Silent podcast #8

Welcome to the eighth InThinking Physics 'Silent Podcast', written back and forth in a shared document. We welcome your suggestions for future topics: put your questions (whether timely or timeless) in the comments.

For today’s podcast, we are going to be returning to the theme of AI and we're really pleased to welcome back, Richard Allaway.  Richard’s subject is Geography and is the author of the internationally renowned site Geography All The Way.  As we found out before, Richard is so much more that a teacher, geographer and resource creator, he is a true technology integrationist.  

Last time we were focused on the students and how it might impact on our teacher, this time the plan is to talk about the positives from the perspective of the teacher so let's get going…  

Tim: Happy Christmas and thanks so much for taking time to return for another Silent Podcast.  First off can I ask you for a simple example of an everyday use of AI that you find particularly useful?

I am currently grappling with two new (to me) courses, DP ESS and DP TOK. I have spent a little time building two 'subject experts' using Claude Projects - essentially custom AI assistants focused on specific topics. I have uploaded all the official IB documentation, as well as various other sources, to provide an 'expert' to whom I can ask questions. This is really useful when I want to check something procedural, such as an exam requirement, find an illustrative example for an ESS syllabus point, or get a first draft of a potential TOK knowledge question to structure a lesson around. All the major AI services offer these 'projects' (Google's Gemini Gems, ChatGPT's GPTs and, as mentioned, Claude's Projects).

I am a DP Geography teacher who prides himself on using contemporary case studies and examples. This means that I need to research and produce new case studies and related teaching materials quite regularly. Google's NotebookLM is my go-to tool here. NotebookLM allows you to upload up to 50 sources (audio files, YouTube clips, documents and webpages), which then gives you your own focused AI upon that topic area. The answers you then get are numerically referenced against all the sources you have added. NotebookLM also allows me to create 'audio overviews', which I then upload to Spotify for students to listen to when reviewing lesson content.

Tim: Which AI engines do you find yourself using regularly and why?

My current go-to is Claude, as I like the interface and the output seems more in line with what I am looking for. I have also turned on Claude's 'memory' feature, so the more I use it, the more it knows what I am looking for and the usual endpoint I am happy with. It is a lock-in, but at the moment, I am happy with it. At school, I am trying to make more use of Gemini. We are a school that uses Google Workspace for Education as our core digital tools, and there is a growing feature set that links Gemini with my email, my Google Calendar and the content of my Google Drive. Some would find this level of insight scary, but for me, it helps surface work I have done before and make connections I might otherwise miss.

I try to keep an eye on developments across all major platforms, as new features are announced nearly weekly. This is no easy task, and I still use a feed aggregator (Reeder for Mac and iOS) daily to follow news releases, blog posts, and new YouTube video releases.

I follow Anthropic's Newsroom as the owner of Claude. Google's The Keyword for all things Gemini and NotebookLM. I tend to keep up with OpenAI and their ChatGPT by following their YouTube Channel.

Tim: You work in an international school, have the linguistic capabilities of AI been useful to you?  I’m thinking specifically about real-time translations or video editing.

Using AI to support EAL students and lesson accessibility generally is something I am really starting to reflect upon. The NotebookLM audio overviews I mentioned earlier can be generated in a wide range of languages, even when the source information is in English. This is quite a powerful feature - you can take English-language resources and make them accessible to students in their home language without any additional work. I have offered these to my students in any language they wish and have had a couple of takers so far. A colleague has had some success using HeyGen, an AI video generator, to create some unit introductions for students who have recently joined his school.

Although it does not use an LLM directly, Google Meet's live translation feature is powered by advanced artificial intelligence. I regularly use translated captions during bilingual meetings, usually with the school's IT team. These practical applications are where AI starts to feel less like a novelty and more like genuine infrastructure for inclusive teaching.

Tim: Your approach seems pragmatic rather than evangelical. How do you stay enthusiastic about AI while keeping a critical eye on its limitations?

I have been in education long enough to have seen plenty of 'this changes everything' technologies come and go. Interactive whiteboards were going to transform learning. Then it was tablets. Then, coding across the curriculum. Each had genuine value, but none delivered the revolution promised. AI is different in scale and capability, but I try to keep that history in mind.

My enthusiasm comes from the practical wins - the time saved, the tasks made easier, the new possibilities for students. When I can produce a first draft of a case study in twenty minutes that would have taken me two hours, that's genuinely exciting. When a student can get patient, personalised explanation of a concept they're struggling with at 10 pm the night before an assessment, that matters.

But I also regularly bump into the limitations. AI confidently produces nonsense sometimes. It can be brilliant at generating plausible text while completely missing the point of what you actually need. The 'hallucination' problem is real, and students need to understand that these tools require verification, not blind trust.

I suppose my critical eye comes from using AI enough to know where it falls short. If you only dabble, you might think it's either magic or useless. Regular use shows you it's neither - it's a powerful tool with genuine limitations, and knowing both is what makes you effective with it.

Tim: How do you approach colleagues who are sceptical or nervous about using AI in their teaching?

I start by listening. Scepticism usually comes from somewhere reasonable - concerns about academic integrity, worry about being replaced, uncertainty about where to even begin, or simply not having the headspace to learn something new on top of everything else teachers are already juggling. Those are all valid starting points for a conversation.

I try not to be the person who bounds up to colleagues, enthusing about the latest feature. That approach tends to make people's eyes glaze over. Instead, I look for the pain points. What's the task they find tedious? What's eating their time? Report writing, creating differentiated resources, answering the same student questions repeatedly - these are the places where AI can offer genuine, immediate relief. If I can show a colleague how to save an hour on something they find draining, that's worth far more than a theoretical discussion about the future of education.

I also try to normalise not knowing. I am still learning. I get things wrong. I try approaches that don't work out. When colleagues see that experimentation and uncertainty are part of the process, it feels less like they're being asked to become instant experts.

For the genuinely nervous, I suggest starting with something low-stakes and personal - using AI to draft an email, summarise a lengthy document, or brainstorm ideas for a lesson. Nothing that goes near students until they feel comfortable. Once they've had a few small wins, the bigger possibilities start to feel less daunting.

The colleagues I worry about least are the sceptics - healthy scepticism keeps us grounded. It's the colleagues who disengage entirely that concern me more.

Tim: Some of our subscribers are regularly using AI but are looking for ideas about how to move to the next level. Any suggestions about possible things they might want to explore?

The only way I develop my AI skills is to push myself to use them. This comes from taking a moment of reflection before I start an admin task. Could this task be done quickly, more effectively, or in a way that will make it easier to repeat in the future? Then it's time to think about the prompt that will get me what I want. For example, I recently needed to process a batch of student reports, checking for consistency in formatting and name spellings. A year ago, I would have just worked through them manually. Now, I pause and think: is there a way AI could help here? Often there is.

A lot of the time, the limiting factor is my lack of imagination when it comes to creating prompts and creative thinking. It is worth looking at YouTube videos from the likes of Jeff Su and Tiago Forte for prompting ideas and upping my expectations of what I can get AI to do, usefully.

Courses and certifications, such as Google's recently released Gemini Certified Teacher, and Common Sense Education's ChatGPT Foundations for K–12 Educators and Advanced ChatGPT for K-12 courses, are always worth a look.

Tim: Thank you so much - you've given me a lot of holiday exploration to get on with.  I hope you have a relaxing break and best wishes for 2026!