Spectrum of Sameness

By Amina Crooks

I am in the camp of AI optimists, believing that, while there will be some cases of misuse, the majority of new advances and developments will enhance and save human lives. I do however; have some niggling concerns as we embark on this journey through the era of AI and learn to use these new AI tools. Like many people, the unknown that concerns me, is whether or not, we will be able to develop timely and sufficient guardrails to protect and conserve core values and ensure we move in the direction of creating greater equality and diversity.

One of these niggling concerns is whether we will be able to maintain the generation and propagation of new and creative content in AI models used for decisioning. Will creativity become a marginalised commodity as we shift to fast and formulaic derivation of content — replay rather than remix?

Crazy right? I can imagine what you are thinking. Why in the world would this be a concern? The world is experiencing a profileration of new and shared content and GenAI as a tool is accelerating the development of new images and text. So let me explain the kernel popping in my brain. The seed of a thought that may or may not be a legimitate concern and then you can be the judge.

The “Spectrum of Sameness” is my way of describing the creation of products that are seemingly diverse but derived from the same foundational elements.

The spectrum of sameness is the current path we have been on since the industrialised revolution. The most obvious examples are in consumer goods for food and clothing. The mega companies that swipe up smaller businesses and brands grow larger, often leaving consumers with the illusion that there is an abundance of variety and a spectrum of choices; whereas, in reality, it all comes from the same few sources.

The creeping up of homogenity into our lives and specifically diet is described beautifully in Dan Saladino’s book “Eating to Extinction, where he asks us to consider these facts:

“The source of the world’s food — seeds — is mostly in control of just four corporations; half of all the world’s cheeses are produced with bacteria or enzymes manufactured by a single company; … most global pork production is based around the genetics of a single breed of pig; and, perhaps most famously, although there are more than 1, 500 different varieties of banana, global trade is dominated by just one, the Cavendish…”

The spectrum of sameness alludes to the misconception of diversity. One would not be alone to point out that most of us we eat a greater variety of foods than our parents or grandparents. Regardless of what city in the world you visit, sushi, curry, and gelato are easily attainable within a single day. Until you realise it is the same kind of ‘diversity’ that is spreading globally. According to Saladino, the human diet has experienced more change in just that last 6 generations than in its roughly 40, 000 generations of existence. We have gone from consuming more than 6, 000 plant species to roughly just 9 (! yikes) worldwide and what’s more, 75 per cent of all the calories that fuel our species comes from rice, wheat, maize, potato, barley, palm oil, soy and sugar. What if the consolidation of consumer goods and manufacturing that occurred in the past few decades with dietary and physical commodities is replicated with content?

What if the current path of content generation at scale focuses too much on appealing to the mainstream and so uses the same sources to optimise efficiency, productivity and scale to such an extent that it perpetuates regurgitated content at mass and marginalises new content that is on the fringes? We become inundated with seemingly new and diverse content which is in fact versions of the same thing, manufactured from the same sources and same foundational data. Fillers but not nourishers. Noise but not knowledge. Brain dumps but not brain pumps. Ok, you get the jist. I will stop while ahead.

At the moment, AI is generating a massive amount of new content in images and text, but all of that is based on the data ingested thus far. How do we ensure new sparks of creativity emerge from human invention and experience? How do we ensure these get picked up and added into the AI melting pot? Will we face increased inequality in access to diverse content? Is it just me, or is the already happening? I often feel bored by seeing the same streams of content in my feeds. A similar bloating effect inflicted from eating too much food that looks different but all tastes the same and leaves you feeling unsatisfied and yet wanting for more. But more of what? The same or something different?

Instead of body obesity, we have brain obesity.

I am not proposing that new content will not continue to get generated. I believe we as humans have an innate need to create and express ourselves. The web and many social apps has played a pivotal role in lowering the barrier to the entry and circulation of new content, especially content of, and to, fringe and marginal populations. But is this the content that will be produced on mass by humans and bots “optimising for eyeballs”? Data is often compared to oil, but unlike oil, data is infinite. That said, the quality of data still matters and the lifecycle costs (i.e. storage, processing, training, etc) are still high, requiring specialised knowledge banks to be developed for specific use cases.

The question remains, what curated data will go into those future models that are used by mass entities for decisioning and will the data be truly diverse, adequately refreshed and optimised for what?