Autism is often defined in the scentific literature as a problem to be fixed, and social robots have been leveraged to enable corrective actions with the aim to improve autistic individuals’ quality of life and care. However, the definition of “autism-as-problem” is criticized by the autistic community and disability scholars. Nevertheless, human-AI Interaction researchers rarely engage with these criticism nor with the methodological approaches usually highlighted in Critical Disability Studies and Critical Autism Studies. This result in the development of Social Robots that focus on corrective actions for autism instead of empowering actions. In this extended abstract, we highlight two approaches, namely Critical Discourse Analysis and Inclusive Collaborative Design, that are highly relevant to the study of social robotics for autism. We exemplify these approaches and highlight research questions that are seldom asked in the study of social robotics for autism.