Ulrich Fischer
@ulrich
97
141
4.19k
fredcavazza.net/2023/03/20/intelligence-artificielle-nous-navons-pas-besoin-de-plus-de-contenus-mais-de-meilleures-analyses/
Mar 26, 2023
2
www.meta-media.fr/2023/03/18/sxsw-2023-who-cares-about-reality.html
Mar 26, 2023
1
www.superception.fr/2023/03/18/newsletter-superception-353/
Mar 23, 2023
1
www.superception.fr/2023/03/11/newsletter-superception-352/
Mar 23, 2023
1
nesslabs.com/mind-garden
Mar 23, 2023
1
www.cbinsights.com/research/generative-ai-publisher-licensing/?utm_source=CB+Insights+Newsletter&utm_medium=email&utm_campaign=newsletter_general_tues_2023_03_21&utm_term=block-1&utm_content=research-client
Mar 22, 2023
1
www.lebigdata.fr/gpt-4-tout-savoir
Mar 17, 2023
1
www.drorpoleg.com/ai-and-the-long-tail/
Mar 17, 2023
1
battellemedia.com/archives/2023/03/ai-sausage-making-and-unconsidered-consequences
Mar 17, 2023
1
writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/
Mar 16, 2023
1
fredcavazza.net/2023/02/24/mythes-et-realites-des-ia-generatives/
Mar 16, 2023
1
medium.com/whither-news/journalism-is-lossy-compression-86380f0bdb50
Mar 16, 2023
1
medium.com/whither-news/writing-and-exclusion-16d4d98a27a
Mar 16, 2023
6
nymag.com/intelligencer/article/ai-artificial-intelligence-chatbots-emily-m-bender.html
Mar 13, 2023
2
www.timesfreepress.com/news/2023/mar/11/opinion-imminent-danger-of-ai-tfp/
Mar 13, 2023
1
bakadesuyo.com/2023/02/memory/
Mar 5, 2023
1
fredcavazza.net/2023/02/10/avons-nous-encore-besoin-des-ordinateurs-et-imprimantes/
Feb 17, 2023
1
la-rem.eu/2023/01/la-fin-annoncee-du-petit-ecran-convergence-et-divergences-entre-svod-television-et-plateformes/
Feb 17, 2023
1
affordance.framasoft.org/2023/01/gpt-3-cest-toi-le-chat/
Feb 5, 2023
2
prospective2023.cmf-fmc.ca/fr/La-creativite-augmentee/la-creativite-sensorielle
Jan 26, 2023
3
aiascendant.substack.com/p/the-fine-art-of-prompting
Jan 22, 2023
1
craigmod.com/roden/075/
Jan 9, 2023
1
johnbattelle.medium.com/is-conversational-media-dead-4d0e17ce44ba
Jan 9, 2023
1
explorewhatworks.com/rethinking-creator-economy-gina-bianchini/
Jan 9, 2023
1
www.howtomakesenseofanymess.com/chapter1/21/information-is-not-data-or-content/
Jan 9, 2023
1
Invisibly fundamental to all of them, I think, is a better mental model of what the AI is actually doing. The naïve assumption is that ChatGPT answers, and Stable Diffusion creates. This is not correct. Large language models are actually trained to take blocks of text and continue them. ChatGPT channels this into dialectical Q&A, but under the hood, it really “wants” to extend whatever prompt you give it, not respond. (This is why, for instance, telling it what not to do almost never leads to better output.)
Similarly, it’s better to think of diffusion models like Stable Diffusion not as generators, but as infinite art museums containing all imaginable images, à la Borges. They use your prompt not to create but to guide you, through their endless chambers, to the nearest with art that matches your prompt.