“Genesis” by Jens Knappe – a Standard in Generative AI Literature

“Genesis” by Jens Knappe – a Standard in Generative AI Literature

In October 2022, more than a month before ChatGPT hit the world, took it by storm, started the generative AI craze we see to this day, and began inflating the biggest financial bubble since Dot.Com, we published “Genesis”. This “Creation Story in Collaboration with an Artificial Intelligence” by Jens Knappe was an experiment without a clear genre identity, highly original, a wild hybrid of something between a photo book with sci-fi comic elements, a journalistic text, and a serious scientific exploration of a bleeding edge technology. Something where abundant artistic phantasy hits painstakingly rigid, down-to-earth scientific examination.

The first reactions to the book were somewhat lame and also disappointing. It had clearly met an utterly unprepared audience. Almost no one recognized the true nature and scope of it, most media at that point still considered all this stuff science fiction or fringe tech. With the ChatGPT accelerator, the magnitude of all this became more apparent to a larger audience. And with the first serious critical acclaim – with a delay of a few months from the original publication date – a real explosion of interest in the book took place, as well as a rush of orders for it, on a scale we had never experienced before.

Jens Knappe AI book Genesis
Jens Knappe AI Book “Genesis”

“Genesis” received critical praise: “An entertaining as well as razor-sharp account of how generative AI, specifically image generators, actually work” (SF Chronicle), “recommended” (MIT Tech Review), “wort und bildgewaltig (eloquent and visually powerful)” (FAS) or simply “a wonderful book” (Photobookjournal). And, something that makes us very proud, not being one of the prestigious scientific publishers to whom this honor is usually exclusively reserved to: “Genesis” has lately become a mandatory textbook in generative AI classes at US universities.

The book can be bought at all major bookstores and outlets worldwide, or simply here:


and researchers should be able to find a copy in a nearby library there:


What’s next? Jens Knappe has lately been taking a closer look at the issue of Artificial Creativity, and recently joined a project with one of the major organizations in the field, on the borderline of Mechanistic Interpretability research, with the – at first glance – simple and innocent-sounding underlying question: “Why do the outputs of these systems look so good and coherent?” The issue of why these AIs do not just throw out wild mishmashes, incoherent gibberish, and pastishes combined from shreds of data fragments they have been trained on has remained a largely unsolved puzzle. It is an enigma that lies at the heart of current generative AI. Why does mere scaling result in what appears to be the acquisition of some kind of world model in these systems? Understanding what goes on in these “black boxes” – reconstructing and reverse-engineering the workings of a vast number of adjustable parameters, trillions of numbers – is on par with the challenge of deciphering processes in the human brain.

In the future, we hope to make some of this research available to the public.