Pascal Lapointe
@paslap.bsky.social
Journaliste scientifique, rédacteur en chef de Science-Presse : @sciencepresse.bsky.social La rubrique de fact-checking Le #DétecteurDeRumeurs. Le cours en ligne Décoder l’info scientifique. www.sciencepresse.qc.ca
Grokipedia appears not to have updated any articles since April 24. @noupside.bsky.social and @rer.bsky.social explore why it matters that the encyclopedia hasn’t edited itself or processed edits from humans in more than 3 months. www.lawfaremedia.org/article/grok...
Grokipedia Stopped Reviewing Edits in April. It Didn’t Tell Anyone.
A halt in Grokipedia’s edit pipeline offers a case study in how an AI-generated reference system governs its own content.
lawfaremedia.org
Les antivaccins sont nuls avec les chiffres? Quelle surprise. :-)
Timothy Caulfield@caulfieldtim.bsky.social · 2d ago🚨 Antivaxxer & stats: WRONG. "Anti-vaccine discourse exhibits systematic patterns of statistical misreasoning": - correlation–causation fallacy - base rate neglect - Denominator neglect - cherry picking Study: Antivaxx statistical misreasoning & disinformation journals.plos.org/plosone/arti...
1/2 « L’enjeu est fondamental : on ne peut aujourd’hui concevoir la souveraineté culturelle, politique et économique sans y intégrer la dimension numérique. Ce qui est en cause est la capacité et surtout la volonté du Canada d’exercer sa souveraineté numérique » www.ledevoir.com/opinion/idee...?
Du «elbows up» au «elbows down»
Saboter un élément central de la politique culturelle canadienne revient à hypothéquer la souveraineté canadienne.
ledevoir.com
Enfilade. « Project Panama was meant to be kept secret. As an internal memo from April 2024 shows, its goal was to scan and destroy "all the books in the world"«
Snopes@snopes.com · 2d ago🧵 Snopes pored over hundreds of pages of documents to confirm that Anthropic bought millions of books in bulk and destructively scanned them to train its AI model.
1/2 «Un robot conversationnel va adapter son discours aux croyances de celui qui le questionne, dans le but de lui plaire. L’IA va agir comme un miroir extrêmement sophistiqué. Chez certaines personnes vulnérables, ce miroir peut amplifier les pensées présentes» plus.lapresse.ca/screens/7dae...
Quand l’IA mène à la psychose
Avez-vous vu le film où Tom Hanks se retrouve isolé sur une île déserte avec pour seul « ami » un ballon de volley-ball ?
plus.lapresse.ca
« Instead, far-right and MAGA politicians, influencers, will immediately use the most divisive language in order to make voters as angry as possible, as fast as possible, regardless of what actually happened. It’s harder to stop the wave of anger than it is to stop the migrants themselves »
Anne Applebaum@anneapplebaum.wsocial.eu · 3d agoVance, Farage and Weidel posted fear-mongering messages about a Spanish migrant crisis that weren't true. They haven't retracted them. www.theatlantic.com/ideas/2026/0...
Trump aime déposer des poursuites bidon juste pour obliger l’autre à dépenser des sous. Dans le cas de la BBC, celle-ci a pris l’avocat de Trump à son propre piège 1/2 «They drop subpoenas on everyone inside and outside the government who was anywhere near Trump on January 6th»
Trump v. BBC | On the Media | WNYC Studios
What happens when a media company fights back
wnycstudios.org
Un rappel utile au moment où on préfère attaquer Fauci : chez le 20% de la population US vivant dans les comtés les plus pro-Trump, il y a eu 70 000 décès de la COVID de plus que chez le 20% vivant dans les comtés les plus pro-Biden.
The Atlantic@theatlantic.com · 5d agoBy targeting Fauci, Senate Republicans are trying to transfer guilt from those who downplayed or misrepresented the pandemic, @davidfrum.bsky.social argues.
L’IA et l’inévitable réglementation. 1/4 «For years, a debate has been simmering in Silicon Valley over the right way to build artificial intelligence. Should it be open and free for anyone to use, or should we tightly control how it’s made and who can access it» www.nytimes.com/2026/07/31/p...
The Fight Tearing Apart Silicon Valley
Anthropic and OpenAI are clashing with the rest of the tech industry over whether “open-source” models from China should be freely available or restricted.
nytimes.com
« The hallmark of the best-known snake oils is their ability to capitalize on contemporaneous social and political tensions despite weak clinical evidence. Once entrenched in the wellness industry, many therapies outlive their original cultural significance but continue to generate profits »
Timothy Caulfield@caulfieldtim.bsky.social · 6d agoCancer pseudoscience and the cultural moment healthydebate.ca/2026/07/topi... by @docmcohen.medsky.social 🙏 In 2022 Makis helped push "a new quack cancer therapy [#ivermectin] into mainstream discourse." "...drive of snake oil sellers to wring profit from cancer’s misery shows no signs of fading."
