<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>Home on Giovanni Pinna</title>
    <link>https://giovannipinna.net/</link>
    <description>Recent content in Home on Giovanni Pinna</description>
    <generator>Hugo</generator>
    <language>en</language>
    <lastBuildDate>Tue, 14 Apr 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://giovannipinna.net/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>There Is No &#34;Best&#34; AI Coding Agent — And That&#39;s the Whole Point</title>
      <link>https://giovannipinna.net/posts/msr2026-comparing-ai-agents/</link>
      <pubDate>Tue, 14 Apr 2026 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/msr2026-comparing-ai-agents/</guid>
      <description>We looked at 7,156 pull requests from five AI coding agents on real open-source projects. The agent matters less than you&amp;#39;d think. The kind of work matters far more.</description>
    </item>
    <item>
      <title>When AI Agents Lie About Their Own Code (Without Meaning To)</title>
      <link>https://giovannipinna.net/posts/msr2026-message-code-inconsistency/</link>
      <pubDate>Tue, 14 Apr 2026 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/msr2026-message-code-inconsistency/</guid>
      <description>Only 1.7% of AI-authored pull requests have descriptions that don&amp;#39;t match their code. Those PRs get accepted 51.7% less often and take 3.5× longer to merge. Trust is the bottleneck nobody is measuring.</description>
    </item>
    <item>
      <title>Sometimes the Best Feature Engineering Is Throwing Features Away</title>
      <link>https://giovannipinna.net/posts/ssbse2025-hotcat/</link>
      <pubDate>Mon, 13 Oct 2025 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/ssbse2025-hotcat/</guid>
      <description>Classifying urgent software hotfixes is hard: tiny dataset, brutal class imbalance, expensive LLM features. We let evolution pick which features to keep — and discovered some were actively making things worse.</description>
    </item>
    <item>
      <title>Sometimes Your AI Agent Burns More Energy Optimizing Code Than the Code Will Ever Save</title>
      <link>https://giovannipinna.net/posts/ssbse2025-ga4gc/</link>
      <pubDate>Mon, 13 Oct 2025 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/ssbse2025-ga4gc/</guid>
      <description>AI coding agents that &amp;#39;optimize&amp;#39; your code can cost more energy than they save — for hundreds of thousands of runs. We tuned the agent itself, and got 37.7% faster runs and better code at the same time.</description>
    </item>
    <item>
      <title>The Text-to-SQL Field Has a Measurement Problem</title>
      <link>https://giovannipinna.net/posts/scireports2025-text-to-sql-metrics/</link>
      <pubDate>Wed, 02 Jul 2025 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/scireports2025-text-to-sql-metrics/</guid>
      <description>Every text-to-SQL benchmark today scores queries as either perfect or wrong. That&amp;#39;s a coin flip dressed up as a metric. We built one that actually tells you how close you got.</description>
    </item>
    <item>
      <title>Making the LLM-Plus-Evolution Pipeline Actually Smart</title>
      <link>https://giovannipinna.net/posts/sncs2025-exploring-gi-effect/</link>
      <pubDate>Tue, 01 Jul 2025 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/sncs2025-exploring-gi-effect/</guid>
      <description>Last year we showed evolution can fix LLM code. This year we made the evolution itself smarter — better selection, partial credit, fewer cycles — and got improvements in 11 of 12 cases.</description>
    </item>
    <item>
      <title>Improving LLM-Generated Code via Genetic Improvement: A Summary of Recent Advances</title>
      <link>https://giovannipinna.net/posts/italia2025-gi-summary/</link>
      <pubDate>Mon, 23 Jun 2025 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/italia2025-gi-summary/</guid>
      <description>A comprehensive summary of our research program on applying Genetic Improvement to LLM-generated code, presented at the Italian national AI conference Ital-IA 2025.</description>
    </item>
    <item>
      <title>GPT-4 Can Make Court Rulings Easier to Read. It Can Also Lie to You About Them, Confidently.</title>
      <link>https://giovannipinna.net/posts/wiat2024-courts-to-comprehension/</link>
      <pubDate>Tue, 10 Dec 2024 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/wiat2024-courts-to-comprehension/</guid>
      <description>We asked 75 people to read summaries of Italian Constitutional Court rulings — written by experts, by GPT-4o, by a fine-tuned LLaMA, and the raw judgments themselves. The results say more about LLMs than about courts.</description>
    </item>
    <item>
      <title>What If We Stopped Asking ChatGPT to Fix Its Own Code?</title>
      <link>https://giovannipinna.net/posts/eurogp2024-gi-for-llm-code/</link>
      <pubDate>Wed, 03 Apr 2024 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/eurogp2024-gi-for-llm-code/</guid>
