AI generated crochet patterns. Are they any good?

Introduction

Recently, AI is being by one of the most mentioned tech. Everyone seem to be concerned about AI – about AI taking someone’s or everyone’s jobs, about AI taking over whole clusters of economy and making redundant all artists and creators. That last thing is particularly interesting to crochet pattern makers as if this is true, this may mean that there is no need, or will be no need in near future for humans to make crochet patterns. Is it true that soon pattens can be generated simply by asking AI to make one and suggest couple of attributes? Let’s take a look.

What AI can and cannot do?

Firstly, it is worth mentioning that AI or Artificial Intelligence as it exists today is nothing to do with intelligence – AI is just a buzzword for commercial purposes. It sounds cool. In broad terms intelligence is one’s ability to apply existing skills and knowledge in the new area, something that allow humans to take on new things using experience. This is very thing that AI is missing.

Every AI instance is “trained” or programmed to do a specific task, for example ChatGPT can give you sensible answers to your questions, Deep Blue can play chess. However, none of them can play poker – you will need to train another instance to do that. Another significant thing is that AI does not produce anything new, at best it can synthesize its output from combination of the inputs – by definition, something that already been made. In other words, AI does not have much of creativity.

This, however, does not mean that AI is incapable of anything. A lot of products of modern artists are not necessarily new and unique – many of those are no more just result of synthesis of past works of their own or other authors with little if any additional effort. It applies to various texts, music, videos as well as crochet patterns. This is understandable as modern publishing platforms, ones like YouTube or Pinterest rewarding quantity over quality. Authors have no choice but to produce more, rather than focus on quality or uniqueness of their works. And in that sense AI can be perfectly fit to mass-produce “average quality” content.

AI potential for crochet patterns creation

At the first look crochet pattern making should one of the tasks that AI can deal with ease. It consists of limited number of basic stitches, and those can be combined into some combinations, which can be mixed and matched with each other. If all of those are fed to AI model and then it trained by giving feedback to outputs it produces, there is definitely some potential for it to produce some good results. It certainly will be able to produce any type of crochet item and offer some range of design for those.

AI generated crochet patterns overview

There are few pattern generators at the moment, most of them are building block based – not really an AI-made. So, we used to ChatGPT for our experiments. Unfortunately, it failed to produce anything meaningful even for something as simple as basic granny square, so we lowered the bar by asking to make even simpler things.

AI-generated crochet flower pattern

On a simple request to generate a crochet flower ChatGPT produced following (some additional information about materials and additional tips is omitted):

Step 1: Make a Magic Ring

Start by making a magic ring. To do this, hold the yarn tail in your palm, wrap the yarn around your index and middle fingers to form a loop.
Insert the crochet hook under the first loop and pull the second loop through. This creates a slip knot.

Step 2: Crochet the Center

Chain 3 (this counts as your first double crochet).
Make 11 double crochet stitches into the magic ring. You should have a total of 12 double crochet stitches including the initial chain 3.
Pull the tail of the magic ring to tighten the loop, closing the center hole.

Step 3: Forming the Petals

Join with a slip stitch to the top of the beginning chain 3.
Chain 2 (this counts as the first half double crochet).
In the same stitch, work 1 half double crochet, 2 double crochet stitches, chain 2, and slip stitch – all in the same stitch. This forms the first petal.
*Skip the next stitch*. In the following stitch, work 1 slip stitch, chain 2, 2 double crochet stitches, chain 2, and 1 slip stitch – all in the same stitch. Repeat this step around the circle for each petal.

Step 4: Finishing Off

After completing all the petals, slip stitch into the first chain 2 of the first petal to join.
Cut the yarn, leaving a tail of about 6 inches.
Pull the yarn through the loop on your hook and tighten to secure.
Weave in any remaining ends using a yarn needle to finish your crochet flower.

As you can see on the photos there is a mistake that causing loss of symmetry of the flower. Let’s try with slightly more complex pattern:

AI-generated crochet coasters pattern

We tried to make a coaster crochet pattern, which came back as this:

Round 1:

Make a magic ring (or ch 4 and join with a sl st to form a ring).
Ch 1 (does not count as a stitch), work 6 sc into the ring.
Pull the tail of the magic ring to tighten. Join with a sl st to the first sc. (6 sc)

Round 2:

Ch 1, 2 sc in each st around. Join with a sl st to the first sc. (12 sc)

Round 3:

Ch 1, *2 sc in next st, 1 sc in next st*; repeat from * around. Join with a sl st to the first sc. (18 sc)

Round 4:

Ch 1, *2 sc in next st, 1 sc in each of the next 2 sts*; repeat from * around. Join with a sl st to the first sc. (24 sc)

Round 5 (Edging):

Ch 1, sc in each st around. Join with a sl st to the first sc. (24 sc)
Fasten off and weave in ends.

Optional Finishing

To create a more finished edge, you can add a round of slip stitches or a picot edging (ch 3, sl st in the first ch to form a picot, skip next st, sc in next st).

As you can see in both cases AI managed to create requested pattern, however with quite bad mistakes, hardly even mediocre work.

Should I use AI for making crochet patterns?

Potentially, AI can be useful tool for generating some parts of the works, not necessary the crochet stitches themselves, but may be for some auxiliary information, like pattern descriptions, intros, material lists, etc. There is need however to be mindful of few things:

Copyright. As AI is effectively taking parts of other works in public domain (which does not mean that copyright isn’t applicable) and producing some derivative work it may potentially be an infringement of the copyright of other authors. Currently, there are some related court cases against AI-software makers in play, so there is possibility that outcome of those cases may impact anything made by AI. Consider that you may find that some of you works with AI-generated crochet patterns may in fact land you into hot water.

Triviality of AI generated crochet patterns. One of the significant attributes of AI generated content is its lack of uniqueness, as it is generally just another compilation of already existing things. Excessive use of it not only undermines author’s originality, but also may potentially cause issues with your content searchability as Google (which also mean YouTube and other platforms relying on Google Search) consistently making steps to prioritize “helpful content” which is increasingly evident is not the one generated by AI.

It appears that at the moment for authors it is more sensible to invest time into creation own library of templates and building blocks for their work, than spending time and money on searching for AI-powered magic wand that will allow them to make interesting content in one click. After all, anything that can be easily created by anyone never have much value.

What would we do?

Given all this, what should we do about emerging AI capabilities? We believe that there are few things:

Watch the space. There certainly will be some development of further capabilities AI with potential for some crochet pattern-specific to emerge. This certainly can bring some changes to way to how authors make and distribute their crochet patterns.

Work on uniqueness of your work. As it was mentioned above, AI is certainly capable of making certain common work currently done by humans, however it is generally cannot make anything new – at least for now. If anything, even when it comes to producing texts (which is much more advanced capability at the moment) it creates quite mediocre content, containing widely available information, very often containing factual errors. As we cannot expected an epiphany from AI generated crochet patterns, human authors can still stand out by making their works using their unique style and approach. However, that may mean that the content quality bar will be set higher – the development we actually would love.

Keep calm and carry on. What should not an author do is to get distracted from the work they are doing. All the new tech has a lot of hype element, promising to change the world, where in many senses it is just another shiny toy, bringing very little of real innovation. Consistent, steady work over significant amount of time – that was does change the world. Let’s keep doing it.

Also, check out our article on crochet machines.