<?xml version="1.0" encoding="utf-8"?><feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en"><generator uri="https://jekyllrb.com/" version="4.3.3">Jekyll</generator><link href="https://valentinguigon.github.io/feed.xml" rel="self" type="application/atom+xml"/><link href="https://valentinguigon.github.io/" rel="alternate" type="text/html" hreflang="en"/><updated>2026-06-19T23:57:59+00:00</updated><id>https://valentinguigon.github.io/feed.xml</id><title type="html">Valentin Guigon</title><subtitle>description: &gt; Personal website of Valentin Guigon, postdoctoral researcher in computational neuroscience, computational psychiatry, and NeuroAI, studying social learning, trust, belief updating, and decision-making under uncertainty. </subtitle><entry><title type="html">Supervising LLMs in multi-step projects is a governance problem</title><link href="https://valentinguigon.github.io/posts/2026-05-13-2026-05-13-supervising-llms-in-multi-step-projects-is-a-governance-problem/" rel="alternate" type="text/html" title="Supervising LLMs in multi-step projects is a governance problem"/><published>2026-05-13T14:45:27+00:00</published><updated>2026-05-13T14:45:27+00:00</updated><id>https://valentinguigon.github.io/posts/2026-05-13-supervising-llms-in-multi-step-projects-is-a-governance-problem</id><content type="html" xml:base="https://valentinguigon.github.io/posts/2026-05-13-2026-05-13-supervising-llms-in-multi-step-projects-is-a-governance-problem/"><![CDATA[]]></content><author><name></name></author><category term="articles"/><summary type="html"><![CDATA[Building a governance workflow for complex AI-assisted work in the repo.]]></summary></entry><entry><title type="html">The problem of experimental control in evaluating genAI models</title><link href="https://valentinguigon.github.io/posts/2026-04-12-2026-04-12-the-problem-of-experimental-control-in-evaluating-genai-models/" rel="alternate" type="text/html" title="The problem of experimental control in evaluating genAI models"/><published>2026-04-12T13:30:59+00:00</published><updated>2026-04-12T13:30:59+00:00</updated><id>https://valentinguigon.github.io/posts/2026-04-12-the-problem-of-experimental-control-in-evaluating-genai-models</id><content type="html" xml:base="https://valentinguigon.github.io/posts/2026-04-12-2026-04-12-the-problem-of-experimental-control-in-evaluating-genai-models/"><![CDATA[]]></content><author><name></name></author><category term="articles"/><summary type="html"><![CDATA[An experimentalist view on Claude Mythos Preview awareness of being tested.]]></summary></entry><entry><title type="html">3.5 days to build a March Madness forecasting system</title><link href="https://valentinguigon.github.io/posts/2026-03-20-2026-03-20-35-days-to-build-a-march-madness-forecasting-system/" rel="alternate" type="text/html" title="3.5 days to build a March Madness forecasting system"/><published>2026-03-20T16:49:13+00:00</published><updated>2026-03-20T16:49:13+00:00</updated><id>https://valentinguigon.github.io/posts/2026-03-20-35-days-to-build-a-march-madness-forecasting-system</id><content type="html" xml:base="https://valentinguigon.github.io/posts/2026-03-20-2026-03-20-35-days-to-build-a-march-madness-forecasting-system/"><![CDATA[]]></content><author><name></name></author><category term="articles"/><summary type="html"><![CDATA[Building a tightly supervised AI workflow for public reporting under severe time pressure]]></summary></entry><entry><title type="html">Neuroscience 2035: The future of neuroscience is integrative</title><link href="https://valentinguigon.github.io/posts/2026-03-08-2026-03-08-neuroscience-2035-the-future-of-neuroscience-is-integrative/" rel="alternate" type="text/html" title="Neuroscience 2035: The future of neuroscience is integrative"/><published>2026-03-08T00:26:40+00:00</published><updated>2026-03-08T00:26:40+00:00</updated><id>https://valentinguigon.github.io/posts/2026-03-08-neuroscience-2035-the-future-of-neuroscience-is-integrative</id><content type="html" xml:base="https://valentinguigon.github.io/posts/2026-03-08-2026-03-08-neuroscience-2035-the-future-of-neuroscience-is-integrative/"><![CDATA[]]></content><author><name></name></author><category term="articles"/><summary type="html"><![CDATA[The previous decade (2015&#8211;2025) was busy building tools, methods, and diagnoses. This new decade will be about integrating across levels of description, species, and data modalities.]]