New to using a tack saw or any other power tool, I had just finished carefully cross-cutting four 7-inch sides of a plain wood box. Despite being new to woodworking, I was confident in the precision with which each piece would be perfectly square at a 90 degree angle.
But when it came to turning the saw at a 45 degree angle to cut the miters along each edge, I ran into a dire situation. The sides of the box were too high to fit the saw fence snugly and make a safe, clean and precise cut. cut. My design was flawed. I should have planned to make a smaller box from the beginning. Should I try to cut it anyway and risk wood splinters flying in unpredictable and irregular patterns around my woodworking shop, or should I go back to the drawing board?
It’s ChatGPT, the artificial intelligence (AI) chatbot that’s now taking the internet by storm, that got me into this mess in the first place. And when I desperately needed advice, a disembodied computer brain left me hanging.
A “simple” experiment
A few weeks ago, I was providing ChatGPT with what I (mistakenly) thought was a simple prompt. “What is the first basic woodworking project I can start with?”
I was encouraged. It’s finally time to revisit a skill completely forgotten in my failed attempt to build a gumball machine in middle school. A replacement woodworking teacher completely took over for me and “helped” cut the wood, causing it to burn badly.
I thought a box would be easier. Indeed, tools like ChatGPT can be a good guide. But I quickly realized that AI was giving us enough information to be dangerous.
AI Ingenuity and Limitations
ChatGPT (Generative Pre-Training Transformer) is the brainchild of San Francisco-based laboratory Open AI, whose mission is to “ensure that artificial general intelligence benefits all of humanity.”
✅ Get the Facts: Artificial Intelligence
• How does ChatGPT actually work?
• Everything You Need to Know About AI Reaching the Singularity
• Artificial intelligence can’t decipher memes – it could save humanity
It is trained on a large amount of publicly available text. ChatGPT uses a neural network, which is basically a series of algorithms or instructions, to understand words just like the human brain does. Chatbots find patterns in human language and respond with what they believe to be the answer the user is looking for, with varying results. So when I commissioned a basic woodworking project, the response was based on a mashup of sources. And I’m not sure if those websites, blog posts, or manuals are trustworthy in the first place. Not to mention their interpretation by machines. As you know, ChatGPT is not sentient. Generates sentences that make sense but lack context, authority, and expertise.
AI vs. amateur woodworker
First, I took ChatGPT’s instructions and their list of recommended materials and tools at face value.
Step 1 seems pretty easy. Using a table saw, cut a 1×8 pine board into his four pieces: the two sides, the bottom, and the top. But wait, shouldn’t a box have four sides instead of two? How on earth would you use a table saw? And how long should each piece be cut? Given that a 1×8 board isn’t exactly 1 x 8 inches, you’ll have to measure your lumber anyway. We already had to adjust our plans as the board dimensions were always slightly off. I chose four 7 inch square faces for the box. This meant that the wood had to be “ripped” or cut lengthwise along the grain.
Thankfully, Roy Berenthorn—pop mechaSenior test editor and in-house woodworking expert at ChatGPT, he kindly pointed out the basics that ChatGPT had left out. He showed me how to safely use a portable table saw in his Easton, Pennsylvania office. So I tore the pine tree into 7 inch wide strips. It was a disappointing experience, so I decided to proceed with cutting the sides of the box using a cordless tacking saw. Roy again taught me the basics of safety and accuracy and supervised me crosscutting.
At this point I expected step 2 would require a massage. The chatbot suggested using the same saw to cut a mitered edge on the box that fits neatly at a 45 degree angle. This was no beginner’s job. Miter joints must be exact. Otherwise the box will not fit snugly. In a perfect world, you would deviate from the ChatGPT instructions and create a butt joint instead. But to stay true to the experiment, we moved forward and arrived at the dead end of the mitered joint. I decided to install a stop block to cut the miter accurately. It was a battlefield. With each cut, a small wooden rocket cracked and flew around the store, and it was nothing like what I had imagined. In the end, the miter was far from perfect, so when I glued the sides together, I could see the light coming through the small gap.
How to become a true teacher?
I decided to end things there. Thanks to Roy’s lessons and help deciphering ChatGPT’s half-hearted instructions, I’ve learned the basics of safety, precision, and design needed for my next attempt: a 4-inch square box with butt joints. I was able to gain practical knowledge about
Unlike a human instructor, ChatGPT was unable to “read” my skill level and adjust plans accordingly. This is the result of having just surface-level intelligence. As you can see, AI has a long way to go. Still, I was grateful for the starting point and second chance for this daunting yet challenging task.
Perhaps in the future, chatbots will be able to visually interpret students like me to get more feedback about where their problems lie, making them true teachers. Meanwhile, you’ll find me at the store asking Roy endless questions.
Deputy editor-in-chief
Prior to joining Pop Mech, Courtney was a technology reporter for the Pittsburgh Post-Gazette, a local newspaper. She graduated from the University of Pittsburgh, where she studied English and Economics. Her favorite topics include but are not limited to giant squid, punk rock and robotics. She lives in suburban Philadelphia with her partner, a black cat, towering over her tower of books.
