Most systems to write code with AI use "AI-driven engineering." That lets the AI make too many decisions and this has led to the proliferation of "AI slop." This article introduces a free online-training I am offering to teach "engineering-led AI" using PAAD (https://github.com/Ovid/paad). You can still take advantage of AI, but now you have control of quality instead…
Code production was never the bottleneck in software development. This is why making it cheap doesn't do what the replacement narrative claims it does.
The arXiv Explorer uses a novel recursive language model technique to allow you to use a normal AI-chat to query megabytes of arXiv documents at a time.
AI has entered an era similar to the internet near the end of 1996. Basic use cases were solid, advanced use cases were not, and money was flushed down the toilet.
Many people don't write alt tags for images, or they write bad ones. However, these are crucial for people using screenreaders to understand the image. AI can make this easier.
Wardley Maps are an amazing business tool for planning, but you can suffer information overload when you read about them online. Read this quickstart to get up and running with them.
I've often noted that many objections to object-oriented programming go away if the objects are immutable. But sometimes, you want them to be mutable.
I'm still regularly amazed at how many developers dismiss Copilot and ChatGPT as 'stochastic parrots.' These parrots are much smarter than they think.
People keep asking why the Perl language doesn't just drag the Moose module into the core for object-oriented programming. Buckle up, this will be a long trip ...
When I get called in to fix a client's software, I find that most large codebases have common errors. Here are some of them and how to avoid them.