Completing a DIY project used to mean taking a trip to the hardware or hobby store and digging tools out of the garage. Today, a new onslaught of generative AI products, like ChatGPT, and easy to use training platforms have added another dimension to these builds that make them not only satisfying to create but lifelike in their execution.
This is the case for three DIY projects that integrated AI into their design to create moving art, easy to use health detectors, and a means for escaping boring lectures.
Kelin Carolyn Zhang and Ryan Mather are independent designers who worked together to design ‘Poetry Camera’. It’s a deceptively simple camera that uses a Raspberry Pi and ChatGPT to capture a photo—and output the AI’s interpretation of the image in the form of an original poem. Zhang and Mather tell Popular Mechanics that using ChatGPT in Poetry Camera helped them not only bring the project to life faster but gave it an essential sense of humanity.
Edward Aguilar was also looking to add a human element to his project when he used ChatGPT to build BuellerBot. Today, Aguilar is the co-founder and CEO of the human-level transcription start-up, Echo Labs. Before taking on that role he created BuellerBot —a reference to the Ferris Bueller’s Day Off roll-call scene—to help get students out of boring video lectures by having AI stand in for them to respond with thoughtful answers when called on.
These AI projects can serve a more practical purpose as well, like Diana Khalipina’s e-health projects, including a ‘tiredness detector’ based on monitoring a user's seat height and how open their eyes are. Khalipina is a freelance web developer based in Paris.
Is AI a new essential element of DIY projects? It could be, according to these designers.
What is the benefit of AI for DIY projects?
For Zhang, building Poetry Camera may not have been possible without the help of ChatGPT.
“Not only was AI the thing that makes the camera itself work, but was also crucial to lowering the barrier to us getting started with the project,” Zhang says. “This was my re-introduction to programming after almost a decade away—ChatGPT helped me to go from knowing nothing to having working code in just 4 days back in January.”
Incorporating AI into Poetry Camera from the very start helped give the project essential momentum, Zhang says. This instant gratification from idea to prototype is something that Aguilar says Bueller Bot benefited from as well.
“This wouldn’t have been possible without this technology being so easily accessible,” Aguilar tells Popular Mechanics. “There are so many ideas that are just hobbies—things you want to explore but don’t want to dedicate months to… with AI we’re able to do nearly everything it would take a team of dozens could do a year, in a weekend.”
Accessibility was an important benefit for Khalipina as well, who used Google’s Teachable Machine to train her tiredness detector.
“[I] had a dream to create an e-health project to analyze the state of health based on video analysis… but it was challenging from a hardware perspective,” Khalipina tells Popular Mechanics. “Later, I found out about Google’s Teachable Machine and I got interested in its real-time video analysis opportunity. It was quick, useful and didn’t require any expensive hardware.”
The integration of AI to a project can be important to its creative soul as well, says Mather.
“This project really needs AI because without it, the poems could only be very crude or random feeling,” Mather says. “There’s this awesome magic moment when people see that the AI has ‘seen’ something really specific or personal in the poem, and it lights them up with a childlike joy and playfulness.”
Human connection and creativity is a core principle of Poetry Camera, Zhang says, and one that a creative use of AI can help instead of hinder.
“In my opinion, AI art projects are all about restraint and perspective—so much is possible with the tool, so the creator’s taste and deliberate decisions around how to deploy the AI are more important than ever,” she says. Decisions like reading the poems aloud, never printing the source photograph, or making the DIY instructions open-source.
“All of these decisions are to encourage the human connection and human creativity, that is our true end goal,” Zhang says.
Overcoming challenges of Using AI
While integrating AI in their projects went off without a hitch, Zhang and Mather say they still faced challenges plaguing most DIYers—specifically, the fragility of hardware.
“I [learned] that hobby electronics are so much less robust than consumer-grade electronics, you kind of have to unlearn the level of carelessness we’re used to in how we treat consumer electronics,” Mather says.
One way to solve this problem, Zhang says, is to be prepared with extra money budgeted and duplicate components onhand. The solution is just having enough money to buy duplicates of everything, enough time to wait for your parts to ship, and enough persistence to try and debug the failure when there’s no clear reason why something broke.
Finding time to work on the project was a hurdle as well, Zhang says. Both Mather and ChatGPT helped alleviate this problem, and “having ChatGPT help with coding was crucial,” she says.
ChatGPT can also offer a shoulder to lean on when the project gets tough, Mather says, by helping debug code or even bea place to vent.
Best Tools or Apps for AI projects
Beyond ChatGPT, Zhang recommends tinkering with other accessible or open-source models from OpenAI or Replicate, to get a feel for how a project works before understanding the nitty-gritty. She also recommends Replit for software prototyping without a full development setup.
In addition to Teachable Machine, Khalipina recommends using GitHub Copilot for assistance in the coding process. And, she says that Vittascience AI has recently been helping her add additional markers of sickness, like coughing.
“I love the idea that the link to my training model can be used by anybody, anywhere, anytime—you don’t need to buy expensive hardware and install it,” Khalipina says. “You can get warned about your tiredness level just by having a computer and internet.”
To stay inspired for new projects, Aguilar recommends keeping up with sites like HackerNews and X to see what kinds of projects people are releasing.
Sarah is a science and technology journalist based in Boston interested in how innovation and research intersect with our daily lives. She has written for a number of national publications and covers innovation news at Inverse.
