Curation vs Creation of Content

Ranty shower-thought time...

Before the Internet, the media was our prime source of information. The media filled several roles, they chose what to write about (selection), they prioritized the importance of the topics they selected (curation), they created stories (creation) and they published and sold them (distribution). Within creation, what was Opinion, Reporting and Journalism was reasonably easy to figure out... often just from what part of the newspaper it was featured.

Now... almost 30 years after the Internet brought us online news, I still don't believe the media industry fully understand that Selection, Curation, Creation and Distribution. What more... I don't think the Tech Giants do either. And really understanding what machines are good at... and what humans are good at when it comes to content is a key survival skill in this age of AI.

Selection is the skill of choosing what to write about (or create content about). Any AI that I have seen needs prompts for it to generate content. And what it generates is based on the vast amount of data that it is trained with. Thus, it creates within a box where boundaries are existing published knowledge. Think about it like this: AI will always select sideways - to create either a remix, a retelling or an incremental improvement (along some dimensions, not all) to an already published thought. If I tell generative AI to create a million fish dishes, it would never come up with the concept of serving a poisonous fish with just enough nerve poison left to numb your mouth and change the experience of eating - unless that fugu fish dish was already part of the training material.?

Curation is the ability to prioritize content for a particular purpose. This was a large part of the media business when print was king. What makes the first page, what's going to be today's opinion piece. Curation has largely been replaced by personalization algorithms. Social Media is the new curation. Which is starting to become a problem... or at least, a point of pretty heated debate. I personally had a period where I commented and engaged with content I did not agree with. My thinking was that it would allow me to understand better the people around me. The Facebook algorithm took this as a sign that I liked that content better than other content, so in a matter of week my Facebook feed was filled with pro-trump memes, anti vax, Q anon conspiracies and it just made me angry. So I quit Facebook. Because the curation sucked.

The more of the same is the Achilles heel of algorithmic curation. Who hasn't been bombarded with ads about vacuum cleaners, just after you bought a vacuum cleaner. Any intelligent system would try to sell you vacuum cleaning bags, or articles about how to vacuum regularly. The point is that curation algorithms are purely based on the understanding of the content and not any understanding of the world. So it doesn't offer a pathway to improvement. If I watch a bunch of YouTube videos' on Chess Openings, when I get to the point where I have watched videos' on the most common ones, the recommendations are filled with videos' on the same openings. Not a single place is there a video about mastering the middle game of chess, which would actually be the logical next step for me to watch. Because algorithms curate, again, sideways. Without any topical understanding, all it recommends is similar content, and not the most likely content that would add to your knowledge. By design it priorities reinforcement of existing thoughts over diversity, contradictory perspective or advancement. Imagine if YouTube showed you categories of recommendations: watch this video for a rebuttal, watch this video for a more moderate view, watch this video if you want to learn more, watch this video if you want a different take... no, it's all just different takes on the same core content. It can't do the categories, because the algorithms lack the ability to categorize into the kinds of categories I mentioned (AI is very good at categorizing, but not necessarily all types of categories).

Third is Creation. No doubt that Generative AI is going to save us so much time during the act of creation. But keep in mind that creation is not the writing. The writing is the forth thing media does, which is distribution. No, creation is to come up with something novel, something new, and then forming that into an article, a tweet, a YouTube video or whatever other format. I believe Generative AI will be revolutionary in our ability to distribute our unique thoughts. Take my article on Selection, Curation, Creation and Distribution of content and turn it into a YouTube video, please. But the core human addition to this is the framework. The ability to take a concept like Content and break it into further sub-symbols, explain the relationships and changes between those sub-symbols and how it aggregates up again to the original concept of Content. That is uniquely human. AI doesn't have original thought... except accidentally... and only then it would take a human to identify it as actually useful.?

This rant is actually key to understanding the nature of Generative AI. It is not autocomplete of your thoughts as some people claim, it is more an ability to convey your thoughts quicker, in more formats, in more ways. But the original thought and concept is still the human domain. The challenge is of course that these unique thoughts is very likely to drown in a sea of reiteration, remixing, restatements of the same old content - made possible with Generative AI. Content that might not even be true. Meme factories or algorithms creating millions of versions of the same false thing and throwing it at the giant wall of the Internet to see what sticks. The key to solving it... it's in Curation. It is to make those algorithms more intelligent. Instead of recommending, it should curate, and then maybe new and exciting content could actually stand out. Because if we can't identify and highlight content that is actually new, actually novel, actually adds to the human knowledge base - from content that is just the old in a different way - then we will drown in our own sea of self confirmation and easily digestible top 10 lists... and, btw, you won't believe number 7.