Your LLM Doesn't Write Correct Code. It Writes Plausible Code.
One of the simplest tests you can run on a database:
I Still Built It - Ewan’s Blog
if an LLM wrote a significant amount of the implementation, apparently the human directing it no longer counts.
Dan Shipper 📧 on Twitter / X
this is true and is a big reason why you don’t need to be a highly technical researcher to use LLMs in surprising and novel ways https://t.co/TuxNzXzToU— Dan Shipper 📧 (@danshipper) July 27, 2025
My AI Skeptic Friends Are All Nuts
My smartest friends have bananas arguments about LLM coding.
My AI Skeptic Friends Are All Nuts
My smartest friends have bananas arguments about LLM coding.
Designing a Language by Asking the Language Models — using an LLM panel as a syntax usability lab (from the kaish project)
Designing a Language by Asking the Language Models — using an LLM panel as a syntax usability lab (from the kaish project) · GitHub
Why your local LLM feels dumber than it is
Quick Introduction We have all been on forums, chats, reddit, discord, youtube, or somewhere and heard “Oh! Model XYZ is AMAZEBALLZ!zomgwtfbbq” then downloaded it (or more likely, some quantized form of it) and said “eww… This sucks!” This post is going to be a rather technical series of experiments to demonstrate the impact of implementation-specific hazards with inference. I will be using the term “reference implementation” to describe the lab that published and offers first-party hosting of ...
What sort of maths are LLMs good at?
For the sake of anyone who might read this blog post in the distant future (a month from now, say), let me mention that I am writing it a few days after OpenAI announced that it had solved ten majo…
Copyrightability of LLM-generated code: Can we license “vibe code” into Free Software? - FSFE
The use of large language models (“LLMs”) has exploded in recent years, including in the generation of source code. But even as their usage gains popularit...
llama.cpp with ROCm
WarningThis is a technical guide and assumes a certain level of technical knowledge. If there are confusing parts or you run into issues, I recommend using a strong LLM with research/grounding and reasoning abilities (eg Claude Sonnet 4) to assist.…
Understanding Reasoning LLMs
Methods and Strategies for Building and Refining Reasoning Models
elvis on Twitter / X
Diagram of the LLM Knowledge Base system.Feed this to your favorite agent and get your own LLM knowledge base going. https://t.co/4AQSFOv4PV pic.twitter.com/nPSNi4Ayqv— elvis (@omarsar0) April 3, 2026
LLM-generated skills work, if you generate them afterwards
LLM “skills” are a short explanatory prompt for a particular task, typically bundled with helper scripts. A recent paper showed that while skills are useful to LLMs, LLM-authored skills are not. From the abstract:
I haven't, but I don't know that it's that kind of model. It's really different than an LLM, since if you want an answer from it you can only ask it multiple choice, true or false ( with confidence ), or "tell me the score of X" questions. So there's no generating proof code or anything directly.
Sep 19, 2026 at 3:31 AM
I just came across this paper arxiv.org/abs/2509.22908 which coins the term "vericoding" for using LLMs to generate formally verified code, and presents benchmark results for several languages
A benchmark for vericoding: formally verified program synthesis
arxiv.org
Martin Kleppmann
New blog post! ✨ I argue that AI will make formal verification go mainstream. martin.kleppmann.com/2025/12/08/ai-formal-verifica… Three reasons for this: 1. LLMs are getting increasingly good at writing proofs using proof assistants. This will make formal verification vastly cheaper than it's been to date.
