Test an AI sweater-size prediction against the finished knit
At a glance
A knitter used ChatGPT and Gemini to choose a pattern size, then measured the finished sweater to check their predictions. This workflow is listed under Knitting, Size Planning, Checking AI Results.
SourceBuilding a tool for these workflows? See weekly rates.
What they did
1) Obtain the Ridge Sweater pattern and choose your yarn. Knit a swatch with that yarn, block it, and measure the resulting gauge. Gather the pattern gauge, finished measurements for its sizes and your desired fit. 2) Give those measurements to ChatGPT and Gemini and ask which pattern size should produce the closest finished dimensions at your actual gauge. Compare the estimates before choosing a size; the author selected Size 5. 3) Knit the selected size, then wet-block the sweater and let it dry completely. 4) Measure its bust and armhole and compare them with the saved predictions. Try it on as a separate check: a wearable sweater does not establish that its predicted dimensions were accurate.
How it went
The bust measured 96 cm against an estimate of about 116 cm. The armhole measurement differed by 0.7 cm. The author liked the fit but judged the prediction experiment unsuccessful.
Worth knowing
The cause of the large bust discrepancy remained unresolved.