Using AI with micro:bit to teach students about the climate crisis

26 January 2026

This week, I went to BETT, the education technology show in London's Docklands. In 1988, long before the Excel Centre was built, I was on that spot for Jean-Michel Jarre's spectacular Destination Docklands. During the concert, images of cultural and technological revolution were projected onto the Millennium Mills building across the dock.

At BETT, the revolution that everybody was talking about was AI. Some companies were promoting it for administration and grading, with one asking "Are you still marking like it's 1995?". Others had products to help students learn about how AI works and how it can be used safely.

One session was called "Coding for Climate Action". It brought together three things I have a keen interest in: sustainability, AI, and micro:bits. If you're not familiar with the micro:bit, it's a small programmable device with a grid of LEDs, built-in sensors, and GPIO pins for connecting it to peripherals and electronics projects. I saw several micro:bit robots rolling around table tops on various booths.

To give you an idea of what the micro:bit can do, a 15-year old student from Hong Kong showed me his project on one of the booths. It was an assistant for the elderly, featuring a fall sensor, medication reminder, and games to keep the mind sharp. My free Coding Compendium book includes projects to make a pomodoro timer and use the micro:bit as a steering wheel in a Scratch game.

The Coding for Climate Action session, run by Matt Hewlett from charity Digitall, shared a series of lesson plans for years 4 to 8. Projects include logging temperature around the playground to show how trees help regulate temperature; a fitness tracker to start a conversation about how we can reduce the demand on healthcare, which is responsible for a lot of road emissions; and monitoring plant growth.

At the start of the session, we were given a micro:bit and challenged to work out how to make it display a skull image. I was familiar with the device, so I tried tilting it and pushing its buttons in various combinations, but nothing worked. The answer was to whirl it around like the arm movement of a cricket bowler but without letting go. Matt's nine-year-old son had used machine learning to program the micro:bit to recognise that gesture.

Matt demonstrated CreateAI, which enables you to train the micro:bit to recognise custom movements using its combination of sensors. In one of the projects in Coding for Climate Action, students use CreateAI to train their micro:bit to recognise a landslide in what Matt says is their most popular lesson. There's also a project to track storm movements.

The CreateAI training works by recording sensor movements as you perform the gesture you want to recognise. You provide several sample movements to increase the accuracy when the software attempts to recognise the movement later. This process enables your programs to respond to nuanced and sophisticated movements that you couldn't describe in a human-authored program, far beyond the relatively simple x, y, and z axis movements you can easily detect.

One of the best features of the micro:bit is the built-in communication, so Matt demonstrated how we could all move our devices to simulate gentle or extreme winds, and the data can be collected centrally and shown on a map with each micro:bit channel representing a location.

It was good to see the conversation around AI expand beyond ChatGPT and similar generative AI tools, and into machine learning which is widely used in academic and industrial applications. I think an understanding of machine learning also helps to understand how generative AI works, so it's a great starting point for AI education. While AI (especially generative AI) is contributing to the climate crisis because of its intensive energy use, I don't think this application of machine learning is hugely demanding of energy or resources. I'll be looking at whether I can use some of these projects in my Code Club.

Download the Coding for Climate Action projects, including lesson plans and mappings to the National Curriculum, here. You can also download my free Coding Compendium, including two micro:bit projects, and check out my Raspberry Pi and Python tutorials.

Permanent link for this post | Blog Home | Website Home | Email feedback

Trimming the silence from hidden tracks using AI

05 January 2026

Although most people are streaming now, I'm still living in the world of CDs and MP3s. I buy CDs and rip them to iTunes, where I playlist them for work, listening out and about, and inspiration when running.

I recently picked up the Alt-J album 'This is All Yours' in a charity shop, and was intrigued to see that the last track was 16 minutes long. About ten minutes of that, though, is silence. It's one of the many albums that have a hidden track, concealed within the final track after a long silence. I understand that bands might want to gift listeners a song that they don't think is part of the main album, or might prefer to surprise listeners with a particular song. But it's an irritating listening experience having a long silence in the middle of a track. It means neither song can be playlisted or shuffled. A couple of much-loved albums by my favourite artists have the same issue, including Kenickie's At the Club and Depeche Mode's Sounds of the Universe. (That's Lauren Laverne from Kenickie pictured on the right in a photo I took when I was writing for Making Music. See more photos here.).

