Why PKL exists

PKL began with a book. I was concerned about the effect social media platforms had on critical thinking, in particular with the way teens were learning to think (or not learning to think). When I was about 100 pages into writing my book, I realized it was obsolete. The threat I’d been writing about was already being overtaken by a bigger one.

Social media platforms presented a form of information overload and encouraged us to react emotionally rather than logically. But they did not think for us. The introduction of ChatGPT and other LLMs did.

While I was not drawn in by the social media platforms, I was smitten with the LLMs. For someone who loved reading encyclopedias as a kid and spent hours on Wikipedia, LLMs were a godsend. You could pursue any curiosity by way of a conversation with an intelligent “being” that left the Turing Test a relic of a bygone age. Within a few years, the question moved from whether AI could imitate a human to whether models like Claude had a soul.

At the same time, LLMs are training us to think less as we outsource more to them. We can use and manage them well because we’ve already learned the skills we’re outsourcing: writing and critical thinking. But what about the generation native to LLMs? How will they manage an LLM’s output if they haven’t spent thousands of hours writing and thinking? Will LLMs become the oracle we trust without questioning? Will we even have the right questions?

My solution wasn’t to go Kaczynski and reject AI and technology. What is constructive about preaching against an inevitability? The better idea, I thought, would be to use AI to create an offramp from a future of idiocracy. To use AI to help us think better — to read widely, imitate deliberately, and understand through practice.