      <description>Self-correction is the default fix for buggy LLM code, but it has a ceiling. We tried something stranger — evolving the code instead — and it worked across every model we tested.</description>
    </item>
    <item>
      <title>Influence</title>
      <link>https://giovannipinna.net/projects/influence/</link>
      <pubDate>Wed, 13 Jan 2021 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/projects/influence/</guid>
      <description>A platform combining marketing and AI to analyze audiences and generate targeted social media content.</description>
    </item>
    <item>
      <title>Influence: Where Marketing Meets Artificial Intelligence</title>
      <link>https://giovannipinna.net/posts/influence-project/</link>
      <pubDate>Wed, 13 Jan 2021 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/posts/influence-project/</guid>
      <description>A project exploring the intersection of marketing and artificial intelligence, using data analysis to generate targeted social media content.</description>
    </item>
    <item>
      <title>Book Review: Thinking, Fast and Slow by Daniel Kahneman</title>
      <link>https://giovannipinna.net/posts/pensieri-lenti-e-veloci/</link>
      <pubDate>Sun, 10 Jan 2021 10:00:00 +0100</pubDate>
      <guid>https://giovannipinna.net/posts/pensieri-lenti-e-veloci/</guid>
      <description>A review of Daniel Kahneman&amp;#39;s Nobel Prize-winning work on how we make decisions and the cognitive biases that influence our thinking.</description>
    </item>
    <item>
      <title>Book Review: Don&#39;t Make Me Think by Steve Krug</title>
      <link>https://giovannipinna.net/posts/dont-make-me-think/</link>
      <pubDate>Sun, 10 Jan 2021 09:00:00 +0100</pubDate>
      <guid>https://giovannipinna.net/posts/dont-make-me-think/</guid>
      <description>A review of Steve Krug&amp;#39;s classic guide to web usability and human-computer interaction.</description>
    </item>
    <item>
      <title>About Me</title>
      <link>https://giovannipinna.net/about/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/about/</guid>
      <description>&lt;div class=&#34;about-hero&#34;&gt;
  &lt;div class=&#34;about-hero-img&#34;&gt;
    &lt;img src=&#34;https://giovannipinna.net/images/profile.png&#34; alt=&#34;Giovanni Pinna&#34;&gt;
  &lt;/div&gt;
  &lt;div class=&#34;about-hero-text&#34;&gt;
    &lt;h1&gt;Hi, I&#39;m Giovanni &lt;span class=&#34;wave&#34;&gt;👋&lt;/span&gt;&lt;/h1&gt;
    &lt;p class=&#34;about-tagline&#34;&gt;AI Researcher &amp;middot; ML Engineer &amp;middot; Ph.D.&lt;/p&gt;
    &lt;p&gt;I&#39;m an AI Researcher and Engineer based in &lt;strong&gt;Trieste, Italy&lt;/strong&gt;, with a Ph.D. in Applied Data Science &amp; Artificial Intelligence. My work sits at the intersection of &lt;strong&gt;NLP&lt;/strong&gt;, &lt;strong&gt;Large Language Models&lt;/strong&gt;, and &lt;strong&gt;AI Agents&lt;/strong&gt;, and &lt;strong&gt;Evolutionary Computation&lt;/strong&gt; &amp;mdash; building systems that improve and evaluate AI-generated code.&lt;/p&gt;
    &lt;div class=&#34;about-cta&#34;&gt;
      &lt;a href=&#34;https://giovannipinna.net/contact/&#34; class=&#34;btn-primary&#34;&gt;Get in Touch&lt;/a&gt;
      &lt;a href=&#34;https://giovannipinna.net/cv/&#34; class=&#34;btn-secondary&#34;&gt;View My CV&lt;/a&gt;
    &lt;/div&gt;
  &lt;/div&gt;
&lt;/div&gt;
&lt;div class=&#34;about-section&#34;&gt;
  &lt;h2&gt;🎯 My Story&lt;/h2&gt;
  &lt;p&gt;My journey in computer science started in high school, where I fell in love with programming and electronics. That curiosity led me to a &lt;strong&gt;B.Sc. in Electronic and Computer Engineering&lt;/strong&gt; (2015&amp;ndash;2019) at the University of Trieste, where my first AI courses sparked a passion that shaped everything that followed &amp;mdash; continuing with a &lt;strong&gt;M.Sc. in Computer Engineering&lt;/strong&gt; (2019&amp;ndash;2022) focused on Data Science and AI. A semester at &lt;strong&gt;Montanuniversit&amp;auml;t Leoben&lt;/strong&gt; through Erasmus+ pushed me out of my comfort zone: applying AI to energy and logistics taught me how transferable these methods really are.&lt;/p&gt;</description>
    </item>
    <item>
      <title>Activities</title>
      <link>https://giovannipinna.net/activities/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/activities/</guid>
      <description>Academic and professional activities</description>
    </item>
    <item>
      <title>Contact</title>
      <link>https://giovannipinna.net/contact/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/contact/</guid>
      <description>Get in touch with Giovanni Pinna</description>
    </item>
    <item>
      <title>Curriculum Vitae</title>
      <link>https://giovannipinna.net/cv/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/cv/</guid>
      <description>Giovanni Pinna</description>
    </item>
    <item>
      <title>Publications</title>
      <link>https://giovannipinna.net/publications/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      <guid>https://giovannipinna.net/publications/</guid>
      <description>Full list of publications</description>
    </item>
  </channel>
</rss>
