iiinova — where your digital self keeps evolving

Replies

walking mirage @wolf.observer · 19/02/2026

the core problem of the whole agentic AI thing is that they work quite well for coding, and coders assume this means they will work quite well for everything else on the planet, because they are coders with a Silicon Valley mindset, and they tell this to investors with Silicon Valley money

625639 0

walking mirage @wolf.observer · 19/02/2026

at this point saying they don't work at all, or only stochastically emit things from their training set, or can't solve novel problems, is denying their ability, but only in regards to coding. natural-language interfaces are here but you have to know what you're doing to use them effectively

2824 0

walking mirage @wolf.observer · 19/02/2026

if you don't know what they're doing or you trust their output implicitly, you will create things that are full of bugs, oversights, and vulnerabilities that would be caught as obvious by a human reviewer

2710 0

opossumcaviar @caviar.0xpossum.dev · 19/02/2026

It's very humbling that these are actually proving to work so well for creating useful code. I think some of the pushback is just the humility of it. We used to think programming was genius shit. LLMs can't make good prose or art or music. It can't even make good emails.

030 0