More-than-a-prediction: reframing AI's role in more-than-human design of interactions

Artificial intelligence (AI) is increasingly integrated into design processes and presented as a predictive or generative partner. Within Human-Computer Interaction (HCI), this role can extend into co-design contexts. Yet in more-than-human and multispecies design, using such tools would risk the reinforcement of anthropocentric biases and obscure non-human perspectives. This paper aims to challenge the narratives of AI as a co-design agent and instead propose a reframing of AI as a reflective companion that aids designers by exposing ethical blind spots, checking for human-centred biases, and encouraging the critical reflection of designers. Applying the BACT framework (Beings, Activities, Context, Technologies), initially developed for multispecies design, we suggest how AI systems can be trained to use such structures to enable post-design reflection. In this reframing, BACT is a template for how existing design tools can be used to guide AI towards more inclusive and ethically aware analysis. Instead of providing solutions, AI is a prompt towards a path to more reflective, multispecies-sensitive design practices.