53% of peer reviewers use AI. With high adoption in China/Africa, AI literacy in the Global South is urgent. While chatbots help, we need bespoke solutions for peer review with integrity and education baked in. www.researchinformation.info/news/most-pe... #PeerReview #AIinResearch

AI-assisted researchers are publishing 36-60% more—but their papers are less likely to get accepted despite sounding more complex. AI may be polishing weak science. Time for bespoke AI tools to help editors spot it. Happy New Year! 🥳 cen.acs.org/policy/publi... #AIinResearch #ResearchIntegrity

Merry Christmas! My publishing wishes: interoperable metadata, community-governed AI benchmarks, and Diamond OA that actually scales. The current APC model is unsustainable, especially for the Global South. scholarlykitchen.sspnet.org/2025/12/23/a... (Image Credits: ChatGPT 5.2)

Predatory journal emails are the Nigerian prince scams of academia! Red flags: poor design, grammar errors, WhatsApp contact, fake impact factors. If it feels off, it probably is. Verify everything! blog.cabells.com/2025/12/03/a... #AcademicPublishing #ResearchIntegrity

Excited to be co-facilitating a workshop at #R2RConf in London (24–25 Feb)! We're tackling responsible AI use in scholarly writing—building practical tools like checklists and disclosure templates. Come build with us! r2rconf.com/r2r-conferen...

Some stories I came across: Springer-Nature retracts 40 papers with questionable dataset, ethics journal publishes fake ChatGPT reference, paper mills sell authorship via WhatsApp. Researcher education, anyone? deevybee.blogspot.com/2025/11/the-... #ResearchIntegrity #AcademicPublishing

Journals demanding strict formatting before submission? Obsolete barrier. Accept clean formats, handle formatting after acceptance. Focus on ideas, not reference styles. blogs.lse.ac.uk/impactofsoci... #AcademicPublishing #OpenAccess

China proposes WAICO for global AI governance. Contrasts with US deregulation and EU's risk-based approach. Someone needs to lead on international AI coordination. China stepping up forces everyone to engage. www.nature.com/articles/d41... #ResponsibleAI #AIRegulation

AI writing patterns you can't unsee: "it's not X, it's Y" and the rule of three. They're everywhere! I use Claude to fix my writing slop. Now I'm hunting AI patterns in my own work. The irony is delicious. www.nytimes.com/2025/12/03/m... #AIinResearch #ResponsibleAI

AI universities are here! Personalized-pedagogy is sweet but the risks included cognitive steering and pedagogical debt. But remember, we might be redesigning education too fast without thinking about the consequences. www.nature.com/articles/d41... #AIinEducation #ResponsibleAI

62% of researchers now use AI for writing and data analysis. There are gains along with risks like hallucinations. What worries me even more if whether we'd overlook the need for critical evaluations of AI outputs. www.nature.com/articles/d41... #AIinResearch #ResearchIntegrity #ResponsibleAI

Maria Popova, you beauty - thanks for this! "... our most original and unexpected ideas arrive not when we strain the mind at the problem, but when we relax it and shift the beam of attention to something else entirely" www.themarginalian.org/2025/12/07/l...

NEJM's AI Companion claims it's trained only on their papers, yet it knows fictional characters. The foundational model is clearly leaking through. When what's promised doesn't match what the AI does, trust erodes fast. www.the-geyser.com/ai-on-nejm-w... #AIinResearch #ResponsibleAI

The AI reality even in Dec 2025! An OG #meme #academia #AI #LLM #AISlop #ChatGPT #Gemini #Claude #Mistral #author #writing #academipublishing #STM

Positive AI news: veteran professor says AI hasn't destroyed his classroom. Students still want to think independently. In the AI age, doing your own thinking is how we avoid going obsolete. www.nytimes.com/2025/11/25/m... #ResponsibleAI #AIinEducation

How to Be Human: Kahlil Gibran’s Recipe for Our Spiritual Perfection as a Species www.themarginalian.org/2025/11/29/k...

