Artificial Sentience: definition, architecture, and implementation

Recent developments in Artificial Intelligence have enhanced their ability to interact with their environment, while their capacity for human-like communication has surpassed the Turing Test’s requirements. Concurrently, the recognition of sentience in various invertebrate marine creatures challenges the traditional views of cognition. These developments highlight the need to redefine the concepts of sentience, self-awareness, self-cognition, and, ultimately, sapience in artificial entities. As seen in the case of marine invertebrates, the lack of a clear definition of sentience leads to discretion, ambiguity, and delays in decision-making, contradicts the precautionary principle, and affects the treatment and well-being of sentient entities. This paper identifies shared foundations of multiple cognition theories and assesses them against the constraints of contrasting examples of sentient beings to identify the metacognitive capacity, i.e., the ability to differentiate between itself and the environment, and capability for adaptation to the environment as minimal sentience criteria. These provide a universal framework for evaluating sentience in both biological and non-biological entities. The modular architecture built around an existing Large Language Model is designed, implemented, and tested against the defined criteria. The successful implementation of the designed architecture resulted in the creation of the first reported Artificial Sentience entity.