The AI-: what owners actually say

Multiple commenters call out the gap between claims and delivery. Anthropic's Mythos was described as 'full with marketing in the guise of trying to save the world' with a 'smells really really weird' vibe — a model so allegedly powerful it can't be released. Apple's AI features were cited as barely special while sales fell worldwide.

Multiple highly-upvoted comments state plainly: 'Nobody ever ask for or give a shit about AI on phone.' iPhone buyers in China care about status, not AI, and you can get 80% of the functionality for 30% of the price from Chinese brands.

YouTube is described as heading toward being '100% over-run with absolute garbage' with AI-generated videos, thumbnails, and comments, making it impossible to determine what content is real. Amazon selling AI-generated mushroom identification books was called 'a disaster waiting to happen' and 'a truly terrible idea.'

One commenter raised the 'automation bias issue' as something that has been 'mostly ignored' — the better models get, the worse this problem becomes. Air purifiers were used as a parallel: creating 'a false sense of security while doing only part of the job.'

Developers report AI-generated code is identifiable by its comments ('You used to be able to tell an image was photoshopped because of the pixels. Now with code you can tell it was written with AI because of the comments'). Extension and open-source maintainers insist submitters must 'justify and explain the code they submit' — if they can't explain why any of the code exists, 'it needs to go in the garbage.'

Multiple commenters acknowledge AI is actually enabling non-programmers to write full applications. 'Non-programmers are writing full apps. Sure, they're simple ones, often just CRUD and UI, but it actually works' — distinguished from previous no-code/low-code waves that didn't deliver.

The Sign Painter short story generated by Claude was praised as genuinely enjoyable, featuring nuanced character work about a craftsman who loved his work but was angry for other reasons. A clinical psychiatrist's external assessment found Claude had 'excellent reality testing, high impulse control' with affect regulation that 'improved as sessions progressed.'

Mythos found a critical OpenBSD vulnerability plus 'several dozen more findings' across a thousand runs for under $20,000. Credible independent security experts have noted a 'notable uptick' in 'actually valuable AI-assisted vulnerability reports,' lending credibility to the claims.

Laser-equipped robots that use imaging and machine learning to identify and destroy weeds were described as a genuine application. Seven-digit price tag but payoff for farms of a couple hundred acres. Multiple commenters agreed 'agriculture is a perfect industry to make more autonomous.'