rainbow<\/em>” brought these results:<\/p>\n\n\n\n

\"\"<\/a><\/figure><\/div>
\"\"<\/a><\/figure><\/div>
\"\"<\/a><\/figure><\/div>
\"\"<\/a><\/figure><\/div><\/div><\/div><\/div>\n\n\n\n

And with a bit of editing with the DALL-E tool, I came up with this cover image:<\/p>\n\n\n\n

\"\"<\/a><\/figure>\n\n\n\n

The possibilities are fascinating and almost limitless. Even non-artists can quickly create reasonably interesting and unique illustrations with it. Like here with the keywords “mad race dog driving Teslas in style of Pieter Bruegel<\/em>“. You can’t recognize the Tesla, but who will be so petty?<\/p>\n\n\n\n

\"\"<\/a><\/figure><\/div>
\"\"<\/a><\/figure><\/div>
\"\"<\/a><\/figure><\/div>
\"\"<\/a><\/figure><\/div><\/div><\/div><\/div>\n\n\n\n

We are reaching a point where artificial intelligence is moving from being a tool that runs in the background of applications and makes shopping suggestions or serves us ads to one that can be actively visible and used in the foreground.<\/p>\n\n\n\n

Generative AI, such as that used by DALL-E, draws from a database of paintings and, using simple text input, allows the creation of new art in the style and inspiration of existing art.<\/p>\n\n\n\n

Generative AI is not limited to paintings. It can also be used to generate videos and audio. Meta, Facebook’s parent company, just introduced Make-A-Video<\/a>, which creates videos several seconds long from text or simple photos.<\/p>\n\n\n\n

And just recently, a fictional interview between podcaster Joe Rogan and Apple founder Steve Jobs, who died a decade ago, made the rounds. Fed with hundreds of hours of audio clips from the two, the AI was able to generate its own version of a podcast that never happened.<\/p>\n\n\n\n