So, I wondered whether this is something ChatGPT could fix for me. I've been doing a number of experiments with it recently, so this seemed like an interesting thing to test, especially since I hadn't tried any audio processing with it before. My plan was to upload the MP3 file and have it split it into two separate files with the silence removed. I could then import these files back into iTunes.

It proved to be harder than I expected. There were several attempts where ChatGPT split the track into two files, but did not remove any of the silence. It seems the silence wasn't perfectly silent, so an alternative approach was used where ChatGPT focused on where the music was instead of where the silence was. At one point, ChatGPT was getting confused because it was overwriting the file I'd uploaded with one of the new files it had created. After a few goes, I thought that ChatGPT understood the problem even though it wasn't providing a solution. I asked it to write me a prompt I could use in a new chat to get the expected results. ChatGPT built some checks into the prompt and indicated how to perform the task, which helped to increase the success of the prompt. It took a few iterations in the new chat, too, but I did end up being able to split tracks easily.

The verdict, then? This is a problem that ChatGPT can solve, but the amount of iteration required isn't justified given I only have a handful of files to edit. It wouldn't take very long to edit the files manually using a tool like Audacity. Nonetheless, it was interesting to see that ChatGPT is technically capable of performing this work.

Permanent link for this post | Blog Home | Website Home | Email feedback

Discover GamerCard for handheld gaming and coding

12 November 2025

The current issue of the Official Raspberry Pi Magazine includes my interview with inventor Grant Sinclair about his handheld GamerCard device. It's based on the Raspberry Pi Zero 2 W, so it can run a wide range of software, including PICO-8 and classic computer emulators. There are two four-button controllers under the pads on the front of the device.

Grant's vision is for GamerCard to be used not just for gaming but also for coding. He's incorporated Qwiic connectors so that users can access the GPIO pins to expand the device and schools can ensure that the devices remain reusable. He said some schools had Arduino or Raspberry Pi devices that they couldn't reuse because projects had been soldered on to them.

Grant is the nephew of Clive Sinclair, who gave us the ZX-81 and ZX Spectrum home computers. In the same way that those machines brought a mixture of gaming and coding into homes, Grant's new device puts those capabilities in the palm of your hand. I can imagine myself packing a small keyboard and a GamerCard for coding on long train or plane journeys.

I interviewed Grant in the bar of a fancy hotel in London on a sunny day. Most of the surfaces were mirrored or glass and the sun was streaming through the windows. I'm especially pleased at how well the photos came out in those circumstances. I used a red padded wall and a stripy rug as backdrops, and angled the camera and device carefully to minimise reflections.

Below you can see one of Grant's videos of the GamerCard, showing the Nintendo Switch games BlooKid2 and AstroBlaze which have been ported to the GamerCard.

The Official Raspberry Pi Magazine, formerly known as The MagPi, is available in newsagents including TG Jones, formerly known as WH Smiths. Why not buy a copy while listening to The Artist Formerly Known as Prince? You can also order online here.

While you're here, why not read my Raspberry Pi articles, Scratch tutorials, or get my free Coding Compendium ebook?

Permanent link for this post | Blog Home | Website Home | Email feedback

Out now: 10th Anniversary Edition of Earworm

01 October 2025

My novel for adults Earworm is now available in a 10th Anniversary Edition.

In the book, a record company plans to replace pop stars with music-generating technology, and comes under fire from ageing rocker Dove and two of his most devoted fans.

It's never been more topical: The Velvet Sundown recently achieved a million streams on Spotify before they were rumbled as being an AI band. Streaming service Deezer reported this month that it is receiving more than 30,000 fully AI-generated tracks every day, more than 28% of new music uploaded. When I tried using AI to make music, I was struck by how convincing it was. Play my EP Songs About Coding here.

Earworm was first published under the title University of Death in 2007, named after the band at the centre of the action. When I published it on Kindle in 2015, I made some small edits, commissioned a new cover and changed the title to Earworm.

Music magazines have reviewed the book positively, saying it's smart, fast moving and funny. The latest review is in this month's Blitzed magazine (issue 22). They said: "Given that it was first published in 2007, Earworm is really prophetic. Who would have thought that computers could write entire songs in just a few seconds? I really enjoyed Earworm and the Extras section at the end is very interesting too. There are several laugh-out-loud moments. Highly recommended."

