Table of Contents
The more I read about AI generating blog posts and other content, the more I feel that we should write for ourselves, rather than our readers. I noticed, recently, that no one is reading my posts. There could be three reasons for that. The first is that what I’m writing interests no one. The second is that when I disconnected Jetpack I broke integration. The third possibility is that posts are being read, but because they are federated I get no feedback on audience.
Inconsequential
Whether I find readers or not doesn’t matter. What matters is that I invest time in thinking, and in elaborating ideas on a daily basis. I write about projects I am working on. It’s also about looking at a blank page, and trying to find something to elaborate on and write about.
This morning I saw an article about how human writing is hard to distinguish from AI.
WordPress wants AI to generate posts but humans are already overwhelmed with content. We’re so overwhelmed with content that people use clickbait to emotionally blackmail us into following links. With AI that problem will be made worse, rather than better. In the age of content farming for books, blog posts, and articles we’re heading towards an age where the fediverse is more important than ever, and where word of mouth is key.
Humans as Authenticators
Galaxus have an article about an AI female creation and my reaction to this is that in the age of algorithm driven antisocial media feeds AI can boil to the surface and remain there. It is for this very reason that antisocial media no longer has any value. In an environment where AI creations rise to the top there is a profound and chronic social disconnect. That social disconnect is why antisocial media companies like TikTok, Instagram, Threads and others are undercutting their own niche, rather than increasing interest.
I see value in social media, while they are social networks. Once they stop being social networks, as is the case with FB, Threads, Twitter and even BlueSky, they stop having value for my desired use case. I use these sites to connect with people. If I feel no connection, then I have no reason to stick around.
In a healthy social network, AI “personalities” would be a curiousity, without followers. In an unhealthy antisocial network people follow an AI bot.
If you don’t expect a social connection with the accounts you follow then you are missing out. I want lively conversations, not unrewarding scrolling. I stopped scrolling for conversations during the pandemic, when I realised that conversations are dead.
Doesn’t the Context Make Blogging Absurd?
That’s the key. AI generated blog posts, that no one will read are abusrd, but human written that no one will read do make sense. I’ve invested my time into developing and expressing my ideas. I’ve invested my time in thinking an idea through. I’ve also invested my time in not wasting it.
If you scroll on Facebook, it will pause your scrolling for five seconds to show an advert, as will Instagram. Both of them are pushing me to scroll through content I have no social or emotional connection with, to show ads for products I never intend to buy. I gave my time to social media because there was a social reward. Before the pandemic that social reward was, gone, and it’s even worse now. Without a social reward social media is dead.
Scaled Up To Novel Proportions
Now, if Instagram is a waste of time, for showing a ninety second video, and twitter is a waste for showing 140 characters (number used for the sake of the argument) then what motivates us to invest half an hour to an hour listening to an AI generated podcast, or six or seven hours reading an AI generated book? The answer is ‘nothing.’
And Finally
If I had asked an AI agent to write a blog post today, I would have provided two to five prompts and it would spit out an answer within seconds. In contrast it took me hours. The key difference, and this is painfully clear, when code pairing with Gemini, is that AI doesn’t understand what it’s doing.
To be specific, Gemini wrote a test, and then it re-wrote eleventy.js. The code that was optimised for the test, broke the logic that Eleventy relied on to work. Gemini was happy to have me try the same thing 20 times. Gemini is like Number Five from Short Circuit. It just runs programmed.
This applies to a blog post, an article, videos, or a novel in that AI can generate a photo, or video, or text, but it has matched patterns and used probability. It doesn’t understand what it has created. It doesn’t feel.
Garmin Connect+ illustrates this with the most recent summary. I found myself wishing that I could write to offer more context that is relevant, but because I can’t it gives general feedback, rather than context based feedback.
That’s why the blank page, and human creativity have value. We’re slow, but the result is based on human perception, rather than machine learning.
Leave a Reply