<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Ai on dev.endevour</title><link>https://devendevour.iankulin.com/tags/ai/</link><description>Recent content in Ai on dev.endevour</description><generator>Hugo</generator><language>en-AU</language><lastBuildDate>Mon, 07 Jul 2025 00:00:00 +0000</lastBuildDate><atom:link href="https://devendevour.iankulin.com/tags/ai/index.xml" rel="self" type="application/rss+xml"/><item><title>State of AI tooling (for me)</title><link>https://devendevour.iankulin.com/state-of-ai-tooling-for-me/</link><pubDate>Mon, 07 Jul 2025 00:00:00 +0000</pubDate><guid>https://devendevour.iankulin.com/state-of-ai-tooling-for-me/</guid><description>&lt;p&gt;I&amp;rsquo;ve been meaning to write this for a couple of weeks, so let&amp;rsquo;s get to it - things are moving to fast to reflect too long; which is it&amp;rsquo;s own risk.&lt;/p&gt;
&lt;p&gt;In March, I wrote about &lt;a href="https://devendevour.iankulin.com/where-im-up-to-with-ai-for-coding/"&gt;how I was using AI in coding&lt;/a&gt; , which was Codeium (now Windsurf) in VS Code for completions, and ChatGPT and Claude online for architecture questions and code gen that was more than half a function.&lt;/p&gt;</description></item><item><title>Where I'm up to with AI for coding</title><link>https://devendevour.iankulin.com/where-im-up-to-with-ai-for-coding/</link><pubDate>Mon, 03 Mar 2025 00:00:00 +0000</pubDate><guid>https://devendevour.iankulin.com/where-im-up-to-with-ai-for-coding/</guid><description>&lt;p&gt;There&amp;rsquo;s still plenty of controversy about LLMs for coding, and not without reason. But I thought I&amp;rsquo;d run through what I&amp;rsquo;ve tried, and where I&amp;rsquo;ve landed for using AI. Also what the pitfalls are, where it&amp;rsquo;s useful and how it&amp;rsquo;s changed my practice.&lt;/p&gt;
&lt;h3 id="issues"&gt;Issues&lt;/h3&gt; &lt;h5 id="training-data"&gt;Training data&lt;/h5&gt; &lt;p&gt;The training data for large language models generally is problematic. There&amp;rsquo;s no doubt that they have been trained on copyright material. With code it&amp;rsquo;s slightly less murky since there is a high availability of good quality open source data with attached licenses to train models on. No doubt this include code written by people who don&amp;rsquo;t approve of it being used by AI, but I think the popular reading of most open source licenses is that using it for training is fine.&lt;/p&gt;</description></item><item><title>Clear explanation of how transformer AI works</title><link>https://devendevour.iankulin.com/clear-explanation-of-how-transformer-ai-works/</link><pubDate>Mon, 28 Oct 2024 00:00:00 +0000</pubDate><guid>https://devendevour.iankulin.com/clear-explanation-of-how-transformer-ai-works/</guid><description>&lt;p&gt;If you&amp;rsquo;re interested in how generative AI works, check out &lt;a href="https://ishananand.com/" target="_blank" rel="noopener"&gt;Ishan Anand&lt;/a&gt; &amp;rsquo;s Youtube series &amp;ldquo;&lt;a href="https://www.youtube.com/@Spreadsheetsareallyouneed" target="_blank" rel="noopener"&gt;Spreadsheets are all you need&lt;/a&gt; &amp;rdquo;. He steps through the basics using an Excel spreadsheet that encompasses most of GPT-2. Just doing that is an impressive (and hilarious) feat, but he also has a knack for teaching, so you&amp;rsquo;ll come away with a good understanding of AI and how some of it&amp;rsquo;s limitations manifest.&lt;/p&gt;
&lt;p&gt;Ishan is selling a course, which I guess these are the first three lessons of, and I got a lot out of them. It&amp;rsquo;s also possible to &lt;a href="https://github.com/ianand/spreadsheets-are-all-you-need/releases/tag/v0.6.1" target="_blank" rel="noopener"&gt;download the spreadsheet&lt;/a&gt; he uses in the course to play with.&lt;/p&gt;</description></item><item><title>LLM coding question comparison using Ollama</title><link>https://devendevour.iankulin.com/llm-coding-question-comparison-using-ollama/</link><pubDate>Mon, 29 Jul 2024 00:00:00 +0000</pubDate><guid>https://devendevour.iankulin.com/llm-coding-question-comparison-using-ollama/</guid><description>&lt;p&gt;Now Ollama has made it simple enough for anyone who can use a terminal to run large language models locally, naturally I&amp;rsquo;ve gone overboard downloading too many to play with. I&amp;rsquo;m increasingly feeling they definitely have a place in the devops/coding arsenal of tools, but which model is best?&lt;/p&gt;
