Source-linked AI summary
With a Thermomix You Lose the Ability to Cook: A Kitchen Machine Analogy for Applications of Generative AI in Education
Nikol Rummel, Valentina Nachtigall, Ernesto Panadero
TL;DR
The paper organizes possible uses of generative AI through established frameworks such as ICAP and presents a comparative perspective on Thermomix and AI use. It argues that the central issue is how tools shape practices and what kind of learners users become, while cautioning that media-comparison designs risk confounding numerous factors.
Problem
The paper asks whether evaluation should focus solely on whether the dish was made.
Method
The paper organizes possible uses of generative AI through established frameworks such as ICAP and compares Thermomix and generative AI scenarios.
Results
The paper concludes that the central issue is how tools shape practices and what kind of learners and cooks users become through their use.
Takeaways & Limitations
Following instructions step-by-step is less likely to improve cooking abilities, highlighting the importance of how learners use tools.
Takeaways & Limitations
Media-comparison designs risk confounding numerous factors and yield limited insight.
Abstract
from arXiv · showhide
The rapid adoption of generative AI tools such as ChatGPT has sparked intense debate about their risks and opportunities for education, as well as the ways researchers should investigate them. In this paper, we approach these discussions through an analogy with the Thermomix, a smart kitchen appliance that has similarly provoked both enthusiasm and critique. By mapping Thermomix use cases onto examples of learning with generative AI, and situating them within the ICAP and SAMR frameworks, we show how different modes of tool use can either support or undermine meaningful engagement and learning. The Thermomix metaphor underscores that the central question is not whether learners employ AI, but how such use shapes their learning processes. In doing so, we provide a conceptual lens for researchers and practitioners to critically examine - and more effectively guide - the integration of generative AI into educational practice.
Introduction
The paper uses the Thermomix as a cultural and technological analogy for examining how generative AI reshapes learning. It emphasizes a continuum from unaided work to full delegation, with tool use potentially supporting or undermining meaningful engagement.
- Continuum of use: The paper rejects a binary choice between doing all work unaided and relying entirely on AI, identifying intermediate uses that support rather than replace meaningful learning.The proposed continuum includes tools that assist activity without replacing the learner.
- Analytical frameworks: The authors illustrate this continuum through generative-AI learning scenarios modeled on kitchen-machine practices and organized using ICAP and SAMR.ICAP distinguishes Passive, Active, Constructive, and Interactive engagement, while SAMR ranges from Substitution to Redefinition.
- Analytical rationale: The cooking analogy is suited to learning with generative AI because cooking involves skill, knowledge, active participation, well-being, and social connection.The paper presents this multifaceted activity as emotionally resonant and therefore useful for examining learning processes.
- Cultural lens: Debates about whether guided cooking enables or erodes culinary competence mirror educational concerns about generative AI, including efficiency, access, authenticity, and skill loss.The analogy is presented as a cultural lens for examining how technologies reshape practices and identities.
- Technological analogy: The Thermomix combines functions formerly distributed across multiple kitchen tools, while generative AI similarly merges search, grammar checking, translation, and note-taking.This convergence reconfigures established practices rather than merely accelerating them.
Four scenarios of Thermomix and generative AI use
The paper develops four Thermomix use scenarios and pairs them with corresponding uses of generative AI in education. Table 1 maps these cases onto ICAP engagement modes and SAMR technology-integration levels.
- Scenario framework: The four scenarios are designed to show that technology can both support and undermine skill development.Each Thermomix practice is juxtaposed with an educational AI use case.
- Scenario framework: Table 1 provides a comparative overview of Thermomix practices, generative-AI uses, and their alignment with the ICAP and SAMR frameworks.The subsequent sections elaborate each scenario with examples and evidence.
- Guided cooking: Guided cooking lets users prepare meals by selecting recipes and following device instructions, potentially enabling cooking without particular skills.The same functionality may prevent skill acquisition or lead users to be viewed as assistants rather than cooks.
- Guided cooking: Some users value guided cooking because it helps people who lack cooking skills or confidence prepare meals independently.The section cites working mothers and online discussions describing the Thermomix as foolproof for otherwise limited cooks.
- Guided cooking: Public reactions to guided cooking remain mixed, combining appreciation of accessibility with concern that reliance may erode cooking skills.Forum comments contrast appliance-enabled independence with hopes that future generations will still cook using pans and utensils.
Generative AI
The paper contrasts AI use that cognitively offloads learning with use that preserves or reallocates meaningful activity. Through ICAP and SAMR, it links full delegation to low engagement while identifying conditional opportunities for inclusion and deeper work.
- Risks of delegation: Novice writers who received AI-produced texts revised them only superficially and invested less cognitive effort than when revising their own drafts.This finding illustrates how accepting AI output can reduce engagement with revision.
- Risks of delegation: Full delegation of assignments to generative AI can diminish reasoning, creativity, revision, and subject learning by removing students’ personal input.The paper connects this pattern to passive engagement and cognitive offloading.
