In traditional approaches to develop software systems that do not have an Artificial Intelligence (AI) or Machine Learning (ML) component, requirements determination activities – also called Requirements Engineering (RE) – are well-established and researched. When it comes to building software systems with one or more AI/ML components, the process is dependent heavily on data with limited or in some cases no insight into the non-functional requirements of the system’s internal workings such as fairness, accountability and transparency, inter alia. In this paper, I review literature to probe how an emphasis on non-functional requirements during requirements determination can help address ethical requirements when implementing AI systems. The results show that currently technical experts often emphasize functional requirements more than non-functional requirements. Where they do, the emphasis is still on system-oriented non-functional requirements such as portability, maintainability, opacity, inter alia. However, AI being an interdisciplinary subject, encompassed in a bracket of sociotechnical systems, there is need to revisit this tendency by technical experts. I propose a framework for classifying ethical requirements under non-functional requirements in order to ensure equal emphasis on environment-oriented non-functional requirements such as privacy, fairness, eco-friendly among others. I also argue that the use of a value sensitive design (VSD) approach will help in the implementation of ethical AI systems in order to address ethical concerns from stakeholders. In addition, existing technical literature has focused more on using AI to manage RE activities (AI in RE), yet very limited research has been done on RE for AI systems (RE4AI). Requirements engineering is generally accepted as the most critical and complex process within the development of sociotechnical systems. The literature also confirms that during requirements determination for AI systems (RE4AI), an emphasis on non-functional requirements could actually help address some ethical issues that are raised by stakeholders.