&lt;p&gt;If you go on HuggingFace to look at a new model you&amp;rsquo;re interested, they often have great comparisons like this.&lt;/p&gt;
&lt;p&gt;&lt;a href="https://huggingface.co/deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct" target="_blank" rel="noopener"&gt;&lt;img src="https://devendevour.iankulin.com/images/performance.png" alt="" class="img-responsive"&gt; &lt;/a&gt; &lt;/p&gt;</description></item><item><title>Using LLMs for coding</title><link>https://devendevour.iankulin.com/using-llms-for-coding/</link><pubDate>Mon, 01 Jul 2024 00:00:00 +0000</pubDate><guid>https://devendevour.iankulin.com/using-llms-for-coding/</guid><description>&lt;p&gt;&lt;a href="https://madmuseum.org/events/ghost-shell" target="_blank" rel="noopener"&gt;&lt;img src="https://devendevour.iankulin.com/images/ghost-in-the-shell_07.jpg" alt="Ghost in the Shell
© Manga Entertainment 1996
" class="img-responsive"&gt; &lt;/a&gt; &lt;/p&gt;
&lt;p&gt;This post looks at the context for some of my thinking about AI for supporting software development, and where I&amp;rsquo;ve landed on it for the time being.&lt;/p&gt;
&lt;h3 id="the-landscape"&gt;The landscape&lt;/h3&gt; &lt;p&gt;I &lt;a href="https://devendevour.iankulin.com/chatgpts-code-writing/"&gt;briefly wrote about ChatGPT&amp;rsquo;s&lt;/a&gt; coding ability at the end of 2022. The wide availability of this tool marked the beginning of what I think can fairly be described as a revolution. The controversies that have crystalised since have not dampened my amazement of this step forward in what compute can do, especially around natural language processing.&lt;/p&gt;</description></item><item><title>ChatGPT's code writing</title><link>https://devendevour.iankulin.com/chatgpts-code-writing/</link><pubDate>Sun, 11 Dec 2022 00:00:00 +0000</pubDate><guid>https://devendevour.iankulin.com/chatgpts-code-writing/</guid><description>&lt;p&gt;&lt;img src="https://devendevour.iankulin.com/images/22.jpg" alt="room full of monkeys typing at computers - stable diffusion" class="img-responsive"&gt; &lt;/p&gt;
&lt;p&gt;This week, the internet has been all about &lt;a href="https://chat.openai.com/chat" target="_blank" rel="noopener"&gt;ChatGPT&lt;/a&gt; , the rather remarkable natural language AI with a very large model. If you&amp;rsquo;re a twitter user, you were probably amazed, but now eventually tired of seeing examples of it&amp;rsquo;s output. I&amp;rsquo;ll add to that with an example of a SwifUI CoreData based todo app it wrote for me from a single sentence prompt below. Rather than look at other people&amp;rsquo;s examples you should definitely go and play with it yourself - it is very impressive. Along with the image based AI&amp;rsquo;s it&amp;rsquo;s made 2022 into a historical year for AI.&lt;/p&gt;</description></item><item><title>Using AI to Generate Icons</title><link>https://devendevour.iankulin.com/using-ai-to-generate-icons/</link><pubDate>Tue, 11 Oct 2022 00:00:00 +0000</pubDate><guid>https://devendevour.iankulin.com/using-ai-to-generate-icons/</guid><description>&lt;p&gt;&lt;img src="https://devendevour.iankulin.com/images/macos-icon-scissors-floating-over-code2.jpg" alt="Open macbook with design image showing - created by Stable Diffusion" class="img-responsive"&gt; &lt;/p&gt;
&lt;p&gt;Since I have minimal design skills, I went back to &lt;a href="https://www.fiverr.com/" target="_blank" rel="noopener"&gt;Fiverr&lt;/a&gt; (the digital gig economy platform) to get some icons done for CodeTrimmer - explaining that I wanted something like a &amp;ldquo;pair of scissors floating over some computer code&amp;rdquo;. At the same time I&amp;rsquo;ve been playing with &lt;a href="https://github.com/divamgupta/diffusionbee-stable-diffusion-ui" target="_blank" rel="noopener"&gt;DiffusionBee&lt;/a&gt; - a free Apple silicon version of the &lt;a href="https://stability.ai/blog/stable-diffusion-public-release" target="_blank" rel="noopener"&gt;Stable Diffusion&lt;/a&gt; artifical intellligence that generates images from text prompts. The image above was created on an M1 Macbook using DiffusionBee.&lt;/p&gt;</description></item></channel></rss>