></summary></entry><entry><title type="html">Better Judgments: How AI forecasting abilities will reshape human metacognition</title><link href="https://valentinguigon.github.io/posts/2026-02-23-2026-02-23-better-judgments-how-ai-forecasting-abilities-will-reshape-human-metacognition/" rel="alternate" type="text/html" title="Better Judgments: How AI forecasting abilities will reshape human metacognition"/><published>2026-02-23T22:05:15+00:00</published><updated>2026-02-23T22:05:15+00:00</updated><id>https://valentinguigon.github.io/posts/2026-02-23-better-judgments-how-ai-forecasting-abilities-will-reshape-human-metacognition</id><content type="html" xml:base="https://valentinguigon.github.io/posts/2026-02-23-2026-02-23-better-judgments-how-ai-forecasting-abilities-will-reshape-human-metacognition/"><![CDATA[]]></content><author><name></name></author><category term="articles"/><summary type="html"><![CDATA[In 2027, is it still worth bothering with human judgment?]]></summary></entry><entry><title type="html">Better Judgments: What good judgments are missing</title><link href="https://valentinguigon.github.io/posts/2026-02-21-2026-02-21-better-judgments-what-good-judgments-are-missing/" rel="alternate" type="text/html" title="Better Judgments: What good judgments are missing"/><published>2026-02-21T22:08:41+00:00</published><updated>2026-02-21T22:08:41+00:00</updated><id>https://valentinguigon.github.io/posts/2026-02-21-better-judgments-what-good-judgments-are-missing</id><content type="html" xml:base="https://valentinguigon.github.io/posts/2026-02-21-2026-02-21-better-judgments-what-good-judgments-are-missing/"><![CDATA[]]></content><author><name></name></author><category term="articles"/><summary type="html"><![CDATA[In 2026, are we still evaluating the world the right way?]]></summary></entry><entry><title type="html">Intelligence is easy; cognition is hard</title><link href="https://valentinguigon.github.io/posts/2026-01-24-2026-01-24-intelligence-is-easy-cognition-is-hard/" rel="alternate" type="text/html" title="Intelligence is easy; cognition is hard"/><published>2026-01-24T18:10:41+00:00</published><updated>2026-01-24T18:10:41+00:00</updated><id>https://valentinguigon.github.io/posts/2026-01-24-intelligence-is-easy-cognition-is-hard</id><content type="html" xml:base="https://valentinguigon.github.io/posts/2026-01-24-2026-01-24-intelligence-is-easy-cognition-is-hard/"><![CDATA[]]></content><author><name></name></author><category term="articles"/><summary type="html"><![CDATA[Distinguishing intelligent systems from cognitive systems]]></summary></entry><entry><title type="html">The mind, the brain and the network</title><link href="https://valentinguigon.github.io/posts/2025-10-14-2025-10-14-the-mind-the-brain-and-the-network/" rel="alternate" type="text/html" title="The mind, the brain and the network"/><published>2025-10-14T03:01:14+00:00</published><updated>2025-10-14T03:01:14+00:00</updated><id>https://valentinguigon.github.io/posts/2025-10-14-the-mind-the-brain-and-the-network</id><content type="html" xml:base="https://valentinguigon.github.io/posts/2025-10-14-2025-10-14-the-mind-the-brain-and-the-network/"><![CDATA[]]></content><author><name></name></author><category term="articles"/><summary type="html"><![CDATA[What's modular? Learning under topological constraints]]></summary></entry><entry><title type="html">Can machines think? 2.0</title><link href="https://valentinguigon.github.io/posts/2025-09-06-2025-09-06-can-machines-think-20/" rel="alternate" type="text/html" title="Can machines think? 2.0"/><published>2025-09-06T20:36:59+00:00</published><updated>2025-09-06T20:36:59+00:00</updated><id>https://valentinguigon.github.io/posts/2025-09-06-can-machines-think-20</id><content type="html" xml:base="https://valentinguigon.github.io/posts/2025-09-06-2025-09-06-can-machines-think-20/"><![CDATA[]]></content><author><name></name></author><category term="articles"/><summary type="html"><![CDATA[Assessing machine vs human cognition in a game of Baseball-Poker]]></summary></entry><entry><title type="html">Neuroscience 2035: What the next decade looks like</title><link href="https://valentinguigon.github.io/posts/2025-08-27-2025-08-27-neuroscience-2035-what-the-next-decade-looks-like/" rel="alternate" type="text/html" title="Neuroscience 2035: What the next decade looks like"/><published>2025-08-27T23:10:23+00:00</published><updated>2025-08-27T23:10:23+00:00</updated><id>https://valentinguigon.github.io/posts/2025-08-27-neuroscience-2035-what-the-next-decade-looks-like</id><content type="html" xml:base="https://valentinguigon.github.io/posts/2025-08-27-2025-08-27-neuroscience-2035-what-the-next-decade-looks-like/"><![CDATA[]]></content><author><name></name></author><category term="articles"/><summary type="html"><![CDATA[The previous decade (2015-2025) has been busy building tools, methods, and diagnoses. This new decade will be integrative.]]></summary></entry></feed>