Social Media 2010 :: AI 2025?

I first joined social media somewhere around 2008. It was amazing. It helped me rekindle friendships with people from past chapters of my life with whom I had lost contact for years or even decades and make friends with people all over the globe who I would never have met otherwise. We shared stories of births, deaths, anniversaries, travels – in other words, all the things that make life … life.

There was a heyday somewhere around the early 2010s in which that joy of connection – with people we knew – kept us active and engaged.

We mostly didn’t think in those days about how these sites were funded. We just enjoyed them. (OK, some people did think about the costs. Anyone remember the petitions that went around begging Facebook to never make us pay to use it? Good times.)

Eventually, of course, the bill came due. Investors wanted returns (that is how it works, of course) and social media has, in recent years, become largely devoid of its initial joy and purpose. On a good day, maybe 10 or 20 per cent of the content the algorithms choose to show me is what I originally came for: genuine updates from people I care about.

The drive for profits and attention has led the social media sites to pump our feeds full of high-sugar content: politics, tornados, AI slop, advertisements for things we really don’t need, and on and on.

Cory Doctorow wrote about this eloquently in his 2023 article Tiktok’s enshittification and it rings true for me. I’ll quote his introduction:

Here is how platforms die: first, they are good to their users; then they abuse their users to make things better for their business customers; finally, they abuse those business customers to claw back all the value for themselves. Then, they die.

I call this enshittification …

Mmm hmmm.

But What About AI?

Fast forward to today.

Generative AI is reshaping entire industries. You can create professional-looking photos, videos, text, whatever from the simplest of prompts at little or no cost.

Important disclaimer: I work for a company that is active in this space, but I’m not directly involved in our AI efforts. What I’m writing here is my own opinion, informed by nothing more than what is announced publicly by my employer and others.

After a period of reluctance and skepticism, I too have found utility here. I’m a software developer and AI tools are surprisingly good coding partners. I still prefer to think for myself about the overall design of a system, but there are places where I’ve found AI far outperforms me.

It’s especially good at explaining to me how code works in other software languages. As an example, I’m working on an open-source Rust project to parse the AsciiDoc language. The best current implementation of an AsciiDoc parser is written in Ruby. I can squint and figure out some Ruby, but tracing through several layers of abstraction in that code is a slow exercise for me. I’ve been quite pleased with the results when I ask AI “Show me where the ___ is parsed and explain how that works to me” or “given the following Ruby regular expression, please write an equivalent regex in Rust.”

So here I am, like so many people, being seduced by the productivity of AI.

Meanwhile, AI providers are building data centers and consuming GPUs like there’s no tomorrow. Some of us – myself included – are paying a little for some AI tools, but I don’t see anyone paying new-data-center money to anyone.

🤔

Déjà vu, anyone? Certainly so for me.

I can’t help but wonder what happens when the mortgages come due on those shiny new data centers.

What will the end-of-honeymoon days look like for AI?