What robots can learn from babies

The idea that robots might one day replace humans is no longer just in the realm of fantasy or science fiction. Roboticists are getting better at programming androids, and are working hard on trying to enable artificial intelligence to learn and develop by itself. To do this, they are more and more looking at how babies and young children learn.

On the other hand, machines that might learn like children provide deep insights into how the mind and body act together to build children’s knowledge and skills. And that’s where such topics become of interest to us: we constantly keep ourselves updated with the latest findings in children’s development theories and practices.

In the last weeks, two articles in prominent publications have focused on differences, similarities and possibilities of learning abilities in babies and in artificial intelligence:

Self-Taught Robots, Diana Kwon, Scientific America, March 2018.

How Babies Learn and Why Robots Can’t Compete, Alex Beard, The Guardian newspaper, 3 April 2018

Some of the aspects of these articles that struck us the most, and confirmed what we knew on the extraordinary learning abilities babies have, are:

  • How special is the way in which babies interact with their environment

While robots can be programmed to read their environment (through cameras, ear microphones, etc.), and can respond only to those inputs they are instructed to detect, active learning from the surrounding environment is not just a prerogative of humans: animals do that too. However, what makes babies unique is that, not only they learn from external inputs and by interaction with their environment, they are social and emotional learners. Babies can read cues, understand people around them and interpret their intentions.

  • Human attention is the starting point of conscious human learning

In the past, much of the efforts of those working on building learning robots have focused on teaching them language. What roboticists have learnt from this, is that human language learning and development is much more than just an expanded vocabulary: it is a social, emotional and sensory process.

As argued by Angela Prodger (expert in early childhood development), interviewed for the Guardian article, “before children can acquire the tools of speech and language they need a sense of being and belonging”.

For a robot, “the acquisition of language is abstract and formulaic. For babies, it is embodied, emotive, subjective, quivering with life”. And, as proven by comparative research, the human-to-human contact is essential for a baby to learn language.

  • The importance of the play and physical activity for learning

This is an important aspect of children’s learning and development which is often ignored by policymakers. And it can be of high importance in, for instance, the debate whether children should start learning to read earlier. In recent years, schools adapt  their curricula and practices more and more to the general public growing expectations that children aged five must already start reading.

From the mechanical point of view, of course, children of that age can do that. However, evidence shows that it is best if kids learn to read when they can make better sense of it. Studies proved that we do not help children become better readers by forcing them to start on the decoding before they can have an underlying understanding of story, experience, sensation and emotion. Quite the opposite. In other words, treat kids like robots during early learning and you put them off.

Instead, it is essential for young children to have the time and possibilities to be physical, to have those early years dedicated to learn mostly through experiments and sensory experience in interaction with their physical and social environment.

It is gratifying for us that what those in the early childhood research community have known for decades about just how sophisticated babies learning capacities are, is now being discovered and valued by new audiences in the science community. Also, we find it exciting that collaborations between child development experts and roboticists might lead to new insights for both fields. We look forward to the next discoveries.

Part of what we do at ICDI is, in fact, to translate the most important scientific research into practice in a way that is relevant and meaningful for the families, communities and professionals we work with, in all parts of the world. Some of our active projects dealing with how young children learn and develop best are:

SEED – Social and Emotional Education and Development, through which we aim to build practitioners’ capacities to establish strong, nurturing relationships with young children growing up in difficult circumstances to better support their psychosocial well-being and development.

Making the first 1000 days count!, where part of the activities focus on the importance of play and social interaction in young children’s learning and growth in Ethiopia.