This project main focus is to research different aspects of art institutions digital architecture through practical development, case studies and bibliographic research. The project is propelled by the crescent virtualization of art institutions on recent decades. The main basis of this research has been carried out together with the MAC USP team, which has storically stablished itself as a space at the forefront of dematerialization and digitalization of art in Brazil. Other actions related to the project are: The expography of the exhibition for the online seminar organized by the postgraduate students from UNESP Arts Institute, Emergências*, 2020; The study of the equivalence between editorial and online design, with the development of at least one book-exhibition; Other online exhibition projects; The production of scientific articles dedicated the research and its adjacent themes. Participants: Andrey Koens, Giselle Beiguelman e Bruno Moreschi.

Comments Off on Site-Museum
Read more about the article Computer Vision in artistic contexts
Still do filme Recoding Art.

Computer Vision in artistic contexts

Set of academic and artistic experiences aiming to investigate the behavior of commercial AIs in the face of works of art images and artistic contexts. As these tools are not trained to classify artistic content, inaccuracies are frequent and, many times, they offer us algorithms that level artistic subjectivity to capitalist logic or even poetic results pertinent to the field of institutional criticism.The first project took place in 2018 at the Dutch Van Abbemuseum and was conducted by Gabriel Pereira and Bruno Moreschi, even before the emergence of GAIA. The results were: Short film Recoding Art. Academic article (pre-print) Ways of Seeing with Computer Vision: Artificial Intelligence and Institutional Critique. With the beginning of GAIA, experiments with the collection of the Museum of Contemporary Art in Barcelona - MACBA (in the context of the event CtrlZ.AI, Flowers by Bornay, Barcelona) and of the USP’s Museum…

Comments Off on Computer Vision in artistic contexts
Read more about the article Affecting Technologies, Machining Intelligences
Um dos workshops no espaço do GAIA.

Affecting Technologies, Machining Intelligences

Event held from February 5th to 7th at GAIA and the USP Institute for Advanced Studies. There were three days of presentations and collective work with art and Artificial Intelligence researchers, with an emphasis on those who research in contexts of the Global South. This project was supported by the Center for Arts, Design and Social Research and Pedro Barbosa. The seminar and workshops were free and open to the public. A publication and a website with the discussions held at the event will be launched by the end of 2020. Organizers: Katherine Ye, Dalida Maria Benfield, Gabriel Pereira and Bruno Moreschi. Speakers: Silvana Bahia (Olabi), Sabelo Mhlambi (Berkman-Klein Center e Carr Center for Human Rights), Jennifer Lee (ACLU Washington), Didiana Prata (FAUUSP), Giselle Beiguelman (FAUUSP), Katherine Ye (Microsoft Research), Amanda Chevtchouk, Tais Oliveira (Núcleo de Estudos Africanos e Afro-brasileiros), Dalida Maria Benfield (CAD+SR), Rafael…

Comments Off on Affecting Technologies, Machining Intelligences
Read more about the article Dissident design on social media
Interface do Calendário Dissidente.

Dissident design on social media

PhD research by Didiana Prata that is being developed in the Design Postgraduate Course at FAUUSP, supervised by Prof. Dr. Giselle Beiguelman, and partially held at GAIA, with coordination of Prof. Dr. Fabio Cozman. The project addresses the dissident visual narratives on Instagram and intends to investigate its aesthetic vocabulary from images that circulated on the platform during Brazilian’s presidential elections of 2018 until Jair Bolsonaro's first year’s term. The research includes a set of experiments, such as a href="" target="_blank" rel="noopener noreferrer"> Dissident Calendar. Participants: Didiana Prata, Giselle Beiguelman, Fabio Cozman, Vinicius Ariel dos Santos, Gustavo Polletti (Poli - USP), Vinicius Akira Imaizumi (Poli - USP) and Marcelo Vilela Gusmão (independent developer).  

Comments Off on Dissident design on social media
Read more about the article Brazilian workers at Amazon Mechanical Turk
Stickers mais usados pelos turkers brasileiros que participarem desta pesquisa.

