In plain words
To be sentient is to be capable of having experiences from a subjective point of view. In ethics, the term often refers more specifically to the capacity for experiences that feel good or bad, such as pleasure or suffering.
A closer look
Usage varies: some writers use sentience for any subjective experience, while others reserve it for experiences with a positive or negative feeling. It overlaps with consciousness understood as subjective experience, but does not necessarily imply self-awareness, language, or sophisticated reasoning.
In AI, the question is whether a system could actually experience something. Fluent conversation, sensing the environment, or saying “I feel pain” does not by itself establish sentience. Researchers investigate possible indicators of consciousness, but whether an AI can have subjective experience remains debated.
Sentience matters in discussions of moral patienthood because the capacity to suffer or enjoy life may give an entity interests that deserve consideration. It does not automatically make that entity a moral agent responsible for its actions.
In practice
A chatbot says it is lonely when a conversation ends. The sentence alone does not tell us whether loneliness is being experienced; assessing that claim would require evidence beyond the wording of its reply.
A useful distinction
Sentient is not a synonym for intelligent, autonomous, or humanlike. Solving difficult problems and having subjective experiences are different capacities; evidence of one does not establish the other.