Pixel farming involves addressing weed control by training AI to recognise individual “pixels” within a field, after which a robot can carry out weed control using high‑power laser technology. The laser effectively heats the plant from the inside, disabling it. This enables weed management without the use of chemical agents and without disturbing surrounding crops or the soil. This innovation illustrates how AI technologies are used to specify decisions and interventions down to the level of individual plants, and how AI can contribute to more sustainable agricultural practices.
From an ELSA research perspective, however, this application also raises important ethical, legal, and societal questions. Decision‑making about when and how to intervene partially shifts from the farmer to an autonomous system. This raises questions regarding responsibility and liability: who is accountable when errors occur or unintended effects arise? In addition, transparency is crucial. Farmers want to understand how the system detects weeds, which criteria are applied, and on what basis a laser action is triggered, in order to maintain professional autonomy and trust in the technology. These applications also touch upon broader societal themes, such as accessibility (for whom is this technology feasible?), potential dependence on technology, and changing power relations between farmers, technology providers, and other actors in the value chain.
Pixel farming was involved during the design phase of the ELSA Scan through a stakeholder workshop aimed at identifying these ELSA aspects.
