What does it mean to assess learning in a world where AI can generate answers instantly?

In this episode of Education Futures, Svenia Busson is joined by Alina von Davier, Chief of Assessment at Duolingo, and Elie Bechara, who works on institutional partnerships for the Duolingo English Test.

Together, they bring a rare combination of assessment science, product innovation, and real-world university dynamics.

As AI tools become widely accessible, traditional forms of assessment, especially high-stakes exams, are being fundamentally challenged.

If students can generate answers with AI, what are we actually measuring?

In this conversation, we explore:

• Why AI is putting pressure on the validity of traditional exams

• How universities are responding to new questions around academic integrity and admissions

• The shift from testing knowledge → competencies and skills

• Why assessment needs to become more continuous, contextual, and embedded in learning

• How the Duolingo English Test is rethinking language assessment using AI

• The importance of designing assessments that reflect real-world performance, not artificial test conditions

We also discuss how assessment can evolve from a moment of evaluation into a tool for learning itself — providing feedback, guiding progress, and supporting long-term skill development.

The challenge ahead is about redefining what we value, what we measure, and what we trust in education.

To explore further what was discussed in the episode:

https://englishtest.duolingo.com/

https://blog.duolingo.com/video-call/

What should universities teach in the age of AI?

In this episode of Education Futures, Svenia Busson speaks with Art Markman, Chief Academic Officer at Minerva University and a leading expert in cognitive science, decision-making, and learning.

With a career spanning academia, research, and applied learning design, Art brings a powerful perspective on how education must evolve — not just to keep up with AI, but to remain relevant in a rapidly changing world.

In this episode, we explore:

• Why education should focus on “far transfer” — the ability to apply knowledge across contexts

• The shift from teaching content → developing durable skills (reasoning, communication, empathy, systems thinking)

• Why most universities are still structured like they were centuries ago

• How Minerva Project is redesigning higher education from first principles

• Why grades and transcripts fail to capture real learning

• How AI is reshaping the role of assignments, assessment, and feedback

• Why trying to ban AI in universities is the wrong approach

• How AI can be used as a tutor, feedback engine, and learning accelerator

To go further:

Minerva University: https://www.minerva.edu/

UT Austin’s Homegrown AI Tutor Platform: https://tech.utexas.edu/news/ut-austin-and-aws-launch-ut-sage-ut-austins-homegrown-ai-tutor-platform

What happens when students start trusting AI more than their teachers?

In this episode of Education Futures, Svenia Busson speaks with Mary Burns, researcher at the Brookings Institution and co-author of the report “A New Direction for Students in an AI World: Prosper, Prepare, Protect.”

With over 40 years of experience in education — from classroom teaching to advising ministries of education across more than 100 countries — Mary brings a rare, global perspective on how AI is reshaping learning systems.

In this episode, we explore:

• The “web of distrust” emerging in education systems

• Why students may begin to trust AI more than teachers

• The risks of cognitive, emotional, and social offloading to AI

• Why AI feels more reliable, neutral, and emotionally safe than humans

• The rise of teacher over-reliance on AI — an under-discussed risk

• Why AI literacy must be holistic (not just tool usage)

• The dangers of sycophantic AI systems and emotional attachment

• The growing gap between AI adoption and regulation

A key framework from the episode:

Mary co-authored a major report proposing a 3-pillar framework for AI in education:

👉 Prosper, Prepare, Protect

• Prosper: Use AI to improve learning and opportunity

• Prepare: Equip students, teachers, and systems to use AI responsibly

• Protect: Safeguard learners from risks (cognitive, emotional, data-related)

📄 Read the full report:

Brookings Institution – A New Direction for Students in an AI World

https://www.brookings.edu/wp-content/uploads/2026/01/A-New-Direction-for-Students-in-an-AI-World-FULL-REPORT.pdf

Distance education for teacher training: modes, models and methods - http://go.edc.org/07xd

Read "eyes wide open" the chapter Mary wrote in "Que educação nos exige – hoje − o porvir? Se não agora, quando. Universidade de Lisboa: Centro de investigação e de estudos em belas-artes, p. 50-69"

English version here: https://static1.squarespace.com/static/5fac2fdb0da84a28cc76b714/t/67b56f484fc8d51dd5546fa4/1739943752612/Mary+Burns-Eyes+Wide+Open+What+We+Lose+from+Generative+AI+in+Education.pdf

What if we asked children how to design the future of education?

In this special episode of Education Futures, Svenia Busson is joined by Selena Marwaha, an 11-year-old builder, coder, and speaker who has already presented at global events such as COP28, COP29, and WISE Summit Qatar.

She is joined by François Taddei, co-founder of the Learning Planet Institute, whose work focuses on collective intelligence, learning ecosystems, and the future of education.

Together, they explore what education could look like if it were co-designed with the next generation.

