Social Robots in Class

What research reviews have found

A 2018 review in Science Robotics by Tony Belpaeme and colleagues looked at social robots used as tutors or peer learners. It reports that they can increase both cognitive and affective outcomes, and that they have achieved results similar to human tutoring on restricted tasks. The authors attribute much of this to physical presence, which screen-based learning technologies lack. That qualifier, restricted tasks, matters for planning. The review supports giving a robot a focused, well-defined job rather than a whole lesson. Use one where a patient, repeatable partner helps pupils practise, and keep the teacher in charge of explanation, feedback and everything else.

Where robots have been tried in schools

A 2024 systematic review in Revista de Educación a Distancia by Ivana Ružić and Igor Balaban of the University of Zagreb gathered studies of social robots in primary and secondary education. It found robots used for language learning, computational thinking, social and emotional development, creativity support, musical instrument practice and library activities, in roles ranging from teaching assistant to peer and learning companion. The NAO humanoid stood out, particularly with primary-age children. For a school, that list is a useful starting point: pick one of these areas, ideally where staff already see a need, rather than buying a robot first and searching for a use.

Concerns the research raises

The same review is frank about risks. Ružić and Balaban highlight problems of transparency, dependence on robots and reduced human interaction, along with privacy risks and possible job displacement, and they call for long-term studies and cooperation among educators, policymakers, developers and privacy experts. Take those points into your planning. Tell pupils plainly what the robot is and is not, keep sessions short so it does not crowd out conversation with adults and classmates, and agree how any recordings will be handled before a camera or microphone is switched on. Reassure staff that a robot assistant is a tool for them to direct, not a step towards fewer teachers.

Testing a robot assistant fairly

A fair comparison needs a control group. Choose one restricted task, such as spelling practice or times tables, and measure pupils' attainment and engagement before you start. Run the robot sessions with one group and the same activity led by an adult or on a tablet with another, then swap the groups halfway. Ask teachers to log what went wrong as well as what worked, including set-up time and technical faults. On this site, QTrobot, Misty II, Abii, Kebbi Air and Sanbot Elf are among the robots you could compare, but judge them on your own results rather than on a polished demonstration.

Sources and further reading

• Social robots for education: a review — Science Robotics (Belpaeme et al.), via University of Plymouth PEARL.
  Review of social robots as tutors and peer learners, and how their physical presence affects learning.
• The use of social robots as teaching assistants in schools: implications for research and practice — Revista de Educación a Distancia (Ružić and Balaban), University of Murcia.
  Systematic review of social robots in primary and secondary schools: subjects, roles and ethical concerns.

Common questions

How long should a robot session last in class?

There is no fixed rule. Short sessions built around one activity, followed by discussion with the teacher, fit both the research on restricted tasks and the practical limits of batteries, set-up time and pupils' attention. Build up gradually and watch for interest fading.

Is a humanoid robot better than a tabletop robot for teaching?

Not necessarily. A humanoid form can help with gestures and social cues, while tabletop robots are cheaper, easier to move and less distracting. Decide by activity: language and social tasks may benefit from a face and arms, whereas practice drills often do not.