Marco Arment

All right.

And so some more from John.

Initial velocity gains disappear in the first few months.

Change failure rate is up 30%, and incidence per pull request is up 23.5%.

Then another article, Comprehension Debt, The Ticking Time Bomb of LLM-Generated Code by Armin Shakar.

Comprehension debt refers to the gap between the code generated by LLMs and a developer's understanding of that code.

In this post, we will explore the implications of comprehension debt, its impact on software quality and maintainability, and strategies for mitigating its effects.

And then finally from John, Your Brain on ChatGPT, Accumulation of Cognitive Debt When Using an AI Assistant for Essay Writing Tasks by Natalia Kosmina.

or something like that, and others.

Anyways, while LLMs offer immediate convenience, our findings highlight potential cognitive costs.

Over four months, LLM users consistently underperformed at neuro-linguistic and behavioral levels.

These results raise concerns about the long-term educational implications of LLM reliance and underscore the need for deeper inquiry into AI's role in learning.

Okay, fellas, thank you for being patient.

What do you got?

A little bit.

So earlier today, I got my keyster to properly start working on the iPad layout and call sheet, which has been trash since call sheet was around.

And it's been too long.

I should have already taken care of it, but I hadn't yet.

And my first cut at this was, all right, let me use the 26.3 beta and let me ask, you know, the Xcode agentic stuff via Claude.

And I don't recall what model I was on, and I should have paid attention.