Although execline is pretty cool.

So maybe I'm done being polite about this. Stop worshipping the damn golden calf and help each other.

3. It is utterly foolish to invest resources into AI while any human has not yet reached their full potential. 1 in 8 people alive today live in absolute poverty.

2. The marginal value that trained models have over engineered heuristics cannot even remotely justify the costs of building them.

Oh I agree. People talk like everything is about money when it is about political power and the fact that too few fallible little men and their enablers control too much.

We can upend the scam by rejecting their tokens and directing our labor into our communities. Become ungovernable.

Musk's history with PayPal tells the real story: a desire for unregulated cross border payments. It's techno-libertarian capitalist nonsense.

Fortunately it's a house of cards. Tokens only have only the value we assign to them. People can just refuse to play the game.

That certainly is not new. So many crypto projects had "founders fees" or "development fees" baked into their minting process. ZCash devs took 20% of every block reward.

Building systems to be safe "from the ground up" is called coinductive reasoning and has been an area of research since before the first computers were built.

Almost no systems in use today have even been proven to do what they claim to do. AI grifters are building rockets made of jello.

And when we don't learn that nuance, we hire 250 non-engineer coders to repetitively perform tasks that could have been automated.

If a pattern is common enough for an LLM to encode it, then it's worthwhile to abstract into a library, or at the very least a template. Choosing the appropriate abstraction is a task for humans because it has long term consequences.

We don't learn that nuance by querying tokens.

I think there's something lost in training people to rely on (sloppy) automation when something feels repetitive.

There's that Bruce Lee quote: "I fear not the man who has practiced ten thousand kicks, but rather the man who has practiced one kick ten thousand times."

It is superficially useful for anything that has been done 1000 times before.

The fundamental problem is that programmers shouldn't be doing things that have been done 1000 times before.

If they are it's bad management. LLMs are a cover for bad, overpaid managers.