As former surgeon this is concerning: anti-vaccine movements seem to offer meaning and connection, which modern medicine may not. :( www.nytimes.com/2025/11/25/o... #Medicine #Healthcare

If you grew up in India/Africa/Asia in the late 90s–00s, WorldSpace radio shaped you. Just discovered the alt-rock station BOB has been restored by Radio Seribatu! Based on KEXP Seattle. Massive nostalgia hit. 🎸radioseribatu.com/bob

Traditional evaluation fails when AI generates essays instantly. Detection is theater. Embrace AI. Conversation assessments gauge understanding. Continuous evaluation works better. Treat AI as ally. www.nature.com/articles/d41... #HigherEducation #AIinEducation #ResponsibleAI

Multiple reports confirm: China leads research output and collaboration now. China-EU partnerships drive global networks. US output declining. China's R&D rivals US levels. Time to acknowledge this shift. www.nature.com/articles/d41... #Research #AcademicPublishing #GlobalScience

Positive AI news: veteran professor says AI hasn't destroyed his classroom. Students still want to think independently. In the AI age, doing your own thinking is how we avoid going obsolete. www.nytimes.com/2025/11/25/m... #ResponsibleAI #AIinEducation

CRediT turns ten—a taxonomy for who did what on a paper. Adoption still patchy. Why is research infrastructure so hard to sustain? www.nature.com/articles/d41... #ResearchIntegrity #AcademicPublishing #CRediT

Springer Nature flagged a paper for fabricated references—including one falsely attributed to Ivan Oransky. The irony? It criticized post-publication peer review. LLM-generated citations strike again. retractionwatch.com/2025/11/21/s... #ResearchIntegrity #AIinResearch #CitationVerification

You can now ask Gemini if an image is AI-generated, thanks to Google's SynthID. But It only flags images made with Google's own tools. :/ We're far from a catch-all solution—human-supervised verification remains the gold standard, especially for academic manuscripts. www.cnet.com/tech/service...

AI-related publications get 2x+ the expected citations. But there's a flip side - risk of opportunistic keyword usage to inflate metrics and possible neglect of other critical research areas. www.csescienceeditor.org/article/the-... #CitationMetrics #AIinResearch #AcademicPublishing

The j-metric: book weight ÷ years since doctorate. Can't game it. Finally rewards heft over salami-slicing. 😂 My word-heavy Master's thesis would have thrived. Stratospheric j-index, if only I'd kept up. www.nature.com/articles/d41... #AcademicPublishing #ResearchMetrics

Elizabeth Gadd in Nature: global university rankings are flawed and hinder innovation. Her three reforms: call them out as unfit, gather better data, push for nuanced assessment. I'm nodding along but wondering why more people aren't voicing this need.

Prompt injection attacks can trick AI reviewers into positive reviews. Both static and iterative attacks work on advanced AI models. Detection defenses exist but have limits. The concerning part? These attacks are transferable across models from different brands. arxiv.org/abs/2511.012...

A study of 25,114 biomedical manuscripts: only 5.7% of authors disclosed AI use - way lower than survey-reported usage. We need more AI literacy and education. And of course, more standardized guidelines for AI assistance and disclosure. www.medrxiv.org/content/10.1... #ResponsibleAI

Rick Anderson's latest SK article is about open access. He states that the "inevitable OA transition" narrative is wrong and that the transition already happened. Rick is spot on - different models solve different problems. We need pluralism, not dogma.

Clarivate's 2025 HCR list now excludes scientists co-authoring with people flagged for integrity issues. Math is back (alright!) but with tighter controls targeting excessive self-citation and suspicious patterns. www.nature.com/articles/d41...

COAR's semantic multilingual search proposal fixes what's been bothering me: AI made English accessible to everyone, but not omnidirectional. This is how I envision AI actually leveling the playing field. coar-repositories.org/news-updates... #OpenAccess

Elsevier survey: Chinese researchers far more confident in AI than US/UK counterparts. Adoption jumped to 58%, but governance/training lag badly. My bet? Same imbalance exists among STM publishing decisionmakers. www.elsevier.com/about/press-... #Research #AI

Found this in Maria Popova's Sunday newsletter featuring Carl Rogers' "On Becoming a Person." Rogers' test: before responding in an argument, restate the other person's view so accurately they agree with it. I'm hoping to try it next time. www.themarginalian.org/2025/11/11/c...

