From the Zed Blog: Writing code is only one part of effective software engineering.

Zed's platform  offers insights and resources on technology, software development, and entrepreneurship. Through articles, tutorials, and personal stories, Zed shares  knowledge and practical advice for individuals looking to succeed in the tech industry. Readers can learn about startup culture, product development strategies, and career growth opportunities to navigate the complexities of the tech landscape effectively.

Large Language Models excel at writing code but struggle with the iterative mental modeling that defines effective software engineering. While LLMs can generate code and update it when given specific problems, they cannot maintain clear mental models of requirements versus implementation, leading to confusion when tests fail or debugging is needed. Current models suffer from context omission, recency bias, and hallucination issues that prevent them from understanding complex software systems. For non-trivial projects, human engineers must remain in control, using LLMs as tools while maintaining responsibility for requirements clarity and code verification.

Why LLMs Can't Really Build Software — Zed's Blog

This take is better than most.
There are so many bad takes out there, someone with a psychology degree, someone who has done fronted work, someone with an MBA, thinks they have the gotcha or even worse they are a linkedin poster, a substack pusher, an wanabe influencer or a thought-leader-poser looker for their first big break.
At the end of the day NONE of these people even know what an attention head is or can explain the FUNDAMENTALS of how their critique relates to a deep understanding of the technology and can prove that this gotcha isn’t cooked up–one part background bias, one-part superficial idea and a dash of hopium. I see this all the time!
This article does suffer from the same issues: pop our mental model onto the stack? just put more words into the context window? When I highlight some code and as it a question about that portion, the model is at least behaving like a human, ‘popping’ our mental model off the stack (not the proper psychological understanding I’m looking for, but I get it).
So it still doesn’t really show me how a proper understanding of these models relates to the critique. That said, it is getting at something that I think is right: above some complexity the models do fail. I just wish SOMEONE would show me why based on the architecture rather than observe and hope.

Really? Algorithms from the 70s can’t fully build complex software projects?
<img src="https://media3.giphy.com/media/v1.Y2lkPTc5MGI3NjExNzRjdzlhOHVnbjVhejRxN2VkbHpqdDAzamRoMjc0Nm1kMXpsN3IyNSZlcD12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/mjMuUOOw8QxFK/giphy.gif" alt="">