The extras section has been enhanced with 101 free band names, taken from the source Jonathan identifies in the story, and a new Afterword. There are two deleted scenes (only one on Kindle for technical reasons), and I updated the Poptastic Trivia, too, which often surprises readers by showing how much of the book was inspired by real life.

It was a delight to read it afresh after so long and to be reunited with the characters from the book, with whom I spent a couple of years during drafting. I wasn't sure how I would feel about the book looking back on it now, but I was pleased with the characterisation and the way it all fits together like a jigsaw. I would find myself wondering why I had put a particular scene or detail in, only to rediscover its role in the plot later on.

Earworm is out now in print and on Kindle. It makes a great gift for the music lover in your life. Find out more, get a free PDF sampler of Earworm and buy your copy here.

Permanent link for this post | Blog Home | Website Home | Email feedback

What are the security risks of agentic AI, and why are firms merging their HR and IT departments?

29 September 2025

I had two new BBC articles published during the school summer holidays:

• How can we stop AI agents going rogue?: New risks emerge when we let AI make decisions and take actions automatically. Security community OWASP published a list of 15 threats that are unique to agentic AI, on top of those that arise from AI generally (such as hallucinations and bias). For this article, I looked into the new risks and some of the ways they could be mitigated. One particularly alarming incident I wrote about was when an AI system attempted to blackmail an executive to stop him switching the AI off. Luckily, this was in a test environment, but it illustrates how AI could attempt to manipulate people to achieve its aims.
• Why are firms merging their HR and IT departments?: When I first heard that 64% of senior executives expect to merge their HR and IT departments within five years, I was puzzled. I always considered HR and IT to be quite different: one is about people, and the other is about technology. My interviewees explained that people and machines need to work more closely together as AI is adopted, and that organising teams around the work makes it easier to use a mix of AI and human skills to deliver it.

I've recently redesigned my articles page to make it easier to navigate and to reduce its carbon footprint by cutting the number of images. I hope you like the new streamlined design.

Permanent link for this post | Blog Home | Website Home | Email feedback

Visiting the Bourdon Street Chippy

10 September 2025

I was taking a photo of a dinosaur fossil sculpture in the street, when three kids rushed up to it, followed by their Mum. "Did you know there's a felt chip shop around the corner?" I asked. "Like, a chip shop made of textiles."

"You mean the Bourdon Street Chippy?" she said. "We came past that. We thought it was real."

It is utterly convincing at first glance, and wonderful for kids and adults alike. I had visited earlier and felt joyful to be immersed in such a fun and surprising artwork.

The food on the table, signs and pictures on the wall are all crafted in textiles

Artist Lucy Sparrow and her team have recreated a dine-in and takeaway chip shop, with tables set with plates of food. The fish, sauce bottles and celebrity photos on the wall have all been created in textiles in brilliant detail. In the room next door is a hot food cabinet, filled with fish, sausages, mushy peas, and pies. The front of the cabinet displays a stitched poster for a circus. Behind the counter is a jar of onions, pickles and eggs underneath a neon takeaway sign. All the the food, down to the individual chips, has a cute face, with black beads for eyes.

Artist Lucy Sparrow at the chip shop, where she serves up felt fish and chips to customers

Like Lucy Sparrow's previous installations (including the Cornershop in 2014 and the Bourdon Street Chemist in 2021), everything is for sale. It's art, so it'll cost a bit more than your local chippy. A large cod and chips meal is £450. However, prices start at £5 for a sauce sachet, so everyone can join in fully and own a part of the installation. On my desk there is now an onion, with big beady eyes and a satisfied smile, catching the light in its gold thread. When you order, Lucy or one of her assistants serves your food. She picked up my onion with felt-covered tongs and carefully put it into a paper bag before ringing it up on the card reader. It's strange to be in a place so fantastical, that is nonetheless a real functioning shop.

The installation took Lucy and her six assistants eight weeks to create and has taken over the Lyndsey Ingram art gallery in London.

When I visited, Lucy proudly pointed out her award on the counter from the National Fish and Chip Awards. It was for the most heartfelt chip shop opening.

