This thesis opposes incorporating artificial intelligence (AI) into our responsibility practices by developing them as artificial moral agents, on the grounds that AIs lack a necessary condition for moral agency: sentience. First, I explain how AI autonomy complicates attributions of moral responsibility. Second, I support the claim that sentience is necessary for moral agency by discussing the substantial conceptual and practical challenges with developing artificial moral agents. Third, I argue that if such agents are possible, we ought not to consider them responsible or punishable because doing so yields counterintuitive responsibility practices. Fourth, I raise the ethical concern that creating AIs that apparently meet the demands for moral agency would confer substantial obligations on their developers and users. Lastly, I suggest that, rather than seeking to create artificial moral agents, we should develop AIs as tools for human use and retain human responsibility in high-stakes settings like medicine and warfare.