In this episode, we explore:

• Why children should be actively involved in shaping the future of education

• Selena’s vision for a school that prioritizes creativity, mentorship, and problem-solving

• Why AI should be used as a co-creator, not a shortcut

• The importance of art, curiosity, and human expression in a tech-driven world

• How to balance AI acceleration with human development

• Why schools should help students navigate emotions, relationships, and purpose

Selena also shares her project Planetary Stories — a platform where children from around the world can share their dreams, challenges, and visions for the future, creating a collective dataset of youth perspectives to inform decision-making.

Go further with these ressources:

• Selena's initiative "Planetary Stories" will be launched on April 22nd (Earth Day), we will share these here in the show notes.
• Mara Mintzer TED Talk "How kids can help design cities" https://www.ted.com/talks/mara_mintzer_how_kids_can_help_design_cities
• Alison Gopnik's work: research on child creativity and learning https://alisongopnik.com/
• Kiran Bir Sethi's work: creator of the Design for Change framework (“Feel, Imagine, Do, Share”) - https://dfcworld.org/SITE

Are we building educational technology faster than we can prove it actually works?

In this episode of Education Futures, Svenia Busson speaks with Natalia Kucirkova, professor at the University of Stavanger and co-founder of the International Centre for EdTech Impact.

Natalia has spent years researching how children learn, and how digital tools can (or cannot) support that process. Her work sits at the intersection of learning science, AI, and impact evaluation.

As AI tools rapidly enter classrooms, one question becomes critical:

Where is the evidence that these tools actually improve learning?

In this conversation, we explore:

• Why engagement is not the same as learning

• The risk of deploying AI tools without evidence of impact

• How EdTech companies can collaborate with researchers to design better products

• Why we need to slow down before scaling untested technologies

• The difference between efficacy, effectiveness, and real-world impact

• Why traditional evaluation methods (like RCTs) need to evolve in the age of AI

• How teachers and schools can make more informed, evidence-based choices

We also discuss concrete tools and initiatives aiming to bring more transparency to the field:

Natalia's Centre for Edtech Impact: https://foreduimpact.org/

AI safety benchmark Natalia has contributed to:

https://korabench.ai/

Edtech Certification Natalia is affiliated with & recommends: https://eduevidence.org/

Other Certifications that exist:

https://www.1edtech.org/

https://edtechimpact.com/

https://iste.org/edtech-product-selection

https://www.edtechtulna.org/

Natalia also explains how EdTech evaluation works in practice — from early-stage testing (A/B testing, rapid cycles) to large-scale studies like randomized controlled trials (RCTs).

The takeaway:

In education, good intentions and high engagement are not enough. If we want technology to truly support learning, we need to measure it well.

What should we protect in education, as AI transforms how we learn, think, and work?

In this episode of Education Futures, Svenia Busson speaks with Nick Krichevsky, a high school teacher and Head of Digitalization at the German International School Johannesburg in South Africa.

With over 15 years of teaching experience across Germany and South Africa, Nick brings a grounded, classroom-based perspective to one of the most pressing questions of our time:

What should education hold onto, and what should it let go of, in the age of AI?

In this conversation, we explore:

• Why AI should serve a clear educational purpose, not drive it

• The limits of a one-size-fits-all education system across different contexts

• What Nick calls the “promise — and misspoken promise — of education”

• Why many students still lack practical skills and opportunities, despite access to schooling

• The importance of effort, friction, and cognitive rigor in real learning

• How AI can support education by freeing up time and resources, not replacing thinking

• Why students must develop ownership of their learning process

• The growing concern among students about AI, self-efficacy, and their future

Nick also shares a powerful vision for the school of the future:

Not one centered on technology, but one built around human relationships, social learning, mentorship, and responsibility.

In this episode of Education Futures, Svenia Busson speaks with education entrepreneur Ed Fidoe, founder of London Interdisciplinary School and co-founder of School 21 in London.

Ed has spent the last decade building new kinds of educational institutions from the ground up. His work challenges one of the core assumptions of higher education: that students should specialize in a single discipline.

Instead, the London Interdisciplinary School is built around a radically different idea: teaching students to tackle complex real-world problems using knowledge from multiple disciplines.

In this conversation we explore:

• Why universities are facing a structural crisis

• Why the traditional single-discipline degree may no longer make sense

• How AI is reshaping what students need to learn

• Why universities should teach students how to identify the right problems to solve

• Why innovation in education requires experimentation and new institutions

We also discuss the Education Futures Master’s at LIS, a new interdisciplinary program designed for educators, learning leaders, and anyone interested in how education systems must evolve in an AI-driven world.

Ed also shares lessons from School 21, including the development of oracy education, teaching students how to think and communicate effectively through speech.

