Started in 2018, Machine Agencies hosts students, artists, researchers and professors in their shared activities and collaborations.
Our cultural practice for a machine agency to come shall be guided by three imperatives that we encourage members to engage with:
- Our cultural practice is a craft practice. As opposed to engineering or art, we shall approach making with machine agency as a craft practice. Craft emphasizes the embodied material aspect of code, let alone electronics, but also extends to the work of making the relations and institutions in which machine agency will participate. Craft is always revisable, adaptable, and adapted. In craftwork there is no single path and there are multiple means to achieve similar ends. At the same time, craft is appreciable as a means and not an end. Craft is not about whatever works but about how it works. A critical element of craft is style. Craft is tied to culture rather than denying culture. The situated practice of craft is not just materially situated but culturally situated. Craft will not permit the god-trick as the craftsperson is always present to their work. In craft the language model and techniques for the production of the model are inseparable so that we may ask valid artisanal questions about how and why a model works as it does.
- Our cultural practice will cultivate endurance. As opposed to speed, we will not move fast and break things. We are looking for a practice that does not necessarily move slow but it is more sustainable, more resilient and more inclusive. Since capital investment in machine learning demands immediacy, we will cultivate alternative pathways which need not capitulate to this demand. There is nothing new in this parable of the tortoise and the hare. We want a machine agency capable of endurance not speed and a form of machine learning that is more holistic and less instrumental, as well as the institutional structures, policies, and investments that enable this.
- Our cultural practice will emphasize the play of agency. As opposed to simulation, we articulate machine agency as open-ended elements of socio-cultural relations. If simulation looks to recreate the human, the city, the society, the mind… then play looks to engage this as a problem not a solution. Making a human or better-than-human actor is a game not a given.
These are our practices for a critical conjuncture of creativity, culture, and technology. Generative AI has made apparent our posthuman entanglements, a vibrancy instrumentalized as automation. That is because we see the AI commons as already here. ChatGPT reveals the possibility of the commons, the Internet as a common knowledge, and in its dispossession of the commons, we can see what we never could.
We want to first name this entanglement as the Already-AI commons in recognition that the global, shared and collective entanglement on which machine learning relies and for which we have no other choice but to participate.
The task of Machine Agencies is to imagine AI commoners otherwise, to borrow a phrase from Beth Coleman. We need a cultural practice as commoners.