Extended Intelligences¶
About the faculty: Pau Artigas
Pau Artigas is an Interactive Web Developer at Taller Estampa. Estampa is a collective of programmers, filmmakers and researchers, with a practice based on a critical and archaeological approach to audiovisual and digital technologies. Since 2017 they have developed an important amount of work focused on the uses and ideologies of AI, an interest that started with a project programmatically entitled The Bad Pupil. Critical pedagogy for Artificial Intelligences (2017-2018).
Some notes about the seminar: AI as automation Exposing.ai - avoiding face recognition with data Feminist Data Set 2017 What we want to automate mean images from Hito Steyerl : regenerative neural neds, she talks about statistical renders: https://newleftreview.org/issues/ii140/articles/hito-steyerl-mean-images
AI as an expression of current ideology Books - The wretched of the screen - Hito - AI in the age of surveillance - Desconfiar de las imagenes - Harum Farocki - Photogrametry - going back to the past and used with data. She talks about cooperative images, used to generate a operation. Harum Farocki - Operational images - Jussi Parikka - Invisible images (Article) - Trevor Paglen how images are used not for human but for triggers ///Research this - The end of Man - Joanna Zylinska - The perception machine - our photographic future between the Eye and AI
(industrial) AI as an infraestructure
Amazon Mechanical Turk - crowdsourcing marketplace where enterprises ask for small tasks. One of its uses is for generating Moderators - AI generated or ppl working for bann some stuff from internet. carbon footprint natural resources human labor intl scale perspective of the resource
Important book OF AI Atlas of AI - Kate Crawford take a look to https://www.katecrawford.net [obs: Research this] Atlas of anomalous AI ///Research this Foreword by bill sherman - Borges has an appearce here [obs: Research this] Ways of being - james bridle [obs: Research this] Mentes paralelas - laura tripaldi [obs: Research this]
Ai as an ubiquitous technology From promt to image: https://stablediffusionweb.com Check many data sets availables here: https://paperswithcode.com/datasets Check the technologies data sets availables here: https://paperswithcode.com/sota Also check keglle: https://www.kaggle.com/datasets ’ Check how gpt3 was trained in Google Gpt3 trainning dataset Dall-e3 Midjourney Microsoft Office Bard Adobe Photoshop https://github.com/openai/whisper https://wellcomecollection.org/ www.adam.harvey.studio/vframe/ ///Research him https://teachablemachine.withgoogle.com/train/image Use colab to make requests to Apis easily: https://colab.research.google.com/drive/1v1QDp2xvT63N3HTojsMs-Pn6wel9WMVn#scrollTo=93AW3QCgKg3O USe replicate to extract the API: https://replicate.com/account/api-tokens https://quickdraw.withgoogle.com/ https://huggingface.co/ ai community
Latent Space Abstract multidimensional space that encodes a meaningful internal representation of externally observed events. Usually they are high dimensional: 300, 512, 2048
Additional Resources¶
Alpaydin, E., 2016. Machine Learning. The new AI. Cambridge, Massachusetts: the MIT Press.
Bridle, James: New Dark Age: Technology and the End of the Future. London: Verso, 2018
Bridle, James: Ways of Being. Allen Lane / Penguin, 2022
Crawford, K., 2021. The Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press.
D’Ignazio, C., Klein, L. F. (2020). Data Feminism. The MIT Press
Joler, V., Pasquinelli, M., 2020. Nooscope.
Kogan, G., 2016. Machine Learning for Artists (Collection of free educational resources). Github.
Miller, A., 2019. The Artist in the Machine: The World of AI-Powered Creativity. Cambridge, Massachusetts: The MIT Press.
O’Neil, C., 2016. Weapons of Math Destruction. How Big Data Increases Inequality and Threatens Democracy. UK: Penguin Random House.
Paglen, T., 2016. Invisible Images (Your Pictures Are Looking at You). The New Inquiry. Brooklyn.
Sautoy, M., 2019. The Creativity Code: How AI Is Learning to Write, Paint and Think.
Sinders, Caroline: Feminist Data Set, 2020
Steyerl, Hito, 2012. The Wretched of the Screen.
Steyerl, Hito: “Mean Images”, New Left Review, 140/141, March-June 2023
Vickers, Ben; Allado-McDowell, K: Atlas of Anomalous AI. Ignota Books, 2020
El mal alumne ––Pedagogia crítica per a AI¶
Extract from the book - Estampa Compulsive experts p. 55 The world of art has been the starting point for two training sessions. In the first one, Wikiart’s categorization categories and its image corpus were used to teach a network to recognize artistic styles. In this situation, artificial vision is forced to confront the idea of artistic style and movements, doing so from a purely formal perspective, avoiding any conceptual consideration, an area that, by definition, is beyond its understanding. In the second case, the training datasets consist of collections from various artistic institutions in Barcelona (MACBA, MNAC, Fundació Miró, Fundació Tàpies, Museu Picasso, and Museu del Disseny). The images in each collection are extremely heterogeneous. Forced to synthesize them, the network ends up producing absurd interpretations. In these cases, artificial vision serves to satirize the discourse on art based on classification, which inevitably homogenizes and simplifies artistic production (img. 24 and 25). While image classification networks are obsessive and compulsive, linking everything to the vocabulary they have learned, these two generate connections between any image and artistic vocabulary (img. 22).
The training of artificial vision networks has also been carried out in other areas. In some cases, efforts were made to have the network focus more on the image than on the objects represented: we are talking about identifying concepts related to composition or representation of space (vanishing point and horizon, assuming the ideal spectator’s perspective) or devices for creating the image (for example, webcam, phone, etc.). In other cases, the network was trained to identify particularly elusive artists such as Cindy Sherman or Joan Fontcuberta (img. 23).