Humanities mode collapse

I woke up this morning with a jolt. It was not a nightmare, but it was borderline, one of those "weird" dreams. I was in a world where everyone spoke in a very constricted vocabulary. Not only was the vocabulary constricted, but the ideas and thoughts themselves also seemed to sampled from a very homogenous distribution. Yet everyone was happy. They seemed to understand and agree with each other.

I feel we are seeing an element of this play out in real life. For now this is only in populations who are "AI pilled"¹ and therefore consuming huge amounts of writing and ideas from a very few sources (Claude & ChatGPT). These models write in quite specific ways, but they also seem to formulate ideas and approach innovation in consistent ways. I notice this especially when they propose new ideas. Often, these are combinations of concepts already discussed in the conversation, presented in similar ways.

I've also started noticing something similar in the startups people are building. The ideas, the products, and even the way founders explain what they're building seem increasingly alike. Of course, founders have always borrowed ideas from each other, and everyone is chasing similar opportunities. But I wonder how much of this convergence comes from using the same models to brainstorm, develop and articulate those ideas.

I don't have anything concrete to back this up yet. It's anecdotal, and something I'd like to explore with data in another post.

But back to the dream. If we're increasingly consuming ideas from the same few systems, are we slowly being trained to speak and think in a particular manner, shaped by a handful of powerful organisations? Are we living in Orwell's 1984², or Huxley's Brave New World³?

In 1984, Orwell introduces Newspeak, a deliberately restricted language designed to make certain thoughts impossible. As one of its architects explains, "the whole aim of Newspeak is to narrow the range of thought." Huxley presents a different, and perhaps more disturbing, vision. In Brave New World, people are conditioned from birth to accept their place in society, to desire what they are given, and to be happy doing so. As the Director explains, the goal is "making people like their unescapable social destiny." My dream seemed to combine both. Everyone spoke and thought alike, and nobody was unhappy about it.

The uncomfortable part is that I count myself among the AI-pilled. I think AI has huge potential to benefit humanity and make us more efficient. But at the same time, I worry about losing my own intelligence as the artificial version becomes both better and more abundant.

I recently wrote about a narrower version of this problem in Code at the speed of my understanding. We're becoming dramatically better at producing software, while potentially losing the very process through which we learned to build good software. I wonder whether something similar is happening to our ability to develop original ideas.

And perhaps that's the distinction we should be thinking about more carefully. We should be building AI systems that don't aim purely to automate our thinking, but to augment our intelligence in ways that elevate our own capacity for thought, rather than gradually diminish it.