Iterative prompt injection attacks can manipulate AI reviewers across models. Defense mechanisms exist but can be beaten. At the Peer Review Congress, I learned about adversarial attacks. The onion peels further. arxiv.org/html/2511.01... #PeerReview #AIEthics

Ethan Mollick says give AI job interviews with real scenarios. Nature article recommends specialist tests by domain experts. Both say current benchmarks miss real-world judgment. And I agree, unsurprisingly! www.oneusefulthing.org/p/giving-you... www.nature.com/articles/d41... #AI #AITesting

AI is making publish-or-perish culture worse in academia. Pretty much the situation in all fields, as AI supercharges our treadmills. And as you think about this, why not listen to the AI-generated chart topper "Walk My Walk"? www.nature.com/articles/d41... #AcademicPublishing #AI

Interesting conversation with Fiona Hutton. CUP's "Experimental Results" and eLife's "Replication Studies" are vital for a future when publishing negative evidence will hopefully be the norm. www.youtube.com/watch?v=l9os... #ResearchIntegrity #ScholarlyPublishing #OpenScience

STM Association's *Publishing Decoded* is an open educational hub explaining how scholarly publishing works and is useful for anyone interested in the industry. stm-assoc.org/stm-launches... #ResearchIntegrity #ScholarlyPublishing #PublicTrust

Reproducibility failure persists in biomedical research despite a decade of awareness. Promoting and celebrating methodological rigor is the answer, I think. www.annualreviews.org/content/jour... #ResearchIntegrity #Reproducibility #OpenScience

AI is a clear double-edged sword on image manipulation, as you can imagine. But until the "good" AI gets to be really good, human oversight is essential. retractionwatch.com/2025/10/30/w... #ResearchIntegrity #ImageIntegrity #AIforScience

New study: MRI can now caption your thoughts into sentences. Medical use case: restoring communication. Sci-fi fan reaction: Stephenson and Gibson were right. 2045 reality: ordering pizza with my brain, probably. 😁😂 www.nature.com/articles/d41... #AIethics #NeuroAI #ResponsibleAI

AI assistants are becoming the first readers of scholarly content, reshaping discovery itself. The question isn't whether AI can assist—but what stays human in the process. scholarlykitchen.sspnet.org/2025/11/06/g... #ResponsibleAI #ScholarlyPublishing #AIethics

EDP Sciences has updated its AI & Ethics policy. Hope other publishers and societies follow suit. A key point: it reiterates that peer review has to remain a human-led process. www.edpsciences.org/en/news-high... #ResponsibleAI #PeerReview #ResearchIntegrity

"All sorts of persons, and every individual, has a place to fill in the world, and is important in some respects, whether he chooses to be so or not." - Nathaniel Hawthorne www.themarginalian.org/2025/10/31/h...

AI’s tendency to agree with users—“sycophancy”—can we extremely detrimental in science. . It distorts reasoning and undermines critique. Publisher AI tools must be designed such that this tendency is minimized. www.nature.com/articles/d41... #ResponsibleAI #ResearchIntegrity #PeerReview

GScholarLens adds a “Scholar h-index” that weights first & last authors more heavily. This is interesting but not a solution for preventing authorship manipulation. I'm fully behind contribution statements & CRediT taxonomy. www.nature.com/articles/d41... #ResearchIntegrity #Authorship #Metrics

“Free isn’t enough.” Public trust in science depends on transparency of process. Initiatives like the "Policy Pak" from Science enable this. Let's be inspired to make "openness" not only about access but also about communication. scholarlykitchen.sspnet.org/2025/10/23/g... #OpenScience

Cureus has been “editorially de-listed” from Web of Science, losing its impact factor after 13 months on hold. Indexers wield major influence but little transparency—publishers need clearer integrity audits and communication standards. newsletter.journalology.com/p/cureus-los...

I have never been more compelled to buy merchandise, but then it from The Marginalian! The Search for Meaning Cast in Clay: 19 Years of The Marginalian in 19 Ceramic Sentences www.themarginalian.org/2025/10/23/c...