Can we fix it? AI-Fixer reckons we can, and its creator aims to revolutionise at-home repairs, putting consumers back in charge and providing designers with the feedback they need to make better products in the first place
W e’ve all thrown away a broken device and bought a replacement, just because tackling a repair looked to be too timeconsuming or complicated.
For Nazlı Tergiozlu, a research tutor at the Royal College of Art (RCA), London, that wasteful habit represented an intriguing challenge. To tackle it, she has designed a chatbot based on artificial intelligence (AI) called AI-Fixer, which aims to help consumers carry out device repairs with confidence. This embodies much of her expertise in product repair for circularity, a topic in which she’s long taken a keen interest, beginning with research in 2011 into why appliances break down.
This exploration led Tergiozlu to realise that, while most broken appliances are repairable, they don’t get fixed. For her PhD research, she looked into the reasons why and identified no fewer than nineteen of these. They included the time that a repair might involve; users’ lack of confidence in being able to carry out the repair successfully; and their frustration and embarrassment about not knowing where to start.
AI to the rescue
With the release of OpenAI’s chatbot GPT-4 in 2023, Tergiozlu began considering how AI could help consumers with repairs, figuring that they might feel more comfortable asking a chatbot questions than a human expert.
“During my PhD, I realised that there are people who want to repair their products. I don’t need to convince everyone. There are so many people who want to repair their products, who want that knowledge, who want products to be more repairable,” she says. “I thought a chatbot could help them ask the questions they need to.”
Tergiozlu partnered with smartphone manufacturer Fairphone, a company that prioritises product sustainability and the circular economy. The partners then held workshops in which repair novices and experts alike used the bot to guide them through repairing devices. As Tergiozlu discovered, it was complete beginners that got the most from it.
Expert fixers with more knowledge and experience preferred to go straight to ifixit.com, find the right manual, understand it in seconds, and solve the problem. Yet, with AI-Fixer’s help, those less experienced managed to successfully repair Fairphone devices in the workshops, and many wanted to provide feedback to Fairphone about parts of the process.
Tergiozlu was aware that although a lot of feedback is posted online about manuals and the repair process of devices, it rarely gets back to the designer from community forums like Reddit or ifixit.com. Now that designers are beginning to integrate AI tools with their websites and digital manuals, Tergiozlu theorises, it may get easier to send feedback directly to designers.
Originally, the aim with AI-Fixer was to replace repair manuals. After all, they are notoriously difficult to read and hard to scan for exactly the right piece of information. But, as the workshops continued, Tergiozlu noticed that people did not only want an easier instruction manual. They also wanted more interactive materials, ones that included photos of each stage of a process and instructional videos.
This, she believes, is what AI can provide.
“I imagine [AI-Fixer] being part of the manual, and then you can chat and ask certain questions to that AI tool when you start reading the manual,” she explains.
“People are already using this kind of web page, for example, for Phillips shavers. The menu – I checked – has been visited over 150,000 times. If companies implement chatbots with their websites and provide proper knowledge, they’d receive the benefit of feedback about which parts of the information fail.”
Right to repair
This technology could be applied to all sorts of products, not just consumer electronics, Tergiozlu reckons.
Although many manufacturers recommend repairing and maintaining devices, few of their customers actually know how to do this. How many of us neglect changing the filter on their washing machines, for example? AI-generated videos created for each different model of washing machine, Tergiozlu argues, could get us changing those filters without wasting time researching how to do so.
Looking ahead, she wants to work with designers of different appliances to see how AI-Fixer might help a wide variety of consumers. “In the next five years,” she says, “I want to develop my research in different scenarios. Maybe I will look at 20 scenarios and select a couple of them, and then implement those scenarios, create the tools and test them with the companies. I want to focus on electrical and electronic equipment appliances at home. I want to provide something that people will use.”
From washing machines to shavers to smartphones, Tergiozlu’s research all points to the same conclusion: that with the help of AI, our gadgets might well prove simpler to repair than we thought, and the insights we gain from fixing them might prove just as valuable to designers. www.ecologicalcitizens.co.uk/ec-projects#ai-fixer.com
