The AI Empire

Hey friends 👋,

In Kurt Vonnegut's 1952 debut techno-dystopian novel, Player Piano, there is a world-class machinist named Rudy.

One day, a group of engineers and researchers visit Rudy in his shop. They ask him for permission to record him while he works. They show him a device they've made that can store human movement on magnetic tape. They tell Rudy they're working on a machine that will make his work easier, safer, and more efficient.

But this machine needs to be trained, they explain to Rudy. And they need it to be trained by only the finest machinists in the country.

Rudy – a man with a gruff, greasy exterior but a fluffy, soft spirit – melts at the flattery. He agrees to help the team and spend some time being recorded. When he goes home, he proudly tells his friends and family about his selection and his excitement to be part of the project.

Over the next few weeks, the engineering team visits Rudy on multiple occasions. Rudy is generous with his time, sympathetic and patient when the equipment fails or there are issues with the recording and they need to restart.

After the last session, the team hands Rudy a beer and toasts to him – thanking him for being part of this hopefully historic moment.

A few years later, Rudy is out of a job. The machine the engineers developed by recording him and dozens of other top machinists doesn't require any machinists to operate.

The entire plant can be run by a single engineer. The work is easier, safer, and more efficient. It's just not work Rudy is required for anymore.

Obviously, there are parallels in this parable to OpenAI and the creation of LLMs. For the better part of two decades, we were told that having our online behavior tracked, analyzed, bought and sold for advertising was the cost to access and maintain the most powerful and empowering communications technology ever to exist – the internet.

But now, just like Rudy, we're seeing that this wasn't a fair deal at all.

The underlying technology LLMs use to output text is interesting and novel. But what makes this technology valuable is all of the data these technological juggernauts have collected or commandeered from us.

Without our data, our work, our knowledge, products like ChatGPT would not exist. We created these tools just as much as Anthropic or OpenAI did. And yet, the companies insist they have the sole right to credit, compensation, and control.

In her new book, Empire of AI, Karen Hao – a tech journalist for The Atlantic and The Wall Street Journal – makes the case that this behavior isn't just representative of late-stage capitalism, but early stage technological colonialism.

OpenAI, Anthropic, and Google are laying claim to resources, rewriting well-established legal structures like copyright law, and making threats of economic violence to force us – the internet natives – into submission.

Well fuck that.

LLMs are a powerful and empowering technology. But we must remember that these machines are powered by us – the people – in real tangible ways. And we have the right and responsibility to demand ownership and control over that power.

The valuations of these companies represent potential gains that should reside in our hands more than theirs. We just have to fight to hold the bag.

Until next time,
Drew