We mentioned:

https://www.lis.ac.uk/

https://school21.org.uk/

https://voice21.org/

In this episode, Svenia Busson speaks with Guila Clara Kessous, a UNESCO Artist for Peace and Harvard teacher, about the intersection of art, education, and peace-building. They explore how performing arts and literature can heal trauma and how "diplomatic entrepreneurship" can help the next generation navigate a world shaped by AI and social media.

Key Topics

• Guila’s Journey: From performing arts conservatory to a PhD with Nobel laureate Elie Wiesel.
• Art Therapy & Bibliotherapy: Understanding how drama and reading (recognized by the WHO) can increase serotonin and well-being.
• Education in the Age of AI: Why we should teach kids how to "prompt" like the game Jeopardy to find neutral information and overcome biases.
• Diplomatic Entrepreneurship: Using culture and art to reconcile communities in conflict, such as through Israeli-Palestinian youth summer camps.
• Non-Violent Communication (NVC): The importance of teaching children to "agree to disagree" and express views respectfully.
• The School of the Future: A vision of an open school in nature that connects the "heart, head, and body".

Resources Mentioned

• World Art Day International Forum: An annual event at UNESCO every April 15th focusing on arts, health, and activism. (Website: worldartdayforum.com)
• Faber and Mazlish Technique: Practical methodology for communication between adults and children ("How to Talk So Kids Will Listen").
• Femina Vox: An international forum at UNESCO dedicated to Women’s Rights.
• Movements for Peace: "Women Wage Peace" and "Women of the Sun," including the "Prayer of the Mothers" song.
• Bibliotherapy: Mention of the power of reading as recognized by the World Health Organization (WHO).

In this episode of Education Futures, we speak with Kristen DiCerbo, Chief Learning Officer at Khan Academy, where she leads the teams responsible for content, product design, assessment, and learning science.

With a PhD in educational psychology, Kristen brings a rare perspective to the AI conversation: learning science first, technology second.

We explore how Khan Academy is building Khanmigo, its AI tutor and teaching assistant, and what it takes to design AI tools that support real learning rather than shortcuts.

In this conversation, we discuss:

• Why screen time is a poor proxy for learning
• How AI tutors can support practice and feedback at scale
• Why foundational knowledge still matters in the age of AI
• The growing concern around cognitive offloading and students delegating their thinking to machines
• How Khan Academy designed guardrails and safety mechanisms for AI used by children
• The tension between gamification, motivation, and real learning
• Why human relationships with teachers remain the strongest driver of learning
• What the school of the future could look like, combining technology with project-based learning

Kristen also shares how Khan Academy applies a risk-management approach to responsible AI, identifying potential harms early and designing safeguards directly into their systems. (more on this: https://blog.khanacademy.org/khan-academys-framework-for-responsible-ai-in-education/)

The takeaway: AI may transform education, but learning will always require effort, curiosity, and human guidance.

Read this to go further:

National Study in Top Journal Finds Khan Academy Learning Gains After Accounting for Key Unmeasured Factors : https://blog.khanacademy.org/national-study-in-top-journal-finds-khan-academy-learning-gains-after-accounting-for-key-unmeasured-factors/

By 2050, Sub-Saharan Africa will be home to nearly 1 billion people under 18.

Today, 90% of 10-year-olds in the region cannot read a simple paragraph (according to the World Bank)

What happens when artificial intelligence accelerates, but foundational literacy remains out of reach for millions of children?

In this episode of Education Futures, Svenia Busson speaks with Paul Atherton, founder of Fab AI, about the future of AI in low- and middle-income countries, and whether it will close or widen the global learning gap.

Paul’s mission is to ensure that the world’s best technologies serve children who lack access to foundational literacy and quality schooling.

We explore:

• Why foundational literacy is the non-negotiable starting point
• How AI could help leapfrog infrastructure gaps in Sub-Saharan Africa
• Why most AI funding focuses on short-term pilots instead of long-term system architecture
• The risk that high-income countries experience exponential productivity gains while others fall further behind
• How rapid, decision-ready RCTs could modernize evidence in edtech
• The difference between AI as autopilot vs co-pilot in learning
• Why friction and effort remain essential to deep learning
• Why Paul worries more about today’s 15-year-olds than 5-year-olds

This conversation is about infrastructure, inequality, and the billion young people whose future will shape the global economy, and whose literacy will determine whether AI becomes a tool of opportunity or a force that widens the gap.

Try Fab AI's new web app which can help evaluate the quality of foundational literacy and numeracy materials for low- and middle-income countries: https://fab-content-curation.web.app/

Read the latest Fab-AI Research Paper: "Context counts: Measuring how AI reflects local realities in education" : https://www.fab-ai.org/initiatives/ai-for-education/edtech-quality/resources/research-paper/measuring-how-ai-reflects-local-realities

Subscribe to Paul's Substack: https://paulfabai.substack.com/