In the age of AI, we need a human-centered society more than ever

Earlier this year, I participated in the SIGCSE conference, the special interest group for computer science education. Cecilia Aragon, a professor at University of Washington, gave the opening keynote and she focused on the idea of “Human-Centered AI”. It was certainly a timely topic, and I agreed with much of what she had to say. As we design new AI tools, it is certainly important that we take into account the needs, the interests, and the experiences of people.

But as I thought about it, I felt that the framing of “human-centered AI” was too narrow. As AI technologies spread through our society and our culture, we need to think beyond the design of the AI tools themselves. So when I gave my talk at the conference the next day, I argued that we need to work towards a more human-centered society, and we need to create a more human-centered education so that young people develop the skills and mindsets they’ll need to create and flourish in today’s fast-changing world.

That might seem obvious. Of course society and education should be human-centered. But that’s not the way the world is heading these days. We live in a world where measurement, efficiency, and optimization are too often valued above human needs, interests, and experiences.

And I worry that the situation is becoming even worse. There are two huge disruptions in the world today: there’s the disruption caused by the proliferation of AI technologies and the disruption caused by the rise of authoritarian governments. Both of these disruptions could reduce the humanness of society, so that people have less choice and less voice in their lives. Because of these disruptions, a human-centered society is more needed than ever before, and also more threatened than ever before.

What can we do to create a more human-centered society? We need to expand opportunities for children to develop the most human of their abilities. That is, we need to help them develop as creative, curious, caring, and collaborative learners. These qualities have always been important, but they are more important now than ever before, given the technological and political disruptions confronting today’s societies. Supporting children in their development as 4C learners should be the core goal of our educational systems and a core goal of parenting.

New AI technologies could be used to support this 4C approach to learning — for example, offering support to children as they work on project-based, interest-driven, community-oriented activities. Unfortunately, that’s not what’s happening. Most of today’s educational applications of AI have a very different set of goals. They are very impressive in their capabilities, efficiently generating context-based instruction and information. But they are not aligned with the needs of a human-centered society. I worry that many of the new AI applications in education will:

• reduce children’s sense of agency and control
• diminish children’s sense of joy and accomplishment
• undermine children’s opportunities for personal expression
• replace opportunities for human connection.

Consider, for example, the “AI tutors” that are now offered by many EdTech companies. Too often, these AI tutors control the flow of the educational process: setting goals, delivering information, posing questions, assessing performance. For example, when Khan Academy introduced its Khanmigo AI tutor, the first example on its website showed the tutor asking: “What do you think you need to do to multiply 2 by 5/12?” I expect that this type of tutor could be effective in helping students learn to multiply fractions. But should that be a high-priority goal in today’s educational system? If the tutor is seen as successful, it could further entrench existing educational curricula and existing pedagogical approaches at a time when fundamental change is needed.

In today’s world, we need a more human-centered approach to education, providing students with opportunities to set their own goals, build on their own interests, express their own ideas, develop their own strategies, and feel a sense of control and ownership over their own learning. This type of learner agency is important in helping students develop the initiative, motivation, self-confidence, and creativity that they will need for success and fulfillment in their work lives, their community lives, and their personal lives in today’s fast-changing world.

In evaluating new educational tools and activities, we should focus not on whether or not they make use of new AI technologies, but rather whether they are aligned with the needs of a human-centered society, helping young people develop as creative, curious, caring, and collaborative learners.

In our Lifelong Kindergarten research group at the MIT Media Lab, we have always been guided by this 4C approach. Consider, for example, our Scratch and OctoStudio coding environments, used by millions of children around the world to create their own interactive stories, games, and animations. In designing these platforms, our primary goal was not to help children develop technical skills, though that is certainly a side benefit. Rather our focus has been to provide children with opportunities to develop their interests, their ideas, and their imaginations through an iterative creative-learing process.

In short, we designed Scratch and OctoStudio to help young people develop the skills and mindsets that are needed to contribute to a more human-centered society. In this way, Scratch and OctoStudio are well-suited for the age of AI, even though they make little use of AI technologies.

The Scratch innovation team, led by Eric Rosenbaum, is now exploring ways to integrate AI features directly into the Scratch platform — but only if the new features stay aligned with the human-centered 4C approach that underlies Scratch’s success.

