First published in response to the UAE’s 2025 national AI education policy, this post is part of a series on how schools can rethink lesson design to reduce AI misuse, maintain academic integrity, and support purposeful AI use where appropriate. These ideas apply to educators globally, regardless of whether national policy is already in place.
When the Debate Was Pre-Written
A Year 8 student delivers a flawless spoken argument in a paired debate. It’s articulate and confidently delivered. But something doesn’t feel right. They don’t respond to challenges. Their tone is stiff. The cue cards are overly polished. A quick check confirms it: an AI chatbot drafted the speech the night before.
This isn’t a case of cheating out of malice. It’s a sign that our lesson design hasn’t caught up. Tasks that once demanded personal thinking are now at risk of being quietly outsourced. The solution isn’t to ban the tools. It’s to redesign the task so that students must still think, respond, and reflect in real time.
Why Oral Tasks Are Now Vulnerable
Speaking and discussion-based activities have long been seen as reliable opportunities to hear authentic student thinking. They allow learners to demonstrate understanding in their own words, respond to questions in real time, and show ownership of their ideas. But the rise of AI has started to compromise the originality of these tasks. With a few well-phrased prompts, students can now generate highly polished verbal responses outside the classroom. These scripts may sound impressive but often lack the spontaneity, depth, or personal voice that oral work is designed to develop.
That does not mean we should abandon speaking tasks altogether. It means we need to examine why they are increasingly vulnerable and consider how to make their design more robust.
AI can script entire interviews, monologues or debates - Students may arrive to class with a pre-written piece that sounds natural but was generated the night before. It’s harder to detect than copied writing because delivery can mask the lack of ownership.
Students can rehearse and memorise AI-generated answers at home - Rather than formulating their own views, students may simply practise reciting AI content that aligns with the task, polished but not personal.
Verbal work risks becoming polished performance rather than evidence of thinking - If the focus is only on delivery and not on how ideas are formed or challenged, students may use AI to bypass the most valuable parts of the task.
It’s also important to note that AI detection tools are still highly limited for spoken work. Schools should avoid overreliance on tech-based monitoring and instead prioritise strong task design and formative checking.
Structuring Oral Tasks for Authenticity
The goal is not to make every element unpredictable or high-stakes. It is to build in opportunities for students to demonstrate ownership of their thinking in ways that cannot be easily replicated or rehearsed in advance. Oral tasks should give space for students to prepare thoughtfully, but also require them to respond and adapt in the moment. This blend of structure and spontaneity can help ensure verbal work remains a valuable window into real learning.
As new multimodal tools emerge, such as AI systems that generate full video responses from typed prompts, it becomes even more important to audit and update your speaking and presentation tasks regularly. At least once per year, consider whether the current formats still invite real student thinking or if they’ve become too easily automated.
🧾 AI Tools and Oral Task Risks
Voice cloning and prompt-to-video tools are becoming more accessible, raising ethical concerns around deepfakes and impersonation. Schools should review safeguarding policies to ensure student voice data is not used inappropriately or without consent.
Consider using the following design adjustments to support thinking and originality:
Use AI for early planning, with structure and purpose. Allow students to use AI to support the early stages of preparation. This could include generating questions, listing arguments, or exploring different viewpoints. This support can take place at home as part of a flipped learning model or during a clearly defined time in class. For example:
A student uses AI at home to help generate follow-up questions for a peer interview task, then selects the ones they find most useful.
In class, a group of students use a chatbot to simulate a debate on a topic, identify which arguments are most convincing, and plan their own response.
A teacher allows students ten minutes to use AI to brainstorm responses, but then collects the output before the final discussion begins.
Introduce live, unscripted elements that cannot be rehearsed. Incorporating moments that demand flexible, real-time thinking makes it harder for students to rely entirely on AI-generated material. You might:
Set up a follow-up round of peer questions after a planned speech.
Introduce a new visual, quote, or graph right before discussion begins.
Ask students to directly address or challenge the previous speaker’s point of view.
Encourage students to ask each other real-time questions such as: “Can you give an example from our last unit?” “How would you counter an opposite view?” Offer sentence starters like: “I’d like to build on…” or “A counterpoint might be…” to support less confident speakers in engaging more deeply.
Make student thinking visible by prompting reflection. Ask students to explain what they changed from their AI-assisted planning, why they made those choices, and how their perspective evolved. This reinforces the expectation that AI is a thinking partner, not the author of their ideas.
