<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel><title>Mert Demirer · AI Fault Line</title><link>https://aifaultline.com/authors/mert-demirer/</link><description>New papers by Mert Demirer in the AI Fault Line guide, and news naming them.</description>
<item><title>News: Is AI Replacing Programmers or Boosting Them?</title><link>https://empiricrafting.substack.com/p/is-ai-replacing-programmers-or-boosting</link><guid>https://empiricrafting.substack.com/p/is-ai-replacing-programmers-or-boosting</guid><pubDate>Mon, 27 Jul 2026 00:00:00 +0000</pubDate><description>The Justified Posteriors podcast (Andrey Fradkin) discusses a new paper by Mert Demirer, Leon Musolff and Liyuan Yang, &#x27;Writing Code vs. Shipping Code&#x27;, which estimates how successive generations of AI coding tools affect software production productivity.</description></item>
<item><title>Paper: Chaining Tasks, Redefining Work: A Theory of AI Automation</title><link>https://aifaultline.com/research/p/title-daa71d8fc2d188be/</link><guid>https://aifaultline.com/research/p/title-daa71d8fc2d188be/</guid><pubDate>Sun, 14 Jun 2026 00:00:00 +0000</pubDate></item>
<item><title>Paper: The Effects of Generative AI on High Skilled Work: Evidence from Three Field Experiments with Software Developers</title><link>https://aifaultline.com/research/p/doi-10-2139-ssrn-4945566/</link><guid>https://aifaultline.com/research/p/doi-10-2139-ssrn-4945566/</guid><pubDate>Mon, 01 Jan 2024 00:00:00 +0000</pubDate><description>Reports on three separate field experiments (at Microsoft, Accenture, and a Fortune 100 firm) measuring how GitHub Copilot access affects software developers&#x27; output, finding sizable increases in completed tasks per week that are largest for less-experienced programmers. As one of the largest employer-scale randomized trials of a generative AI coding tool to date, it adds enterprise-level evidence to the debate over whether AI narrows or widens skill gaps among workers.</description></item>
<item><title>Paper: The Impact of AI on Developer Productivity: Evidence from GitHub Copilot</title><link>https://aifaultline.com/research/p/doi-10-48550-arxiv-2302-06590/</link><guid>https://aifaultline.com/research/p/doi-10-48550-arxiv-2302-06590/</guid><pubDate>Mon, 13 Feb 2023 00:00:00 +0000</pubDate><description>A controlled experiment with recruited developers building a JavaScript HTTP server found that those with GitHub Copilot access finished the task 55.8% faster than a control group without it, with especially strong effects for less experienced programmers. It remains one of the most-cited direct measurements of an AI coding assistant&#x27;s productivity effect and is frequently used as a benchmark figure in broader estimates of AI-driven productivity growth.</description></item>
</channel></rss>
