# AI+Education Summit 2026: Creativity in the Age of AI – Catalyst or Hindrance? Source: https://www.youtube.com/watch?v=wp3Jd-c1GSI Recap page: https://rapidrecap.app/video/wp3Jd-c1GSI Generated: 2026-02-19T18:42:25.482+00:00 --- ## Quick Overview The panel discussion concluded that while Generative AI (GenAI) can act as a powerful catalyst for creativity and accelerate learning by lowering the barrier to entry for complex tasks, it requires careful scaffolding and new curricular approaches to prevent negative transfer effects, such as reduced creativity and false confidence, which were observed in studies involving middle school students. **Key Points:** - GenAI is currently perceived as a tool, but should be treated as a curricular topic where facility grows as students progress through education. - A study on middle school creativity showed negative transfer in creativity tasks when using AI assistance, meaning students who used AI scored worse on subsequent unassisted creative tasks. - The fundamental assumption that a strong product indicates a strong learning process is challenged because GenAI can produce strong outputs without the user engaging in a strong learning process (e.g., writing code without understanding security). - The speaker (Chris Chafe) suggests that AI can elevate the ceiling of creativity by allowing learners to tackle more complex problems, such as those involving music composition or physics simulations, that were previously inaccessible. - The study presented by Guilherme Lichand found that students assisted by AI showed lower performance on the Divergent Association Task (DAT) compared to the control group, indicating a negative impact on academic self-concept. - The key takeaway is that safeguards are currently insufficient, and educators must rethink curriculum design to focus on process-oriented learning rather than just product assessment when using generative tools. ![Screenshot at 26:56: The slide summarizing the study's key findings, explicitly stating 'Negative transfer in creativity tasks' and concluding 'Safeguards were not enough!', which emphasizes the core finding that raw assistance can impede learning.](https://ss.rapidrecap.app/screens/wp3Jd-c1GSI/00-26-56.jpg) **Context:** This video captures a panel discussion from the 2026 AI+Education Summit focusing on whether Artificial Intelligence acts as a catalyst or a hindrance to creativity in education. The panelists, including Chris Chafe (Music), Mehran Sahami (Computer Science/Engineering), Guilherme Lichand (Education), and Hari Subramonyam (Education), debated the role of GenAI in learning. The discussion pivoted on the distinction between learning as a process versus evaluating learning solely on the product, using examples from music composition, physics simulation tools, and a formal study on middle school creativity. ## Detailed Analysis The panel addressed the role of Generative AI (GenAI) in creativity and education, debating whether it acts as a catalyst or hindrance. Chris Chafe highlighted that AI can lower the barrier to entry for complex creative tasks, such as in music composition or physics, by automating difficult technical processes, thereby allowing learners to focus on conceptual blending and creative expression. Mehran Sahami reinforced this by noting that large portions of computer science code bases are now written by AI, which is productive but risks creating a false sense of security and bypassing deep learning, leading to security vulnerabilities in the final product. Guilherme Lichand presented evidence from a study titled 'GenAI Can Harm Learning Despite Guardrails: Evidence from Middle-School Creativity,' which showed that students assisted by AI demonstrated 'negative transfer' in creativity tasks (Alternative Uses Task) and experienced a negative impact on their academic self-concept (Divergent Association Task). The study indicated that while assistance improved performance on the immediate task, it hindered subsequent unassisted performance and lowered self-concept scores. The core challenge identified is the educational system's inherent focus on evaluating the product rather than the learning process, a flaw exacerbated when AI can easily generate high-quality outputs without deep student engagement. Hari Subramonyam concluded by suggesting that AI should be viewed as a tool to foster creativity and learning by creating simulations, rather than just a consumption device, and that educators need to develop scaffolding models that promote genuine learning over mere efficiency. ### Panel Introduction - Brian Johnsrud introduces the panel members: Chris Chafe, Guilherme Lichand, Mehran Sahami, and Hari Subramonyam. ### Product vs. Process (Mehran Sahami) - Learning is a process, but education often evaluates it via a product, a fundamental assumption broken by GenAI's ability to produce strong products without deep learning. ### AI as an Aid to Spur Creativity (Brian Johnsrud) - Cites New York Times article questioning if AI can generate new ideas; notes AI can suggest hypotheses previously unconsidered, showing it can be a true catalyst. ### Curricular Orientation - Topics vs. Tools (Mehran Sahami): Argues GenAI should be treated as a curricular topic to build facility, not just a tool decided upon by individual instructors. ### Study Findings (Guilherme Lichand) - Presents data showing negative transfer in creativity tasks (AUT) and negative impact on academic self-concept (DAT) when students use AI assistance, emphasizing that shortcuts bypass necessary cognitive processes. ### Conclusion - AI should support expressive creation and purposeful learning, not just efficient consumption, requiring new scaffolding models to avoid negative impacts. ![Screenshot at 0:01: Title slide for the '2026 AI+Education Summit' featuring the theme 'Creativity in the Age of AI: Catalyst or Hindrance?' and listing the panelists.](https://ss.rapidrecap.app/screens/wp3Jd-c1GSI/00-00-01.jpg) ![Screenshot at 0:05: Brian Johnsrud opening the discussion, noting that creativity is having a moment.](https://ss.rapidrecap.app/screens/wp3Jd-c1GSI/00-00-05.jpg) ![Screenshot at 1:56: Chris Chafe speaking about his work in computer music and acoustics, contrasting the complexity of pre-AI programming with current capabilities.](https://ss.rapidrecap.app/screens/wp3Jd-c1GSI/00-01-56.jpg) ![Screenshot at 12:24: Slide illustrating the process of learning through creation, showing a progression from simple shapes to complex structural diagrams, with the caption: 'Creation is where the knowledge becomes structured and usable!'](https://ss.rapidrecap.app/screens/wp3Jd-c1GSI/00-12-24.jpg) ![Screenshot at 26:36: Slide detailing the structure of a study comparing unassisted vs. assisted learning on creativity tasks \(AUT/DAT\), showing the assisted group performing worse on the DAT transfer task \(Panel C\).](https://ss.rapidrecap.app/screens/wp3Jd-c1GSI/00-26-36.jpg)