« peptides, the cure du jour of the “biohacking” and wellness worlds, capable of fixing everything from poor sleep and bad skin to an inability to get the perfect tan, should you believe the podcasters and influencerss » www.theguardian.com/us-news/ng-i...
‘It’s a peptide wild west’: how RFK Jr is clearing way for a market of untested drugs
A century ago, compounding pharmacies helped people get bespoke drugs. Now these less-regulated businesses are fueling the peptide boom
theguardian.com
1/2 «Ce relativisme ayant contaminé l’esprit d’une bonne partie des citoyens, c’est tout naturellement sur leur ressenti que ceux-ci s’appuient lorsque vient le temps de donner une opinion, de porter un jugement ou de décréter ce qui est vrai, bien et juste» www.ledevoir.com/opinion/idee...
Les enfants de Protagoras et la suspicion envers les experts
L’art de donner un vernis de crédibilité à des options ou à des lubies politiques en flattant le bon peuple.
ledevoir.com
Ça va être important de le rappeler : « A user’s perception of a media brand cited within an AI response determines whether they trust the answer overall. In a nationally representative UK study, we found a direct correlation between trust in the source and the trust in an AI answer »
Press Gazette@pressgazette.co.uk · last wk.New survey finds direct correlation between trust in the source and the trust in an AI answer A user’s perception of a media brand cited within an AI response determines whether they trust the answer overall. pressgazette.co.uk/comment-anal...
Jolie justification de l’importance de s’informer, par Louis Cornellier. 1/2«S’informer, ce n’est pas comme écouter le hockey, jouer à des jeux vidéo ou faire un voyage dans le Sud. Ça ne devrait pas être une affaire de choix individuel ou de goût; c’est une responsabilité, un devoir»
Le monde en été
Dominique Moïsi explique le désordre du monde par l’entrechoquement d’émotions exacerbées.
ledevoir.com
Deux épisodes de la guerre aux faits le même jour. 1/2 «President Trump warning signs and exhibits installed outside the Smithsonian’s National Museum of American History to alert the public that the museum does not, in his view, accurately reflect American history» www.nytimes.com/2026/07/24/a...
Trump Orders Warnings Placed at ‘Inaccurate’ Smithsonian History Museum (Gift Article)
President Trump said the signage at the National Museum of American History would correct what a disputed White House report described as “extreme political activism.”
nytimes.com
« The report calls for the government to invest in young researchers, without noting that the administration fired scores of young scientists and terminated hundreds of grants to early career researchers »
Carl T. Bergstrom@carlbergstrom.com · 2w ago1. The new Trump / Kratsios plan for US science appears explicitly designed to destroy US universities. Diverting money to tech donors is probably just a bonus. Gift link.
Can we stop "teaching" people to use outdated techniques to "spot" AI video. It does an immense amount of harm to teach people methods that produce false negatives!
NJ.com@nj.com · 2w agoAI-generated tornado images circulated on social media during New Jersey storms, prompting experts to share tips for spotting fakes.
« This is a wickedly hard problem. We have to stop telling ourselves it will be easy to solve »
New York Times Opinion@nytopinion.nytimes.com · 2w ago“If you insist on believing climate change isn’t real, or that its effects must be trivial, then whenever you find yourself overwhelmed by those effects you have to place the blame somewhere,” David Wallace-Wells writes.
C’était inévitable : « Mr. Winters’ case appears to be the first to claim that a chatbot’s health advice harmed someone seeking guidance about a medical condition »
The New York Times@nytimes.com · 2w agoA Florida pastor sued OpenAI on Wednesday, claiming that ChatGPT offered him “dangerous medical recommendations” that led to delayed care for a life-threatening pulmonary embolism. His lawsuit could be the first to argue that a chatbot’s advice harmed someone seeking medical guidance.
Avec l’utilisation de plus en plus répandue des outils d'intelligence artificielle par les étudiants, des universités se tournent vers des logiciels qui prétendent pouvoir détecter les textes rédigés par l’IA. Ceux-ci sont toutefois très peu fiables, selon un article publié dans la revue Nature.
Des détecteurs d’IA à la crédibilité défaillante
Avec l’utilisation de plus en plus répandue des outils d'intelligence artificielle par les étudiants, des universités se tournent maintenant vers des logiciels qui prétendent pouvoir détecter les textes rédigés par l’IA. Ceux-ci sont toutefois très peu fiables, révèle un article publié en juillet dans la revue Nature.
sciencepresse.qc.ca
Le groupe scientifique international indépendant sur l'IA mis sur pied par l’ONU en 2025 vient de publier son premier rapport. La conclusion : l’IA possède un potentiel considérable, mais se développe plus rapidement que notre capacité à bien l’encadrer.
