Christine Baker
Artificial intelligence can summarize text, find bugs in code, and create images. It can even record and summarize panels at a conference.* New tools and use cases are being developed as we speak, each model more powerful than the last. What does this reality mean for teachers and learners?
In its third instance, the AI+Education Summit hosted by the Stanford Accelerator for Learning and the Stanford Institute for Human-Centered Artificial Intelligence (HAI) brought together researchers, educators, tech developers, and policymakers for pivotal conversations on how to shape a thriving learning ecosystem with human-centered AI technologies. The convening showcased cutting-edge, research-based applications of AI in learning and facilitated dialogue about how to ensure AI serves education ethically, responsibly, and equitably.
The summit was guided by big questions: What is the future of learning we aspire to? What makes us uniquely human? How do we center humanity in our learning and AI ecosystems? How can research best support a bright future of learning?
The conference featured four expert panels and two rounds of speed talks followed by small group discussions for attendees across sectors to share ideas, opportunities, and challenges. During the lunch hour, AI and education seed grant recipients funded by the Accelerator and HAI shared posters with their in-progress research projects. The Accelerator’s AI Tinkery hosted a pop-up and the Generative AI for Education Hub presented a demo of its new research repository, a tool for education leaders to explore the latest studies on AI and education.
Following are some insights from the summit.
How AI is Transforming Education
Victor Lee, associate professor of education at Stanford and faculty lead for AI+Education at the Accelerator, raised the question of what it means to be "AI literate,” and the implications for schools.
Lee has studied how high school students are currently using AI, such as to check grammar and support group work. He also surveyed teachers on their professional development needs related to AI, finding that teachers want to understand how to use AI to teach, how to teach about AI, and how AI actually works. With emerging legislation around AI literacy, such as a new California bill requiring it to be incorporated into curriculum, Lee stressed the need to develop a common language and framework, encompassing user, developer, and critic perspectives.
“This is going to be an active conversation. We need to figure out what everyone in school should know about AI, and I would encourage that we figure it out soon,” Lee said.
Michael Frank, Benjamin Scott Crocker professor of human biology at Stanford, is a psychologist and cognitive scientist who develops language models to better understand how children learn and develop. His BabyView project uses head cameras to collect data on children's language input and processing, and then uses the data to train and evaluate AI models.
Frank’s work is expanding to collect information on child development globally through a project called the Learning Variability Network Exchange (LEVANTE), a framework for collecting open, linguistically and culturally adaptable data that can be then used to further hone the tool.
"We can use AI models as our scientific models to push forward the science of child development, but we can only do that if we have data that represent children's input and broad, diverse data sets that help us evaluate those lines,” Frank said.
Emma Brunskill, associate professor of computer science and a faculty affiliate of the Accelerator, discussed AI’s potential to accelerate the pace of education research. Brunskill and her team are testing ways that AI could help speed up innovation, which often takes decades, by simulating and optimizing educational interventions and evaluating them with human experts.
Responsible Use and Ethical Dilemmas
While AI has transformative potential for learning, the day’s conversations emphasized the need for thoughtful design and ethical considerations.
Rob Reich, McGregor-Girand professor of social ethics of science and technology, highlighted the tension between using AI for automation, replacing human skills, versus augmentation, facilitating and enhancing human abilities. He argued that developing AI tools with the goal of mimicking humans is outdated and misguided, and instead, the focus should be on how AI can amplify and cultivate human intelligence.
“I view this as basically a design challenge,” said Reich.
Erin Mote, chief executive officer at InnovateEDU, leads work focused on safety, privacy, security, and transparency in AI systems. When it comes to AI and education, she has found that privacy is the top concern for policymakers, parents, and communities. Mote emphasized the need for clarity and accountability on whose job it is to protect student data and privacy, as well as an intentional focus on mitigating bias in AI tools for education.
State and District Leaders at the Helm
Education policy leaders from New York, North Carolina, and Washington shared how they are approaching the AI revolution.
Tara Carrozza, director of digital learning initiatives for New York City Public Schools, the largest school district in the United States, shared her district’s work leading professional development for 10,000 staff members, creating a K-12 AI Policy Lab, and building partnerships with research institutions, philanthropists, and the technology sector. Kris Hagel, chief information officer at the Peninsula School District in Gig Harbor, Washington, shared his district's proactive approach to establish an AI action research team, collaborations with other districts, and partnerships with universities.
Catherine Truitt, former North Carolina superintendent of public instruction, emphasized the need for state-level guidance on AI use, noting that only 26 states have issued such guidance. Without it, she said, teachers often lack the tools to navigate its presence in their classrooms and schools or districts may issue blanket bans, presenting an equity issue. Higher education and teacher preparation programs play a key role in integrating AI into education, too, she said.
Where Are We Headed?
Researchers and technologists presented emerging areas of exploration for AI and learning, including collaborative classroom tools, virtual reality, and epigenetics, the study of how environmental factors change how genes are expressed. Across sessions, speakers emphasized that AI should support, but not replace, human relationships, and equity must be at the forefront when it comes to the design of and access to AI tools. To these ends, participants largely agreed that teachers, students, parents, and policymakers must be involved in the design and implementation processes of new technology.
“Education is the most important sector in our country,” said Carrozza. “If we want to be globally competitive, we really need to engage every community member in this process. All you wonderful people in this room: how do we work better together? Because we have to orchestrate how we're doing this together. We need to take co-design as a policy, not as a nice to have.”
Mote, speaking from her perspective as a parent and school founder, acknowledged that AI is a rapidly evolving frontier.
“I think it's inevitable that we have to help young people think about how to navigate what we don't know yet,” she said. “What are we doing to actually build resilience and the ability to pivot? Because if there's anything I'm certain as a mom of two kids under the age of 10, gosh, are they going to have even way more pivots than I did.”
*Generative AI was used to record, transcribe, and summarize event sessions. The author fact-checked and edited these summaries for use in this story.
Photos by Christine Baker