Launched at the Cambridge AI in Education Summit, the international initiative will examine how students’ capabilities develop through AI use and what they retain when support is removed.
ETIH Innovation Awards judge Tina Austin at the Cambridge AI in Education Summit 2026, where she launched the UnBlooms Longitudinal Research Consortium.
ETIH Innovation Awards judge Tina Austin has launched an international research consortium to investigate whether AI-supported learning produces lasting improvements in students’ critical thinking, metacognition, and independent judgment.
The Longitudinal Research Consortium was publicly launched at the Cambridge AI in Education Summit in October 2026. According to Austin, the initiative has attracted interest from more than 40 institutions internationally, bringing attention to a question facing educators as generative AI becomes embedded in teaching and assessment.
Austin is Project Director of Public Interest Technology Pathways at the University of California, Los Angeles (UCLA), and specializes in responsible AI and assessment redesign. A judge for the ETIH Innovation Awards 2026, she was also recently recognized as a LinkedIn Top Voice in AI and education.
Her research focuses on how institutions can establish whether students are developing capabilities they can apply independently, beyond the work they produce with AI assistance.
Measuring learning after AI support is removed
Through her research analysis on Substack, Austin examines the evidence behind claims about AI’s effects on education. She argues that studies reporting both benefits and harms can leave important gaps in understanding how learning develops over time.
One concern involves studies suggesting that AI harms learning without first establishing students’ independent capabilities or providing structured preparation. Austin argues that students may have limited opportunities to develop the skills needed to evaluate AI critically before researchers assess their performance.
Studies reporting benefits can present a different problem. They frequently measure performance while students have access to AI tutors or immediately afterward, leaving questions about whether improvements persist days, weeks, or months after support is removed. Even research covering longer periods may rely on narrow measures, such as homework performance.
Austin also highlights the challenge of comparing studies of AI scaffolding, the structured support used to help students complete a task. Differences in how interventions are designed, delivered, and sustained can make findings difficult to reproduce or apply across disciplines.
Research may also use different definitions of AI use and learning, alongside inconsistent outcome measures, limited starting-point data, or inadequate comparison groups. Austin argues that these methodological gaps can support dramatic claims about cognitive decline or educational transformation without sufficient evidence of lasting change.
Building evidence across disciplines
The consortium builds on Austin’s UnBlooms framework, which she presented at OpenAI in 2025. The framework rethinks learning and assessment around students’ reasoning before, during, and after AI use.
As part of this work, Austin developed the Metacognitive Awareness Scale (MAS), described in a peer-reviewed workshop paper. Metacognition refers to students’ ability to monitor and evaluate their own thinking, including recognizing when they need to question an answer or reconsider an approach.
The scale examines evidence of that thinking in students’ work, including how they verify information, explain their decisions, and revise AI-assisted outputs. Its focus extends beyond whether a finished assignment is accurate to the reasoning students demonstrate while completing it.
The paper reports preliminary shifts from accepting AI outputs with limited scrutiny toward more active verification and self-critique. Austin also reports promising early experience with her own students and pilot implementations at other universities, while acknowledging that further validation and longer-term evidence are needed.
The consortium will investigate how these capabilities develop through repeated AI use across different subjects and educational settings. Examining what happens when the technology is removed will form part of its work to understand whether students retain their learning and can apply it independently.
For educators and institutional leaders, the findings could help inform decisions about which AI-supported teaching strategies produce lasting learning and how assessments should change. They could also help institutions distinguish students’ own understanding from improvements in the work they produce with assistance.
Austin also leads an Assessment Redesign team within IEEE’s responsible AI in education initiative, connecting her research with practical approaches to teaching, assessment, and responsible adoption.
Further information is available through the consortium website. Institutions interested in learning more can contact Austin at tinaaustin@ucla.edu.
ETIH Innovation Awards 2027 open for entries
The ETIH Innovation Awards 2027 are now open for entries, with 26 categories recognising companies, institutions, projects and individuals making a measurable impact across education technology. The global awards span AI in education, digital learning, higher education, skills and workforce, inclusion, education operations, evidence and impact, with entries managed through ETIH’s new online awards platform.
Entries are open worldwide, with the early-bird deadline on Friday 18 December 2026 and the final entry deadline on Friday 26 February 2027 at 23:59 UK time.
Explore the 2027 categories and judging criteria, read the How to Enter guide or start an entry through the ETIH Innovation Awards 2027 platform.
