A Vision for AI-driven Neural Implants: Towards Ethically Responsible Translation

Abstract

Neural implants are technologies that directly interact with the brain and are being developed to restore impaired neurological functions, such as vision, hearing, or speech. Increasingly, artificial intelligence (AI) is used in these technologies to interpret complex brain and sensory data. Several neural implants, such as many brain-computer interfaces (BCIs), are still in the research phase. This thesis examines the ethical aspects of AI-driven neural implants and the question of how these technologies can be translated in an ethically responsible way. Three applications are used as use-cases: a visual neural implant for people with blindness, a speech-BCI for individuals with locked-in syndrome, and an improved version of the cochlear implant for people with severe hearing loss. The research questions are answered through literature studies, interview studies with developers and prospective users, and conceptual analysis. The research shows that while these neural implants offer important opportunities, they also come with uncertainties and high expectations. Expectation management is therefore essential, particularly in clinical research. Additionally, neural implants have the potential to enhance users’ autonomy, for example by improving their independence. Yet, their autonomy could be undermined if the system makes mistakes causing a loss of user control. Furthermore, neural implants and their intensive clinical trials can be, not only physically, but also mentally invasive. In particular, the end of clinical trials raise ethical questions. For instance when implants are removed while users benefitted from them. A central conclusion of this thesis is that the ethically responsible translation of AI-driven neural implants requires realistic communication, early involvement of the perspectives of (prospective) users, and the definition of long-term responsibilities of researchers.