Today I learned about technologies being separators. I watched a fantastic video that only has a few views and subscribers, and I think it will explode soon. As of Sunday May 17th, it had 3.5K views and 205 subscribers. I refreshed the page an hour later and it was already at 600 subscribers.
My main point is in that video. I just wanted to combine it with my own interpretation.
It's hard to know where to start, but I like one conversation I had with Gemini. (I deleted my Pro account and now all my conversations are gone, so be careful.)
It went something like this: why do humans think so much in dualities, when things are more of a spectrum? Part of the answer was: because of boundaries. A thing is a thing because it is not another thing. If it's not that thing, there's a boundary. Crossing the boundary blurs the line between one thing and another.
An easier way to picture it: boundaries wrap stuff. A boundary is a bucket of things. It can feel like one thing in our mind, until we learn we can pull something out, and the thing we pull out becomes its own thing, while the bucket has less in it.
Technology helps us get things out of buckets. It's a separator.
Writing did this to knowledge. Before writing, knowing something required remembering it; knowledge and memory were the same bucket. Writing pulled them apart. You can now know something without remembering it. Just read it again.
The pattern repeats:
- GPS split getting somewhere from knowing the way.
- Refrigeration split eating food from making food.
- The telephone split a voice from the person speaking.
- Batteries split energy from its source. Before batteries, heat meant fire, grinding meant a windmill or flowing water, light meant a flame in the room. Energy and the source of energy were the same bucket. A battery lets you store energy and use it later, the same move writing made on memory.
LLMs are doing this to intelligence and identity.
Google already blurred the line. The old "Big Brother" jokes were really jokes about an entity that knows things no human could know. LLMs went further. They produce expert-tailored answers on demand, but there's no human behind the expertise. The "you are a helpful assistant / doctor / dream interpreter" system prompt is a mask, not a self. Identity is performed; intelligence is real output.
Before this, accessing expertise meant engaging with a person who carried it. SaaS came close. It's codified expertise inside a UI, but no one called Salesforce an expert. Experts were the humans you needed in order to build the SaaS. LLMs collapse that. Claude Code writes better code than most working programmers, and the same trajectory applies elsewhere. Expertise is being lifted out of the people who used to hold it.
Other things AI seems to be separating from their usual containers: effort from work, humans from relationships, language from understanding, authority from institutions, listening from a listener.
The practical takeaway: when you find yourself arguing "it's not really X, it's actually...", check what bucket you're defending. You're rarely defending the whole bucket. You're defending the specific things inside it that the new technology is pulling out. Naming which things makes the argument honest, and usually more interesting.
Prompted by this video, expanded with my own framing.