A ‘journey to trust?’ Machine learning, faith, and the discourse of ‘trusted’ autonomous systems

While societal trust is eroding, military power is consolidating itself in the desire to perfect AI-enabled technology with the view towards greater ‘trusted’ autonomy. Much of the international community agrees that autonomous weapons systems must be trusted, should they satisfy ethical and legal requirements. To this end, numerous AI ethics and legal principles have been drafted to help assure autonomous systems development. However, drawing on assemblage theory and discourse analysis, we shift the focus away from whether trust in autonomous systems is possible vis-a-vis ethical and legal principles. Instead, we examine the performative role that trust plays. Specifically, we confront the discourse of ‘trusted’ autonomy to question the meaning of so-called trusted autonomous systems, particularly since, as we claim, machine learning algorithms agitate and undermine trust. We argue that trust is underpinned by an under-acknowledged concept deeply tied to militarization: faith. We claim that what is actually meant in the trusted autonomy discourse, and which remains a tacit question to the public in the rhetoric, is not to simply have trust in autonomous systems, but to have faith. Faith commitments to military technology development reveal that belief in trusted autonomy and the drive towards militarization reinforce each other. We aim, therefore, to make explicit the often-implicit power relations that go into the making of trusted autonomous systems discourse.