Professor at Mines Paris & Curie Institute. Machine learning for life sciences and therapeutic research. Shallow learning + fat data. #BiInSci. Cofounder of Paris WiMLDS. #MECFS patient.
Public Key
npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Profile Code
nprofile1qqsyynzxmg3lmrwu64yxacunh30elx64r440znl0kruj3uzfprr4wugpz4mhxue69uhhyetvv9ujuerfw36x7tnsw43qz9rhwden5te0wfjkccte9ejxzmt4wvhxjmcre7u35
Show more details
Published at
2026-05-21T09:27:06Z Event JSON
{
"id": "397a595b5553827a8ab9d315f7021de2b5429c12909c366cdec7d548c87854d2" ,
"pubkey": "424c46da23fd8ddcd5486ee393bc5f9f9b551d6af14fefb0f928f04908c75771" ,
"created_at": 1779355626 ,
"kind": 0 ,
"tags": [
[
"proxy",
"https://lipn.info/users/cazencott",
"activitypub"
],
[
"client",
"Mostr",
"31990:6be38f8c63df7dbf84db7ec4a6e6fbbd8d19dca3b980efad18585c46f04b26f9:mostr",
"wss://relay.ditto.pub"
]
],
"content": "{\"name\":\"Chloé Azencott\",\"about\":\"Professor at Mines Paris \u0026 Curie Institute. Machine learning for life sciences and therapeutic research. Shallow learning + fat data. #BiInSci. Cofounder of Paris WiMLDS. #MECFS patient.\",\"picture\":\"https://lipn.info/system/accounts/avatars/109/291/639/298/711/356/original/b51677d1d118db96.png\",\"banner\":\"https://lipn.info/system/accounts/headers/109/291/639/298/711/356/original/32026b2ba1a156eb.png\",\"nip05\":\"[email protected] \",\"fields\":[[\"Website\",\"http://cazencott.info\"],[\"Pronouns\",\"she/her\"],[\"Profile picture\",\"A cropped shot of a white woman with short brown hair, a thick red jumper, big glasses and a tired air.\"],[\"Banner\",\"The cover of my book \\\"Introduction au Machine Learning\\\"\"]]}" ,
"sig": "7e4f184974a820de6ed4840754bd48e8e8328b0be4343d1e839f4ae17f64b3d21713425a29b0bca051122e5ab1862957106ec16ffc73adcf085dbf052d403025"
}
Last Notes npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott Now at #jobim2026 Sarah Cohen-Boulakia is giving a keynote on #reproducibility by design in #bioinformatics We're very proud in our community of our sharing culture: thousands of open data sources, 33k tools on bio.tools, thousands of analysis pipelines on git... but still, computational reproducibility is not a given! Which version of what database was used? Which version of what software, with which parameters? npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott @nprofile…tn9h Peut-être pas le plus important par rapport aux informations déformées, mais "(Très) grande silhouette élégante, cheveux fournis tombant sur un regard bleu" je m'attendais pas à ce style non plus... npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott Getting ready for the "AI & Cancer" day of Cancéropole IdF to discuss what machine learning can and cannot do in genomics. My message: there's a lot that predictive models can do (with the usual caveats of proper evaluation), but as for interpretability, which we need for the generation of novel scientific hypotheses, robust tools with statistical guarantees are few and far between. Will share slides later, if anyone is interested. https://www.canceropole-idf.fr/seminaires/sem2026-ia/ #machineLearning #genomics npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott The claim that Android's Mythos is so skilled at finding and exploiting software vulnerabilities that it's too dangerous to release has, unsurprisingly, nothing to do with security and everything to do with marketing. "The flagship demonstration document turns out to be like the ending of the Wizard of Oz, a sorry disappointment about a model weaponizing two bugs that a different model found, in software the vendor had already patched, in a test environment with the browser sandbox and defense-in-depth mitigations stripped out. Anthropic failed, and somehow the story was flipped into a warning about its success." The Boy That Cried Mythos: Verification is Collapsing Trust in Anthropic | flyingpenguin https://www.flyingpenguin.com/the-boy-that-cried-mythos-verification-is-collapsing-trust-in-anthropic/ npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott Reading about #mansplaining, I am reminded of that one time I was introduced to a tech transfer manager as a tenured member of a machine learning team. The guy then proceeded to explain machine learning to me. This was at a lunch, and half the people at the table stopped eating with their mouth half open. And now I remember another one. I was giving an intro to machine learning to an audience of biomedical researchers. One man had the following question: "I am wondering why you're calling it machine *learning*, when it obviously ought to be machine *teaching*. Maybe you've been confused translating from the French 'apprendre', as it can have both meanings? Have you checked the translation?" Yes, it had been said when I was introduced that I did my PhD in the US and that I was a machine learning expert. And then I shared that story on Twitter, and a man who was *not* a fluent French speaker told me that my story was probably wrong, because "apprendre" only means "to learn" and never "to teach". npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott Hey hive mind, do we have any numbers on training data size / model size / training and query energy consumption of the most recent "AI" tools (the ones sold by Anthropic, OpenAI and the like)? https://www.technologyreview.com/2025/05/20/1116327/ai-energy-usage-climate-footprint-big-tech/ is almost a year old now, same for most evaluations on https://huggingface.co/spaces/AIEnergyScore/Leaderboard I've found https://www.simonpcouch.com/blog/2026-01-20-cc-impact/ for Claude Code but that's a lot of guesswork npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott @nprofile…sxvw That is emphatically not the topic of my post. Calling an automated assistant Emmy after Noether is insulting, independently of how ethical, effective and efficient a tool it is. npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott Hiiiiiiiii !! Détails ici : https://cazencott.info/index.php/pages/Introduction-au-Machine-Learning https://lipn.info/system/media_attachments/files/115/055/819/970/220/661/original/0e7e01ac51262c6d.jpeg npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott "So, you're against using genAI agents in education? What's your stance on calculators then? Gotcha here!" My dude (why is it always a dude? No, don't answer me), your argument has more holes than a colander. Calculators work. They do exactly what the people who want you to buy/use them claim they do. Not so with genAI chatbots. Perhaps even more importantly, the existence of calculators does not negate the importance of teaching kids basic arithmetic. That is because what's important is not just the result, but also how you get to this result, and why you want this result in the first place. So even if genAI chatbots were failproof, I would still want students to be able to do the work themselves, because what matters is not the much the outcome as the process. And as using a calculator properly requires the ability to have some notion of what the result should look like, so as to reduce the very real possibility of a user error, using genAI assistants require the ability to look critically at their output in order to fix all the errors (that, again, aren't due to user input so much as to their being error-prone by nature). npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott To close the evening, digital marketer and podcaster Aurélie Giard-Jacquet is telling us about the superpowers of women in ML and DS. Aurélie is the author of techlipstick, the number 1 French podcast on women in tech. https://lipn.info/system/media_attachments/files/114/671/738/566/828/516/original/a96a639986a3a3b9.jpg npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott @nprofile…gfl2 Tu sais que ça m'a déjà été suggéré sans ironie ? npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott PS : Je n'ai pas fini de mettre à jour ledit chapitre 6, mais je vous le partagerai quand il sera prêt. PPS : Bien sûr nous ne serons en mesure de ne sanctionner que des usages détectés, pas la peine de me le faire remarquer. npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott Si ça peut servir, la petite note sur l'utilisation d'outil d'IA générative que j'ai rédigée à propos du projet numérique de mon cours de science des données : La tentation de s'aider d'agents conversationnels ou d'assistants de code qui s'appuient sur l'IA générative (tels que ChatGPT, Claude, LeChat, ou Copilot) pour réaliser un projet numérique peut être grande. Néanmoins, cette utilisation entrave l'apprentissage, notamment car en fournissant directement une solution (possiblement erronée par ailleurs), elle ne permet pas de mener le cheminement mental nécessaire à l'acquisition de nouvelles compétences. Plusieurs études montrent l'effet délétère de l'utilisation de ces béquilles sur l'apprentissage (Bastani et al., 2024), sur les capacités cognitives des étudiantes et étudiants qui les utilisent de manière suffisamment peu critique (Zhai et al., 2024), ou encore sur la qualité du code écrit par des développeuses et développeurs (Perry et al., 2023). L'utilisation d'agents conversationnels ou d'assistants de code est en conséquence prohibée pour la réalisation du projet et sera sanctionnée. Notons aussi que l'utilisation de ces outils soulève de nombreux problèmes éthiques, écologiques et géopolitiques qui seront abordés dans le Chapitre 6 de ce cours. Références H. Bastani et al. (2024). Generative AI Can Harm Learning. http://dx.doi.org/10.2139/ssrn.4895486 N. Perry et al. (2023). Do Users Write More Insecure Code with AI Assistants? https://doi.org/10.1145/3576915.3623157 C. Zhai, S. Wibowo and L.D. Li (2024). The effects of over-reliance on AI dialogue systems on students' cognitive abilities: a systematic review. https://doi.org/10.1186/s40561-024-00316-7 npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott The initiative calling for the ban of "conversion therapy" in the European Union has collected enough signatures (around 1 180 000 as I type) 🎉 But you still have till the end of today, May 17, to add your signature and support putting an end to hateful practices that have been described as torture by the United Nations. Wouldn't that be a suitable thing to do on the International Day Against Homophobia, Transphobia and Biphobia? https://eci.ec.europa.eu/043/public/ #IDAHOT2025 🌈 npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott EU citizens! The initiative calling for the ban of conversion therapy across the whole Union is missing more than 400 000* signatures and the deadline is May 17. Conversion therapy, which aims at repressing the sexual orientation or gender identity of LGBTQ+ persons, is qualified as torture by the United Nations If enough signatures are collected, the Commission will have to consider the proposal. https://eci.ec.europa.eu/043/public/ Edit: less than 300 000 now (May 15, ~8pm)! Keep sharing and signing! Edit (May 16, ~9am): only about 180 000 to go! This can be done! npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott Choose France for Science. But not social