- Passive use and substitution: Guided cooking and AI-generated assignment completion exemplify SAMR Substitution because technology replaces manual or cognitive processes without requiring active engagement.The paper characterizes this use as aligned with passive ICAP engagement.
- Passive use and substitution: AI substitution may promote inclusion and participation for students who struggle with assignments because of special needs or language barriers.The paper presents this as a possible benefit despite the expectation that substitution will not foster high cognitive engagement.
- Reallocating effort: Delegating low-level tasks can free time for searching, reading, synthesizing literature, and critically revising drafts.This opportunity depends on prioritizing more demanding activities after substitution.
- Conditions and trade-offs: Productive engagement with AI-generated material requires prior knowledge, while reliance on basic functions risks diminishing foundational skills.Learners need enough knowledge to evaluate and use generated material critically and creatively.
#2 “As a cook I can alter or add ingredients to some extent”
A more active Thermomix user adapts guided recipes to personal preferences and abilities rather than following instructions mechanically. The analogous AI use involves refining inputs, evaluating outputs, and drawing on prior knowledge.
- Active adaptation: Some Thermomix users treat guided instructions as a flexible framework by skipping, repeating, or modifying steps.The paper notes that users can adapt recipes, although alterations and truly original recipes remain limited.
- Active adaptation: Thermomix users can alter or add ingredients to some extent, but the appliance limits the use of fully personal recipes.This scenario therefore combines user agency with technological constraints.
- Critical AI use: In learning, students can cross-verify AI information with credible sources and adapt or elaborate generated outputs.These practices require understanding AI limitations, identifying credible sources, and comparing information across references.
- Critical AI use: Students may refine or clarify prompts when dissatisfied with AI output, using prior knowledge to seek higher-quality results.Prompt refinement is presented as another way to remain actively involved in the learning process.
- Evidence on active use: More experienced academic writers revise AI-generated text more thoroughly, while prompt-engineering practice is associated with higher self-efficacy and greater knowledge of AI concepts.Students also preferred feedback produced through structured prompt techniques over feedback from ad-hoc queries.
- Evidence on active use: The paper argues that refining inputs and critically evaluating outputs can make generative-AI engagement more active and productive.The cooking analogy connects this pattern to cooks adapting recipes according to their skills and preferences.
Active use and augmentation
Meaningful technology use depends on active engagement rather than passive acceptance. In learning, generative AI can support augmentation when learners critically work with its outputs and draw on relevant knowledge while recognizing its limitations.
- ICAP: In the ICAP framework, active learning behaviors involve focused manipulation of learning materials and are assumed to demand greater cognitive engagement than passive behaviors.Examples include note-taking, underlining, and pausing or rewinding videos.
- Active use: Learners can actively engage with AI outputs by cross-verifying information, refining prompts, and revising text.These activities require and foster active engagement rather than passive acceptance.
- SAMR: In the SAMR augmentation stage, technology enhances existing activities without necessarily redesigning the underlying task.The passage connects technology-assisted cooking and learning with augmented manual and cognitive processes.
- Analogy: Thermomix users may gain freedom to modify recipes when the appliance takes over basic cooking processes, enabling greater creativity.The analogy links this dynamic to AI-assisted learning, where automation can free attention for revision and verification.
- Augmentation: Relevant prior knowledge and awareness of AI limitations support learners’ use of generative AI for augmented activities.Examples include revising self-generated text and cross-verifying textbook information.
#3 “I have become a more creative and versatile cook”
The Thermomix is presented as supporting creativity and versatility by making it easier to discover, adapt, and experiment with recipes. The paper also notes possible benefits for healthier eating, while the supporting health evidence is limited.
- Versatility: Users and professional cooks employ the Thermomix for varied recipes and complex dish components, supporting its association with versatility.The cited examples include manual functions and preparation of challenging dishes.
- Creativity: Recipe search by event, meal type, or dietary attributes helps users discover and experiment with new dishes more easily.The database supports flexible searches, including occasions, meal categories, and vegetarian attributes.
- Health: Reports describe users trying unfamiliar ingredients, developing recipes, and preparing more balanced or healthier meals with the Thermomix.The health-related evidence includes user reports, a product test, and a small single-participant intervention.
- Conclusion: The paper concludes that a smart kitchen appliance can influence cooking habits toward creativity, versatility, and potentially healthier eating.The reported health evidence includes a non-peer-reviewed source and a single-participant pilot study.
Generative AI
Generative AI can support learning when students use its outputs to develop ideas, evaluate information, and reflect rather than simply consume answers. In redesigned assignments, AI can transform information-search tasks into constructive activities involving meaning-making and critical thinking.
- Generative AI: ChatGPT can support knowledge construction by helping learners brainstorm ideas, create outlines, and evaluate or reflect on information.Teachers would need to adapt assignments to encourage these uses.
- Generative AI: Used critically, AI becomes a catalyst for creativity and reflection rather than a source of ready-made answers.The paper links this use with developing and testing learners’ own ideas.
- Modification: In SAMR’s modification stage, technology significantly redesigns tasks instead of merely substituting for or augmenting existing processes.Thermomix examples include experimenting with complex dishes that traditional tools would make difficult or time-consuming.