Rapport de l’ONU sur l’IA : un potentiel sous surveillance
Le groupe scientifique international indépendant de l’intelligence artificielle mis sur pied par l’ONU en 2025 vient de publier son premier rapport. La conclusion : l’IA possède un potentiel considérable, mais se développe plus rapidement que notre capacité à bien l’encadrer.
sciencepresse.qc.ca
« The experts identified five risks — dangerous capabilities, competitive pressures, weapons and cyberattacks, concentrated power, and false information — as the most likely to produce the most severe harms over the next five years »
Timothy Caulfield@caulfieldtim.bsky.social · 2w ago🚨 These are the most urgent AI risks, according to 272 experts mitsloan.mit.edu/ideas-made-t... - dangerous capabilities - competitive dynamics - weapons and cyberattacks, - power centralization - AI spreading false or misleading information.
« desperate attempts to protect an alternate reality they've been fed algorithmically »
Charlie Warzel@cwarzel.bsky.social · 2w agosuch a good, sad window. these side quest conspiracies are just desperate attempts to protect an alternate reality they've been fed algorithmically. the utter popularity of this film is evidence that they live in a bubble of insanity. so they recruit new "evidence" so as not to confront that truth
« When we look back with nostalgia on the Spartans or even the boys who stormed Normandy, we have no way of knowing what their T levels were. Testosterone was first isolated in experiments in the 1930s and clinical use of serum testing wasn’t widespread until the 1970s »
Timothy Caulfield@caulfieldtim.bsky.social · 2w agoPete Hegseth, The Odyssey, and the #testosterone myth www.statnews.com/2026/07/21/c... @statnews.com TRT mythology growing for years, but won't make you brave or "manly" "No moral panic is more unoriginal than the fear that contemporary men are no longer as manly as the heroes who preceded them"
Intéressante histoire de la saisie par la police, en 2023, du matériel informatique du journal d’une petite localité du Kansas, saisie qui s’est par la suite avérée illégale. www.wnycstudios.org/podcasts/otm...
A Newspaper Raid and a Small Town Whodunnit | On the Media | WNYC Studios
In August of 2023, police raided a small local newspaper in rural Kansas, seizing computers and cellphones. Later, the paper’s publisher would discover there was no legal basis for th...
wnycstudios.org
Une analyse qu’on oublie souvent, et qui fait la différence dans bien des élections: « 75% of polled supporters of President Trump said they were "definitely going to vote" in the midterms. By comparison, 70% of non-MAGA Republicans and 63% of Democrats had the same response
Democrats, stop vying for Republican voters. It's a lost cause. | Opinion
Democrats would be wise to remember that Republicans expressing any frustration with Donald Trump would likely vote for him again if it were possible.
usatoday.com
« In rebooting “Little House,” a text that, for better and worse, has come to dominate our understanding of the era of Western expansion, the show once again holds up a mirror to the nation. We might use it to honestly ask ourselves how we got here, and who we have become in the process »
New York Times Opinion@nytopinion.nytimes.com · 3w agoThe Netflix reboot of “Little House on the Prairie” presents a much more complicated vision than the 1970s-era show, “fitting for a country deep in a painful and violent battle over its own history,” Glynnis MacNicol writes.
1/2 Une des origines de la fausse rumeur sur l’ingérence chinoise invoquée par Trump.
Ruth Ben-Ghiat@ruthbenghiat.bsky.social · 3w agoDesperate times mean desperate measures, if you are an unpopular leader afraid of going the way of your friend Viktor Orban. In my latest, I wrote about the ludicrous powerpoint Trump talked about in his speech and why it reinforces the fanaticism of Trump & MAGA. open.substack.com/pub/lucid/p/...
De plus en plus de gens utilisent des agents conversationnels basés sur l'intelligence artificielle (IA) pour rechercher de l'information concernant leur santé. Cependant, les réponses qu’ils obtiennent risquent d’être biaisées, révèle une nouvelle étude.
Santé : quand l’IA perpétue les préjugés
De plus en plus de gens utilisent des agents conversationnels basés sur l'intelligence artificielle (IA) pour rechercher de l'information concernant leur santé. Cependant, les réponses qu’ils obtiennent risquent d’être biaisées, révèle une nouvelle étude.
sciencepresse.qc.ca
Fascinant. Cette journaliste a appris qu’une biographie d’elle-même, écrite par l’IA et remplie d’erreurs, était en vente sur Amazon. « I clicked on the author, one John Crane Miller. “The Biography of Kashmir Hill” was one of 10 biographies that Mr. Miller had published in a single week »
Kashmir Hill@kashhill.bsky.social · 3w agoGot this text a couple months. Had to investigate it. www.nytimes.com/2026/07/16/t...