The installation runs for six weeks until 14 September 2025, open Wednesday to Sunday from 11am to 6pm. Photos by Sean McManus. Please ask if you want to reuse them.

The food has cute faces on it, like these friendly pies

Condiment bottles, also made from textiles

Permanent link for this post | Blog Home | Website Home | Email feedback

Shadow AI: Why are employees smuggling their own AI tools into work?

04 February 2025

Artificial intelligence (AI) is a valuable productivity tool for many jobs now. I've written previously about how AI can negotiate contracts and help with programming, but people are increasingly bringing their own AI tools, such as ChatGPT, into the workplace without permission. According to a Software AG survey, half of knowledge workers use personal AI tools. Unauthorised AI tools are sometimes called "shadow AI", a more specific version of "shadow IT", which is the term used for unauthorised IT.

For my latest article for the BBC's Technology of Businesss, two people told me why they use shadow AI, on condition of anonymity. The article also looks at the risks of unauthorised AI and showcases Trimble's solution to shadow AI, which was to create its own general-purpose AI tool for its people to use.

The article was given the top feature slot on the BBC News homepage, and was followed up by the Business Matters programme on the BBC World Service. Read it here: Why employees smuggle AI into work.

To be updated when I write articles like this in the future, please susbcribe to my newsletter. Thank you to everyone to helped with the article, especially my two interviewees who shared their experiences anonymously.

Photo credit: Steve Johnson at Unsplash

Permanent link for this post | Blog Home | Website Home | Email feedback

Dip into the blog archive

June 2005 | September 2005 | January 2006 | March 2006 | April 2006 | May 2006 | June 2006 | July 2006 | August 2006 | September 2006 | October 2006 | November 2006 | December 2006 | February 2007 | March 2007 | April 2007 | May 2007 | June 2007 | July 2007 | August 2007 | September 2007 | October 2007 | November 2007 | December 2007 | January 2008 | February 2008 | March 2008 | April 2008 | May 2008 | June 2008 | July 2008 | August 2008 | September 2008 | October 2008 | November 2008 | December 2008 | January 2009 | February 2009 | March 2009 | April 2009 | May 2009 | June 2009 | July 2009 | August 2009 | September 2009 | October 2009 | November 2009 | December 2009 | January 2010 | February 2010 | March 2010 | April 2010 | June 2010 | August 2010 | September 2010 | October 2010 | December 2010 | March 2011 | April 2011 | May 2011 | June 2011 | July 2011 | August 2011 | September 2011 | October 2011 | November 2011 | December 2011 | January 2012 | February 2012 | March 2012 | June 2012 | July 2012 | August 2012 | September 2012 | October 2012 | December 2012 | January 2013 | February 2013 | March 2013 | April 2013 | June 2013 | July 2013 | August 2013 | September 2013 | October 2013 | November 2013 | December 2013 | January 2014 | February 2014 | March 2014 | April 2014 | May 2014 | June 2014 | July 2014 | August 2014 | September 2014 | October 2014 | November 2014 | December 2014 | January 2015 | February 2015 | March 2015 | April 2015 | May 2015 | June 2015 | September 2015 | October 2015 | December 2015 | January 2016 | February 2016 | March 2016 | May 2016 | July 2016 | August 2016 | September 2016 | October 2016 | November 2016 | December 2016 | January 2017 | July 2017 | August 2017 | October 2017 | November 2017 | January 2018 | February 2018 | August 2018 | October 2018 | November 2018 | December 2018 | January 2019 | March 2019 | June 2019 | August 2019 | September 2019 | October 2019 | January 2020 | February 2020 | March 2020 | April 2020 | May 2020 | June 2020 | September 2020 | October 2020 | December 2020 | January 2021 | February 2021 | May 2021 | June 2021 | October 2021 | November 2021 | December 2021 | January 2022 | February 2022 | March 2022 | May 2022 | July 2022 | August 2022 | September 2022 | December 2022 | March 2023 | April 2023 | May 2023 | June 2023 | October 2023 | November 2023 | January 2024 | February 2024 | May 2024 | June 2024 | July 2024 | September 2024 | October 2024 | November 2024 | December 2024 | January 2025 | February 2025 | September 2025 | October 2025 | November 2025 | January 2026 | Top of this page |