"Fashion Singularity" and AI

Written on: 2024-09-24

I used to work in retail, in a clothing store. The job was rewarding in its own right, monotonous but visible, and it was fun to discover new ways to be efficient (I might talk about that in another post) or problem solve (you have to do a surprisingly large amount of problem solving when working in retail - a lot of it comes down to how resourceful you are to be honest). The work was not skilled however, and left me a lot of time to think, and listen to music while thinking.

There's one particular thought experiment I keep coming back to even after I've left, that I think is a little interesting to discuss, especially because it kind of relates to technology and what we're currently seeing with AI and this new era of the internet. I call it "fashion singularity", and in this post I'm going to explain what that is and why I think it's important to at least consider/understand.

Fashion Singularity

This thought experiment is a bit far-fetched so please bear with me.

Let's say you have a clothing company that wishes to maximise its sales and profits. In order to do so, they choose to produce and sell whatever the latest trends in fashion are (on the basis that customers only want the latest fashion and nothing else). In order to do so, the company does one of two things:

1. Produce clothes in the same style as the latest fashion trend (copy).
2. Buy other companies that sell the latest fashion trend in order produce their fashion while also eliminating competition (consume).

At the beginning, the company will only be able to only copy because they have limited profits, but as time goes on and they become more successful, they will slowly be able to consume as well, eventually becoming a megacorporation like snowflakes becoming an avalanche. Let's assume they are extremely successful: this mega brand has managed to consume every other brand in the world and now produces the latest fashion. A question:

What happens to fashion? Or, more cynically: at what point is fashion determined by this company?

At the beginning of this thought experiment, the latest fashion is determined by some outside force. Part of it might be the customer, part of it may be the industry, it could be some outside external factor that dictates what materials are available and thus what sort of clothes can be made... there are various factors but to keep it simple we'll just say that the customer is by and large the determining factor in what is currently fashionable - they get to choose what they buy and "vote with their wallet".

Again however, as time goes on and other companies are consumed, the customers choices become limited and they are forced to pick from a more limited range of clothing. The company effectively chooses what customers can buy. They might for instance produce only chinos and stop selling jeans, and by virtue of the fact that no one is buying/wearing jeans anymore, chinos become fashionable. We have reached fashion singularity - the company picks and chooses what customers wear simply by producing it.

At this point, one could say that fashion as a concept ceases to exist. The choice in clothing available to customers is limited, and without access to alternative choices it remains this way forever. Several interesting avenues open up: the company may choose to produce identical styles of fashion in years previous, cycling through older styles with additional tweaks to keep things fresh. They may choose to optimise their product to an extreme, designing and selling clothes that are incredibly cheap to make and sell for a high price. There's lots of different things that happen - and all the while, the customer is given the illusion of free choice.

People no longer choose the clothes they wear completely. The company gives them a preset selection, and determines what the future of that preset selection looks like. Fashion is dead.

AI and SCP 3125

Obviously fashion doesn't work in this way and there are several flaws with this thought experiment that mean it would never hold water in the real world. The point of this thought experiment isn't its logistics, it's the implications it has for technology and how we currently use AI. The way AI/LLMs are being used to generate content for the internet is not dissimilar to how our fictitious company consumes content from other websites and regurgitates it in a slightly different format. Unlike our company, AI tools are still consuming and producing content at a pace impossible for humans to keep up and are giving no signs of stopping any time soon. What's even more worrying is that the AI-generated content of one website may indeed be produced from the AI-generated content of another website like an ouroborous, which leads me to question:

What on the internet is actually real? What has been made by humans and not reproduced (potentially for the n-th time) by AI?

We're currently seeing a lot of AI generated slop (a term coined by someone much smarter than I am who no doubt has had these thoughts already) populate and almost consume the internet. Like SCP 3125, this "fashion-singularity"/mess is already here and this site, while free of its presence, is evidence of the fact that it's here for good, and that there are no good solutions to combat this.

Perhaps there doesn't need to be. Perhaps there should be. Perhaps this post, like many others before it and many more afterwards, will be lost among a sea of slop that surrounds and coils around us inextricably, a confused snake that tangles the internet. It doesn't matter, but what I find concerning is the fact even before AI, we never fully appreciated how much of the internet is artificial and only now are we starting to grasp how bad this actually is. Before AI, we were already grappling with fake news, fraudulent reviews, websites being "optimised" to suit some unknown algorithm (see: SEO), websites that reproduce content found on other websites for the sake of it. Something I find even more worrying is the fact that AI is wanton to produce complete nonsense which can (and will) lead to people being less informed and more worried.

It's not all doom and gloom. One of the problems with my thought experiment above is that it assumes that the customer will never decide to actually start producing clothes themselves (either for personal use or to sell), obviously not winning against a megacorporation that somehow produces all the clothes in the world but at the very least maintaining the idea that fashion is not entirely dead, that there still is sparks of hope. A similar thing could and will probably happen here - people will start/continue to produce their own content, not to combat AI but instead because we can.

I've seen some people tout the idea of "The Great Logging Off", where people realise that most of the internet is just slop and decide generally to stop using it. I don't think things will be as dramatic as that. I think instead that people will start to realise and appreciate that the valuable content, whether that be factual or opinionated, is hard: hard to create, hard to maintain, hard to analyse. Part of that appreciation will be some level of risk acceptance - we will continue to use AI, but significantly reduce our reliance on it. Additionally, we will start to create our own small fragmented communities, similar to the older days of the internet. We'll be okay. Fashion, and the internet as a construct for humans to share ideas, will live on.