Assess for clarity of reasoning rather than surface polish. When reviewing verbal responses, focus on indicators of genuine engagement and thought. Rubrics should reward:
Students who elaborate, clarify, or adapt their response.
Those who show they can justify their viewpoint and react to challenge.
Signs that the language and phrasing align with the student’s usual communication style.
These strategies can also help scaffold oral language development for EAL learners. By offering sentence stems, vocabulary banks, and structured speaking frames, you create inclusive opportunities for all students to participate in real-time discussion confidently.
✔️ Perfect phrasing with little interaction - If a student delivers a highly polished verbal performance but avoids engaging with peers or adapting their ideas during the task, it may indicate heavy reliance on AI-generated scripts. Look for signs of surface-level polish without genuine thought.
✔️ Cue cards or notes that are identical or too refined - When several students use the same phrasing, sentence structure, or even the same jokes, it often means the material was copied from the same source. Notes that are overly formal or inconsistent with the student’s usual writing style should be questioned.
✔️ Inability to respond to probing or follow-up questions - AI can generate impressive responses, but it cannot prepare students for every possible question. If a student struggles to explain, expand, or rephrase their ideas when prompted, this suggests they may not fully understand or own the content.
✔️ Unusual or inconsistent vocabulary - If the vocabulary used in the speech feels advanced or unnatural compared to the student’s speaking ability in other contexts, this can signal external generation. Ask students how they selected certain phrases or if they can explain their word choices.
Sample Prompt to Try
“Use an AI tool to help you generate five possible discussion questions and draft your answers. Submit your notes. In class, you will be asked one new question you have not seen before. Your assessment will focus on how clearly and confidently you explain your ideas and respond to follow-up questions.”
📚 Resources to Support You
🔓 Oral Response Planning Sheet This structured sheet guides students in outlining key ideas, organising their response and identifying where AI helped in preparation. It includes space for additional thoughts added during the discussion, making it a helpful tool for flipped learning or scaffolded in-class preparation.
🔒 Student Voice Rubric – AI-Aware Criteria A downloadable rubric for teachers to assess speaking tasks, with specific descriptors that reward spontaneity, personal reflection, and real-time thinking. Includes example feedback comments and a column for evaluating appropriate use of AI during the planning phase.
You may also wish to capture oral assessments on video for later review, moderation, or reflective learning. If so, ensure you have secure storage solutions in place and that parental permissions are obtained in line with your school’s safeguarding policy.
🧠 Reflective Prompt How much of your next speaking task could be generated by AI and how will you build in checks for live, student-owned thinking?
🗂️ Full Series: Teaching Smarter: Designing Lessons for the Age of AI
✅ Post 1: The AI Dilemma: Why Pedagogy Needs to Adapt - Why traditional task design is no longer fit for purpose in an AI-enabled world.
✅ Post 2: Redesigning Written Work in the Age of AI: Essays, Reflections and Reports - How to adapt extended writing tasks so AI supports pre-writing, not replaces original thinking.
✅ Post 3: AI and Oral Tasks: Structuring Authentic Discussion and Verbal Responses How to safely integrate AI into planning for presentations, interviews and spoken assessments without losing student voice.
🔜 Post 4: Making Projects AI-Resilient: From Polished Products to Visible Thinking - How to redesign project-based learning so that AI can support the research phase but not overshadow the process and originality.
🔜 Post 5: Rethinking Routines: Retrieval, Scaffolding and Quiz Tasks in an AI World - How to adapt daily classroom activities and low-stakes tasks to reduce AI misuse and deepen cognitive demand.
🔜 Post 6: Assessment in the AI Era: Tracking Thinking, Not Just Outcomes - Strategies for building process-driven, AI-aware assessments that showcase genuine student learning.
🔜 Post 7: Building a Culture of Integrity in an AI-Enabled Classroom - How to lead conversations, policies and shared expectations that embed responsible use of AI without resorting to bans.
🔜 Post 8: Your AI-Aware Lesson Design Framework: Practical Planning for the Future - A printable, teacher-ready planning model to embed everything from this series into daily practice.
📢 If this post helped you rethink how you approach speaking and discussion tasks, forward it to a colleague who’s wrestling with AI in the classroom or share it with your department during planning.
Leading and Learning Conversations - Sarah Findlater is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
