Socially intelligent robots are becoming increasingly important in human-robot interaction (HRI). The concept of social intelligence in robots has evolved from early AI research, which focused on simulating human intelligence, to a more embodied and situated approach that emphasizes the robot's ability to interact with and adapt to its environment. This paper explores the dimensions of HRI, discussing the requirements for social skills in robots and introducing the conceptual space of HRI studies. It also presents two examples of HRI research: one on the development of a cognitive robot companion and another on the use of robots as educational or therapeutic toys for children with autism. The paper concludes by examining different paradigms regarding the 'social relationships' between robots and people. The paper discusses the nature of artificial intelligence, highlighting the shift from a 'classical' AI approach to a 'nouvelle AI' perspective that emphasizes the embodied nature of intelligence. It argues that social intelligence, a key aspect of human intelligence, may be a crucial component for creating socially intelligent robots. The paper also explores the different approaches to HRI, including robot-centred, human-centred, and robot cognition-centred perspectives. It discusses the importance of social skills in robots, particularly in applications where robots interact with humans, such as in service robotics, education, and therapy. The paper presents a case study on robot companions, emphasizing the need for robots to be socially intelligent and to adapt to human preferences and needs. It discusses the challenges of designing socially intelligent robots, including the need for robots to be able to interact with humans in a socially acceptable manner and to adapt to changing environments. The paper also highlights the importance of 'robotiquette'—a set of rules or heuristics guiding a robot's behavior in its interactions with humans. The study concludes that social intelligence is essential for robots to be accepted and integrated into human environments.Socially intelligent robots are becoming increasingly important in human-robot interaction (HRI). The concept of social intelligence in robots has evolved from early AI research, which focused on simulating human intelligence, to a more embodied and situated approach that emphasizes the robot's ability to interact with and adapt to its environment. This paper explores the dimensions of HRI, discussing the requirements for social skills in robots and introducing the conceptual space of HRI studies. It also presents two examples of HRI research: one on the development of a cognitive robot companion and another on the use of robots as educational or therapeutic toys for children with autism. The paper concludes by examining different paradigms regarding the 'social relationships' between robots and people. The paper discusses the nature of artificial intelligence, highlighting the shift from a 'classical' AI approach to a 'nouvelle AI' perspective that emphasizes the embodied nature of intelligence. It argues that social intelligence, a key aspect of human intelligence, may be a crucial component for creating socially intelligent robots. The paper also explores the different approaches to HRI, including robot-centred, human-centred, and robot cognition-centred perspectives. It discusses the importance of social skills in robots, particularly in applications where robots interact with humans, such as in service robotics, education, and therapy. The paper presents a case study on robot companions, emphasizing the need for robots to be socially intelligent and to adapt to human preferences and needs. It discusses the challenges of designing socially intelligent robots, including the need for robots to be able to interact with humans in a socially acceptable manner and to adapt to changing environments. The paper also highlights the importance of 'robotiquette'—a set of rules or heuristics guiding a robot's behavior in its interactions with humans. The study concludes that social intelligence is essential for robots to be accepted and integrated into human environments.