The Ravenous Machine

The A.I.'s need for data is fed by web users unknowingly, through seemingly harmless activities. Security systems, such as reCAPTCHAs, also train AI for use in the military, raising ethical issues of exploitation and transparency.

This project investigates the 'insatiable hunger' for data of artificial intelligence (A.I.), trained by labelling and describing the data that make up the datasets. Often these activities are carried out unwittingly by web users, who are offered tests, sweetened by graphics and elements, such as kittens, that convey the feeling that they are doing something harmless.

Through photographs, video frames, A.I. generated images, and graphic interventions, I explored how systems designed for the security of users and web services and sites (such as reCAPTCHAs) actually also contribute to enhancing A.I. based systems, for instance those of autonomous driving systems, but also in the military field (for obstacle recognition or target detection by war drones).

This generates a short circuit between what is supposed to be for protection and security purposes and what instead exploits human skills to train this new technology; between what is used to recognise a human being from a bot and what instead helps bots to imitate human beings, in an increasingly accurate and alarming way.

Critical questions arise from all this about ethics, transparency, the unconscious exploitation of people and our future.