Graphi Novel Reviews


Dream Machine
A Portrait of Artificial Intelligence

(2024) Appupen and Laurent Daudet, MIT Press,
£22, hrdbk, 160pp, ISBN 978-0-262-55129-8

 

Originally published by Flammarion Press in Paris, 2023 as Dream Machine ou comment j’ai failli vendre mon âme à l’intelligence artificielle [Dream Machine or how I almost sold my soul to artificial intelligence], the difference between the French and English subtitles perhaps reflects a shift in attitude towards AI as this graphic novel crossed the Atlantic. Daudet is a physics professor and self-described AI entrepreneur who met Appupen, (real name: George Mathen) while the latter was attending a comic arts residency in Paris in 2021.

Surprisingly perhaps, given the subject, the artwork, rendered in black and white with blue-grey highlights, is straightforward, with no innovative arrangement of panels or dramatic use of borders. Nevertheless, it is quite evocative and nicely balanced which makes it easy for the eye to follow. Perhaps that is intentional as a counter to the weightiness of the information contained in the story which spans ten chapters, each of whose titles include the words ‘dream’ or ‘dreams’.

The plot, such as it is, revolves around Hugo, the founder of a company that has developed powerful Large Language Models (LLMs) which have begun to attract the attention of a much larger concern, called ‘Real’ which was founded by an eccentric billionaire (sound familiar?). Although Real presents itself as a gaming company, seeking to expand its all-encompassing ‘metaverse’ around the world (again, sound familiar?!), it soon becomes clear that it has much bigger and more profoundly worrying ambitions. As our hero is wooed by representatives from this behemoth and is tempted by the huge sums of money involved, he bumps into a variety of friends and colleagues who warn him about the consequences of selling out and further advancing the spread of AI. His havering throughout the bulk of the book affords the opportunity for a series of clunky info-dumps covering what ‘AI’ actually is, how it’s being developed, what the ethical implications might be and so on.

To add to the somewhat belaboured feel of the things, each chapter begins with a sequence featuring Hugo’s alter-ego ‘Super-Hugo’ (yes really) who, masked and caped (yes, *really*!) fights off various threats such as a data deluge and, more plausibly, killer drones. According to Appupen (writing on the companion website, dreammachine.in) the purpose of these is to give a glimpse into the ‘REAL.E’ game and to spark our imaginations. Honestly, however, in this post-Moorean age, it comes across as somewhat flat footed.

More interestingly, but also contentiously, the final chapter features a selection of alternative scenarios ‘written’ by ChatGPT itself based on prompts by the authors, with illustrations confected by the Stable Diffusion model using over 500 drawings from the first chapters of the book. It is perhaps worth noting here that in a recent ruling, a US judge found that Stable Diffusion may have been built ‘to a significant extent on copyrighted works’ and was ‘created to facilitate that infringement by design’ (aibusiness.com/responsible-ai/artists-copyright-claims-against-ai-image-generators-advance-in-court).

On the Dream Machine website Daudet insists that “What we make out of AI is not a technical issue, it's a political choice.” We have heard such sentiments expressed before in the history of science, of course, and some of us may well raise sceptical eyebrows. Whatever one’s view on that, as an attempt to give us the means to make such a choice, this work falls rather flat: it offers nothing new for those who already know something of LLMs and artistically, or otherwise, it contains little to entice the average comic book reader.

Steven French

 


[Up: Fiction Index | Top: Concatenation]

[Updated: 25.4.15 | Contact | Copyright | Privacy]