With these new Scratch prototypes, young people can use AI in the process of creating their own projects — thus expanding the range of what they can create while also enabling them to learn about AI in meaningful and motivating context. For example, the Scratch team has developed new programming blocks that enable young people to turn their Scratch characters into AI agents that respond to text queries.

Other new prototypes include an AI-based “creative learning assistant” within the Scratch platform, enabling Scratchers to ask for advice or suggestions as they work on their projects. One non-negotiable requirement is that children (not the AI agent) should control the flow of the creative process: it’s the child (not the AI agent) that initiates the interaction and decides whether and how to follow up.

The Scratch approach to AI is distinctive in that it can draw upon the huge collection of hundreds of millions of Scratch projects, created by children around the world. The Scratch creative learning assistant can provide provide links to projects that can serve as inspiration and offer code for remixing.

Unlike most approaches to AI in education, which focus on 1:1 interaction between the individual student and the AI system, the Scratch approach has a strong social component. The Scratch AI agent can help young people find other members of the Scratch community with similar interests, so that they can share ideas or collaborate — or, at least, gain inspiration from one another’s projects.

I like to think of the Scratch AI agent not as a “tutor” or “coach” (framings used by many other AI systems for kids) but rather as a new type of resource that young people can seek out when they want help. Currently, when Scratchers are looking for advice or inspiration, they might talk with a friend, or refer to a reference guide, or do an online search, or watch a YouTube video. Each of these resources plays a different role. An AI agent could be an additional resource, with its own advantages and limitations. AI becomes another resource for children to use as part of their creative process — but, importantly, it remains their creative process.

To ensure that their new AI explorations remain aligned with the human-centered 4C approach, the Scratch team developed a set of “guiding stars” and “lines in the sand.” I believe that these guidelines could be useful for all developers and educators as they consider the use of AI technologies in the educational process.

My ideas about human-centered society and human-centered education have been influenced by my collaboration with Carla Rinaldi, who sadly passed away earlier this year. For many years, Carla served as President of the Reggio Children Foundation, supporting the world-renowned pre-schools and kindergartens in Reggio Emilia, Italy. The Reggio schools have inspired educators around the world with the ways they organize their classrooms — and engage the entire community — to support children in developing their imagination, curiosity, and creativity.

Carla always emphasized that children in Reggio Emilia are “full citizens from birth.” In Reggio, children learn from the community, but the community also learns from children. Reggio Emilia is not just human-centered but child-centered.

For Carla, educational choices were always political choices. The first schools in Reggio Emilia opened in 1945, in direct reaction against the fascist government in Italy. The original pluralistic, pro-democracy ideals of the Reggio schools have continued to guide all aspects of the schools throughout the years. With the rise of authoritarianism around the world today, we can all learn important lessons from the child-centered and human-centered approach of Reggio Emilia.

One of the guiding visions of the Reggio approach is the idea of “the hundred languages of children.” This idea was beautifully described in a poem by Loris Malaguzzi, the founder of the Reggio schools (and Carla’s mentor). The poem starts:

The child has
a hundred languages
a hundred hands
a hundred thoughts
a hundred ways of thinking
of playing, of speaking

The poem captures the spirit of the Reggio schools, which are filled with a diverse array of materials and activities, designed to support children in exploring, creating, playing, and learning in a multitude of ways. Throughout the day, Reggio students have opportunities to explore their interests, develop their ideas, and express themselves creatively.

As my research group develops new technologies, like Scratch and OctoStudio, we are guided by this vision, wanting our technologies to support exploration and experimentation just like the diverse materials in Reggio classrooms. We evaluate the success of our technologies, activities, and workshops based on the diversity of projects that children create — that is, how well are we supporting the hundred languages of children.

Later in Malaguzzi’s poem, he issues a warning:

The child has
a hundred languages
(and a hundred hundred hundred more)
but they steal ninety-nine

Malaguzzi worried that traditional schools often “steal ninety-nine” of children’s hundred languages, forcing students to conform to pre-determined paths and goals. The technological and political disruptions of our times are accentuating this problem, further limiting children’s choices and voices.

A human-centered education is needed to push back against these pressures. We need to engage people from all parts of the educational ecosystem — teachers, parents, administrators, developer, policymakers — in a movement supporting human-centered education, to ensure that children can develop their hundred languages, and grow up as creative, curious, caring, and collaborative learners. That’s the best way to ensure a pluralistic, democratic, human-centered society.