As I embarked on this AI-enabled knitting project, I thought quite a bit about my goals for how AI can integrate into my life without violating any of my personal principles. First, I am incredibly sensitive to the risks that AI brings to the human world – job loss, creative infringement, etc. Secondly, I’m passionate about keeping my knitting practice focused on self-discovery, meditation, and inner peace.
Surprisingly, I actually believe that AI can help me become a self-sufficient knitter in a way that supports a feeling of blissful disconnection. How, might you wonder? Well, by using AI to learn more about garment construction, I hope to knit “by feel,” where I know the general size and shape that I need and can make the specifics up as I go.
This does not mean that I won’t choose books and patterns for specific items. I know that I will! But, simple knitting allows me to get into a flow state that brings me peace. Having the option to knit without a pattern means that I don’t need to check in for next steps, which keeps me locked into the meditative trance that knitting helps me achieve.
With that said, throughout this AI project, I am focused on learning flexible and classic techniques. I will not be attempting to duplicate any existing pattern, nor will I be selling any patterns generated by AI. Instead, I’ll share the general process I used and how it went. (Spoiler Alert: Don’t always trust what AI tells you!)
First AI Project: Getting Started
To prepare for my first convo with ChatGPT, I needed to have some basic info, like gauge and the ultimate style that I was going for. This required a bit of prep (knitting a swatch) and a bit of research (various styles/ease).
Input 1: Gauge To get started, I knit a gauge swatch with the Knitting for Olive Cotton Merino (Undyed) and Soft Silk Mohair (Cream) using US 7 needles. For the record, I was risky knitter and chose to knit my swatch flat even though my future project would most likely be in the round. Truth be told, the gauge variation between flat and circular knitting for me is pretty minor, so I felt like this wouldn’t be an issue.
Input 2: Style Since I am already the most comfortable with the top-down raglan sweater construction, I wanted my first project to be a raglan. That way, I can focus on the finer points that I don’t understand and maybe prevent possible rework.
Input 3: Measurements For this first project, I decided to look up some examples of various sweaters (not all raglan, actually) to determine the finished garment measurements that I would be striving for. I decided that I wanted approximately 8 to 9 inches of negative ease in the bust circumference for a relaxed fit.
Next, I asked ChatGPT what measurements it would need to help me with creating a relaxed fit raglan. Here’s its reply:
After measuring gauge from my swatch, I asked ChatGPT to help me write a pattern and provided all of the essential finished measurements. I think ChatGPT could have determined appropriate ease based on how I described my desired fit, but I didn’t want to risk it.
Next, ChatGPT asked some clarifying questions about fit, neckline, construction, and sleeve shape:
This is where the learning really begins. I clarified fit and neckline easily, but in reality, I didn’t actually know which construction options would be the best for the raglan increases, so I told it the standard raglan increase of every other round (with my fingers crossed).
At this point, ChatGPT had everything it needed to provide a pattern, and I was ready to cast on!
Early Lessons:
– In hindsight, I wish I had told ChatGPT the yarn I intended to use and asked more questions about drape.
– I should have asked my own clarifying questions about the raglan increase frequency. This could have helped me learn. Instead, I chose to learn through the experience of doing.
In my next post, I’ll share more of the pattern details and my early progress. Subscribe to easily see what happens next!
Leave a Reply