How will AI change the world of knitting?
Last year was a big year for artificial intelligence. It brought the use of new software (hello ChatGPT), heated discussions on AI’s potential risks, and demands for responsible policies for its development. The general public began to understand what AI is — and what it can do.
With so many creative fields already affected by the development of AI, it is only a matter of time before we will start seeing its effects in hand knitting as well. Could AI models be used to generate knitting patterns?
“I cannot think of any reason why not. Given enough training data related to knitting, it should be possible,” says Dorota Glowacka, an associate professor at the Department of Computer Science at the University of Helsinki . “However, I don’t think the existing AI models are actually any good at this, at least not yet.”
“AI can be used to generate creative solutions or pictures, but that does not mean that the AI is creative in a similar way as us humans,” explains Samia Touileb, an associate professor in Natural Language Processing (NLP) at the Department of Information Science and Media Studies at the University of Bergen, Norway. “For an AI to create knitting patterns, it does not require any creative component, it only re quires a large set of knitting patterns from which it will learn to identify common regularities and jargon.”
For the creatives working in the field — designers, tech editors etc. — this poses threats. Are they in danger of losing their jobs because of AI? From a knitter's perspective, one important question is this: how should the use of AI be acknowledged in the design?
“Personally, I think that design work using AI should be labelled as such, and I can then avoid it if I choose,” says Glowacka.
Laine worked with Glowacka and Touileb to list some ways AI might change knitting culture:
1. A tool for designers
There is a lot that goes into creating a knitting pattern. Touileb says that AI could make designers’ lives easier by, for example, helping them grade a pattern for specific sizes, or a wider range of sizes. “I think It could also become a tool that designers use to create or combine intricate stitch patterns and visualise them before starting the swatching process.”
2. Increasing accessibility
AI technologies could be used to translate patterns and make them available more widely, especially outside of the English - speaking community. “AI could also be used to help people with disabilities. Using AI for text-to-speech technologies could allow designers to create narration for the patterns, making it much easier for the visually impaired to keep knitting,” explains Touileb
3. Helping knitters customise
What if you could use AI to find yarn replacements from your yarn stash, or calculate modifications based on your sizing? “AI models can also be used to reduce waste and yarn consumption. We could create models that would predict the exact amount of yarn necessary for a garment, making the process more sustainable,” Touileb adds.
4. Awareness of creative rights
After news of OpenAI models that memorise a wide collection of copyrighted materials, many authors and designers have become more aware of their own rights. “We can all perhaps agree that training (or “teaching”) AI models should not be done using proprietary data without consent. AI is here to stay, so we need to agree on some sort of norms related to that,” Touileb points out.
5. Less creativity and versatility
If we start relying on AI to generate new patterns, we might end up creating models that lack uniqueness. “Bias in AI has been a huge research topic in the last few years. In a nutshell, if the training data used to create AI for knitting patterns contains far fewer examples of patterns from a particular culture, then that type of pattern will not be generated as often as others,” Glowacka explains.
Text. Päivi Kankaro
Illustration: Pauliina Holma
This feature was first published in Laine issue 20.