science. https://france2030.agencerecherche.fr/ChooseFranceForScience-2025/accueil.php #ChooseFranceForScience https://lipn.info/system/media_attachments/files/114/386/134/459/617/922/original/8ef8075bb7b2c028.png npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott @nprofile…5fxq un super concept, non ? npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott Le questionnaire (mal ficelé et bien orienté) «IA et enseignement» de la mission ministérielle du même nom est encore ouvert jusqu'à demain. Quelques éléments de mes réponses https://cazencott.info/index.php/post/2025/03/23/Questionnaire-Intelligence-Artificielle-et-Enseignement ou sur @nprofile…ledw : https://academia.hypotheses.org/61444 npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott @nprofile…0pej de rien, c'est quand j'ai vu que ça commençait à être du boulot que je me suis dit qu'il fallait partager :-) npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott J'ai passé beaucoup trop de temps et d'énergie à rédiger et étayer mes réponses au questionnaire à destination de « l’ensemble des usagers et personnels de l’enseignement supérieur » dans le cadre de la mission Taddei-Pascal sur « l’intelligence artificielle et l'enseignement », alors je vous les partage : https://cazencott.info/index.php/post/2025/03/23/Questionnaire-Intelligence-Artificielle-et-Enseignement npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott To keep their federal funding, Columbia University agreed to ban masks, empower 36 campus police officers with new powers to arrest students, and appoint a senior vice provost with broad authority to oversee the department of Middle East, South Asian and African Studies as well as the Center for Palestine Studies. https://www.wsj.com/us-news/education/columbia-university-trump-demands-federal-funding-e94d41ca I don't have the words. npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott I last traveled to the US in 2019, for a mix of reasons including not wanting to fly, the covid pandemic, my health. Then worries about anti-abortion policies. . And now a French researcher was denied entry because he criticized the Trump administration. It's not like I was planning to fly to the US any time soon. And I can't say I'm surprised, given how everything else is going. And it's such a small thing, my own little life story, compared to the big picture. But, damn, I used to live in this country. For 5 whole years. That it's not possible to go back—it hurts. https://www.liberation.fr/international/amerique/un-scientifique-francais-refoule-a-son-entree-aux-etats-unis-pour-des-messages-anti-trump-paris-grince-des-dents-20250319_JD7JAIVY4ZHE5CI33F7TBNR4CI/ npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott I have a lecture during the Parisian #standUpForScience rally. So I made a slide. https://lipn.info/system/media_attachments/files/114/120/470/399/438/055/original/beb5da27124a4926.png npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott #standUpForScience est l'occasion de rappeler qu'en France on a 60 universités publiques sur 75 en déficit en 2024, un milliard d'euros en moins dans le budget 2025 pour l'enseignement supérieur et la recherche, et «l'islamo-gauchisme qui gangrène l'université». npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott The last #ParisWiMLDS Meetup of 2024 is announced! Tuesday, December 10 at 6:45pm at Mines Paris PSL (right next to the Luxembourg garden). Featuring Meredith Whittaker @nprofile…ka48 on AI, privacy, and the surveillance business model and Raphaëlle Rémy-Leleu @nprofile…668z on ecofeminism vs techno police. Followed by a cocktail sponsored by https://itn.dev/ and https://theallyance.one/ Mark your calendar, free your evening, and register for free at https://www.meetup.com/paris-women-in-machine-learning-data-science/events/304558096/?slug=paris-women-in-machine-learning-data-science&eventId=304558096 npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott Too bad there was no woman in the vicinity of this Nobel-worthy paper to share the award, really, one tries one's best, but it's so hard to find women who do science worthy of a #NobelPrize. h/t @npub1kxm…x64z https://lipn.info/system/media_attachments/files/113/267/317/823/403/215/original/194d2aeb43152eaa.jpg npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott Very much looking forward to reading this preprint by @npub14m5…fhnd, @npub1ynl…873s and Sacha Luccioni! [2409.14160] Hype, Sustainability, and the Price of the Bigger-is-Better Paradigm in AI https://arxiv.org/abs/2409.14160 #machineLearning npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott So I just gave this keynote on deep learning for clinical applications, right. Then some dude comes up to me, thanks me for my talk (always appreciated), and then tells me how nice he thinks it is that I'm handling the modeling myself. I'm a bit puzzled, I guess my PhD students have been handling it at least as much, but I just say thanks because socialization y'all, and then it hits me... does this guy think it's cute that I'm doing math? Is "professor of machine learning" misleading somehow? npub1gfxydk3rlkxae42gdm3e80zln7d428t27987lv8e9rcyjzx82acseqv9un Chloé Azencott That's still way too many fascists, but THEY DID NOT WIN! There's a lot of work to do to send them back to no seats at all, but tonight, we can take a deep breath and rest. Be careful if you're planning to celebrate outside, lots of neo-nazi groups are planning to be violent tonight. https://lipn.info/system/media_attachments/files/112/746/622/399/707/555/original/2cf68580c1c56524.jpg