How to Get Skills That AI Can’t Replace
There’s a lot of AI push these days, and “what if AI takes my job?”. We should want to stay relevant, especially in the tech sector where AI appears good return on velocity. Not investment. Velocity only.
AI (actually LLMs), fundamentally, is a math trick. It’s taking vector math and dice rolls to generate a string of next probable words.
It feels like intelligence, but it’s really exploiting humans’ inability to process massive amounts of data. Humans are essentially giant aberration detectors (i.e., we’re really good at comparing things, but not massive data).
Consider your local car dealership mile, running for years. How are there always people going in? How do they stay in business? It feels weird, right? There are a lot of people. A LOT. It feels weird because humans are horrible at knowing quantity over time. We get around this by squashing time into immediacy with things like math and spreadsheets and writing things down.
This is what makes AI look intelligent. It’s a math trick, we’re “unraveling” the math that we’re objectively bad at and putting it into an abstraction (words) that makes the underlying trick seem impossible. It’s hinging on the same emotional connection that humans have to math (i.e., none) and exploiting it.
We’ve essentially wound up with a street performer pretending that he can generate rabbits out of hats, and people saying “infinite rabbits! world hunger solved!” without pausing to think that A) you can’t actually only eat rabbits (see: rabbit starvation) and B) there aren’t actually infinite rabbits in the hat.
This narrative will never stop being sold and the bets/debt will escalate until something explodes catastrophically.
So what? We have some orgs that blow up because they didn’t think they needed to plant crops or hunt real food because “infinite rabbits”. When these orgs fail, the problem will be placed on the fact that the entire population of rabbit-eaters died. Not why they died of rabbit starvation, but just they died.
As in, people who relied solely on AI let their Actual Intelligence atrophy.
“Bad workers, bad hires, I guess.”
The velocity is so high with AI it will take ages for execs to troubleshoot the real problem. Orgs that rely on AI are built on dice rolls.
What’s the antidote? Stay creative. Figure out why things work, not just how to make them work. Anyone can operate. Architecture, causality, making genuine sense out of data - these are things AI will never be able to do. It lacks the creativity to see new aberrations. But that’s what humans do. That’s what we’re good at.
AI is a velocity producer. That’s it. It moves fast. It’s copy pasting from stack overflow at 100x speed. That’s great, if you know what you’re doing.
Still don’t believe me? Try this - if you have a hobby that you’re really knowledgeable about, or have interesting or edge-case ideas about - e.g., a video game, movies, academia, whatever - have a chat with AI about it, and really get into the weeds. Read its responses very carefully. You will probably find that if you didn’t know the things you know, you would have misunderstood or even learned something entirely wrong.
What if your query to AI was about something with high consequence and you blindly accepted it?
If you keep repeating that cycle without judgment, that’s a speed run to Idiocracy.
That’s a race to the bottom in the worst way. Not about money, but about genuine creativity and intelligence.