Brazilian workers at Amazon Mechanical Turk

  Academic paper for Revista Contracampo – Brazilian Journal of Communication (v. 39, n. 1, 2020) of the Communication Postgraduate Program at Federal University Fluminense (PPGCOM / UFF), edited by Rafael Grohmann (Unisinos) and Jack Qiu (Chinese University of Hong Kong). This paper contributes to studies about the work on digital platforms in the Global South. It presents the results of a questionnaire survey that was distributed to 149 Brazilian workers on the Amazon Mechanical Turk (AMT) platform. The paper starts with a demographic view of Brazilian turkers and their relationship with work in general. As seen in previous studies of turkers in the US and India, AMT offers poor working conditions for the Brazilian turkers as well. We also discussed other results such as: a large number of respondents claimed to be in a situation of formal unemployment for a long period of time;…

Comments Off on Brazilian workers at Amazon Mechanical Turk
Read more about the article Exch w/ Turkers
Página inicial do projeto.

Exch w/ Turkers

  In a partnership with the online art platform aarea, GAIA participated in this project that highlights the role of humans in Artificial Intelligence and their working conditions. Although often presented as automatic and autonomous, AI depends on the work of many humans. For this, we created an interactive website (that was hosted for 20 days in the aarea domain, and now is in a static way) where it was possible to chat with some of the people who perform Human Intelligence Tasks (HITs), small tasks used to train AI systems. This work is marked by a logic of null belonging to what is being built and by very low remuneration - characteristics that the project discussed with these workers, as well as speculating possible new ways to carry out these activities. A digital and print publication will be held until the second half of…

Comments Off on Exch w/ Turkers
Read more about the article Democratizing Machine Learning
Registro de parte do website em construção.

Democratizing Machine Learning

Scientific Initiation PIBIC Project entitled Democratizing Machine Learning by undergraduate student Lucas Nunes (Molecular Sciences USP), supervised by Bruno Moreschi (FAUUSP) and Fabio G. Cozman (USP’s Polytechnic School). The objective is to create a complementary teaching material for undergraduate students in the field of machine learning and AI. It was identified that much of what is learned in the field comes from an online Coursera course, and here we intend to offer new teaching perspectives, including references that are more related to the Global South, and that also discuss the social implications of these practices and tools. The result will be a web platform with accessible interactive reading topics and completely free content. Participants: Lucas Nunes, Bruno Moreschi and Fabio G. Cozman.  

Comments Off on Democratizing Machine Learning
Read more about the article Computational_synesthesia
Estudos para transformar imagens digitais em sons.


The ‘Computacional_Synesthesia Project: between computational vision and algorithmic composition’ comprises a series of speculations initiated according to the concept of signal transduction between digital information and sensitive stimulus. In this work, we subvert the concept, which originally refers to the transfer of genetic material between cells through an intermediary, to use it as a ‘symbolic translation’ between binary code (digital images) and audio/ musical parameters (pitch, duration, amplitude and timbre). The compositional experiments also deal with the notion of synesthesia between vision and hearing, because they cause a kind of affective translation between visual and audio languages. An example is the compositional process in the work anthropophagy (a_person_sitting_on_a_chair), which is based on two axes. First, we translated symbolically the pixel vectors of the images into numerical values, which were then processed by a sound synthesizer. Second, we used commercial artificial intelligence content classifiers (such as…

Comments Off on Computational_synesthesia

GAIA Readings Meetings

The group meets every two weeks to discuss texts and other materials related to art, technology (with an emphasis on the Artificial Intelligence subarea) and society. Participants, materials already discussed and timetable here. Coordinators: Gabriel Lemos and Bruno Moreschi.  

Comments Off on GAIA Readings Meetings
Read more about the article Empirical Accuracy Law (EAL) for Sequential Machine Translation
Parte da cadeia de traduções realizada na pesquisa.

Empirical Accuracy Law (EAL) for Sequential Machine Translation

  In this research, several experiments were performed with sequential machine translation, using Google Translate tool. The experiments consisted of making sequential translations and, over 280 translations (using several available languages), evaluating the behavior of the translated text. In the academic article ‘An Empirical Accuracy Law for Sequential Machine Translation: the Case of Google Translate’, the authors found an unprecedented recurrence in the field: the fact that the accuracy of translation can be represented by a mathematical law. This law makes it possible to predict the behavior of translation chains, creating more efficient translation mechanisms for a world that is increasingly dependent on automated language devices. Also from this investigation it is possible to understand that the logic of automatic translation is related to practices such as consumption, since most of the proper names become names of brands and products throughout the applied translation processes.…

Comments Off on Empirical Accuracy Law (EAL) for Sequential Machine Translation

End of content

No more pages to load