- Constructive engagement: Constructive engagement involves generating outputs beyond provided materials, such as original ideas, independent texts, and critical personal responses.The paper compares this mode with manual Thermomix use for creating new and complex dishes.
- Modification: AI-supported assignments can prompt learners to evaluate and reflect on AI-provided information and draw their own conclusions.This restructures searching and synthesis into activities centered on meaning-making, critical thinking, and self-reflection.
- Modification: The modified learning activity may require AI because completing the equivalent sequence without it would take more time and multiple steps.The paper frames this as a task that could not be completed in the same form without the technology.
#4 “A new way of cooking”
Thermomix communities show how smart technologies can embed interactive exchange and feedback in established practices. The paper transfers this logic to generative AI, where conversational interaction can support debate, questioning, and real-time refinement of understanding.
- Thermomix community: Thermomix apps and recipe databases support a user community that shares recipes, tests them, and suggests adaptations.The community also discusses cooking challenges and provides answers in forums.
- Thermomix community: The Thermomix case illustrates how smart technologies can generate interactivity and feedback within an established practice.Cooking provides the valued context in which community participation and knowledge exchange occur.
- Generative AI: ChatGPT’s interactivity and real-time feedback are presented as supporting effective, personalized, and adaptive learning.It may also act as a collaboration partner in individual and collaborative knowledge construction.
- Generative AI: Conversational AI situates learners in exchanges where understanding is negotiated and refined in real time.The paper connects this interactional role with debate, critical questioning, and co-construction of ideas.
- Generative AI: Generative AI chatbots can serve as dialogue partners for creative problem solving and Socratic questioning.These uses move beyond passive or merely active engagement toward ongoing conversation.
Interactive use and redefinition
Interactive tool use can promote complex cognitive engagement and redefine cooking and learning practices. Across the scenarios, technology ranges from passive substitution to interactive redefinition, expanding or constraining human agency depending on use.
- Frameworks: These uses align with interactive cognitive engagement in ICAP and task redefinition in SAMR, linking tool use to effective cooking and learning processes and outcomes.
- Interactive engagement: Interactive activities involve constructive dialogues, such as debating positions and asking or answering questions, and are identified as the most effective mode for promoting learning.
- Interactive engagement: Discussing recipes in Thermomix communities and interacting with ChatGPT about independently developed ideas exemplify complex interactive cognitive engagement.
- Practice redefinition: The Thermomix community transforms cooking by supporting recipe development, experimentation, feedback, and sustained discussion beyond traditional cooking contexts.
- Practice redefinition: Generative AI similarly redefines learning by providing responsive feedback and reflective stimulation, making individualized and adaptive learning experiences feasible.
- Spectrum of use: The four scenarios progress from passive substitution to interactive redefinition, showing that technology can either constrain or expand human agency.
What makes a cook? Reflections on identity, agency, and evaluation
The analogy questions whether using external tools changes what counts as cooking or learning, and whether evaluation should address only outcomes or also the processes producing them. It frames competence as potentially including the skilled mobilization, evaluation, and adaptation of external resources.
- External resources: Knowing how to find, interpret, and adapt a recipe may be as much a marker of culinary expertise as retaining it in memory.
- External resources: The corresponding learning question is whether authentic competence can include mobilizing generative AI and other external resources to solve problems and construct understanding.
- Identity and competence: The analogy raises whether reliance on external devices shifts learners from making knowledge to managing information that must be understood, evaluated, and organized.
- Agency and creativity: Generative AI may threaten originality or, when used skillfully, support deeper engagement and more ambitious outcomes; the key issue is how use reflects agency.
- Evaluation: Because teachers and examiners often see final products without knowing AI’s contribution, generative AI raises questions about authorship, assessment, and trust.
- Evaluation: Educators face whether assessment should focus solely on final outcomes or also on the processes through which those outcomes were achieved.
Conclusion
The Thermomix analogy offers a conceptual lens for examining generative AI through ICAP and SAMR while moving discussion beyond polarized media-comparison debates. Its central conclusion is that researchers and educators should examine how tools shape learning processes, identities, agency, and evaluation.
- Purpose: The paper uses the Thermomix as a comparative perspective to contribute a distinct angle to debates about generative AI’s risks and opportunities in education.
- Research lens: The analogy organizes possible generative-AI uses through the established ICAP and SAMR frameworks, providing researchers with a conceptual lens.
- Research priorities: The authors argue that inquiry should move beyond asking whether students use AI toward understanding the specific conditions under which it fosters or hinders productive learning processes.
- Research priorities: Media-comparison designs may confound numerous factors and provide limited insight, motivating research that disentangles when, how, and for whom generative AI enhances learning.
- Practical reflection: For educators, the metaphor supplies language for discussing AI’s technical functions alongside identity, agency, evaluation, and the kinds of learners students become.
- Core conclusion: The paper concludes that the central issue is how technologies reconfigure the practices and identities of people who use them, not merely whether tools are used.