@krishnakumarv.bsky.social
Emotionally engaging formats beat factual ones, and personal responsibility appeals do nothing or backfire. Something I think about a lot, both for professional communications and my own conversations with skeptics.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
El Niño goes viral — and scientists learn what makes people care about the climate
Evoking emotions is more likely to prompt people to engage with climate efforts than are data alone.
www.nature.com
September 24, 2026 at 4:23 PM
Emotionally engaging formats beat factual ones, and personal responsibility appeals do nothing or backfire. Something I think about a lot, both for professional communications and my own conversations with skeptics.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
What does global mean in academia? To me it means global northwest. It should not, but it does. Why not local rankings instead, calibrated across geographies?
www.researchprofessionalnews.com/rr-news-afri...
www.researchprofessionalnews.com/rr-news-afri...
https://www.researchprofessionalnews.com/rr-news-africa-pan-african-2026-9-do-african-universities-need-their-own-locally-relevant-ranking/
www.researchprofessionalnews.com
September 23, 2026 at 3:56 PM
What does global mean in academia? To me it means global northwest. It should not, but it does. Why not local rankings instead, calibrated across geographies?
www.researchprofessionalnews.com/rr-news-afri...
www.researchprofessionalnews.com/rr-news-afri...
Tenure-track faculty fell to 31.8% from 53.1% since 1987, and peer review runs on that pool. I'm for paid review, but AI-assisted tools feel more feasible.
scholarlykitchen.sspnet.org/2026/09/14/a...
scholarlykitchen.sspnet.org/2026/09/14/a...
Ask the Chefs: Peer Review Capacity — Uncomfortable Truths and Overlooked Conversations - The Scholarly Kitchen
For Peer Review Week 2026, we asked the Chefs two questions: What is one uncomfortable truth about peer review that the scholarly publishing community doesn’t talk about enough? And, beyond…
scholarlykitchen.sspnet.org
September 22, 2026 at 3:37 PM
Tenure-track faculty fell to 31.8% from 53.1% since 1987, and peer review runs on that pool. I'm for paid review, but AI-assisted tools feel more feasible.
scholarlykitchen.sspnet.org/2026/09/14/a...
scholarlykitchen.sspnet.org/2026/09/14/a...
Springer Nature now grades AI use by how much it affects interpretation, evaluation or decision-making. Subjective, but the better way to go, and OUP's survey shows plenty of disclosure gaps.
group.springernature.com/gp/group/med...
group.springernature.com/gp/group/med...
Springer Nature launches updated AI policies and guidance to support responsible AI use | Springer Nature Group | Springer Nature
New framework provides practical, future-focused guidance for authors, reviewers and editors in an evolving AI landscape London | Berlin | New York |
group.springernature.com
September 22, 2026 at 1:47 PM
Springer Nature now grades AI use by how much it affects interpretation, evaluation or decision-making. Subjective, but the better way to go, and OUP's survey shows plenty of disclosure gaps.
group.springernature.com/gp/group/med...
group.springernature.com/gp/group/med...
An AI review that reads everything submitted at once, done right, is a good thing, isn't it? I also back the two-layer paper, a human narrative over a machine-readable evidence base.
www.linkedin.com/pulse/has-sc...
www.linkedin.com/pulse/has-sc...
https://www.linkedin.com/pulse/has-scientific-paper-become-too-big-human-peer-review-dino-di-carlo-2swuc/
www.linkedin.com
September 21, 2026 at 12:55 PM
An AI review that reads everything submitted at once, done right, is a good thing, isn't it? I also back the two-layer paper, a human narrative over a machine-readable evidence base.
www.linkedin.com/pulse/has-sc...
www.linkedin.com/pulse/has-sc...
Generation has become cheap and verification has not, so this piece splits the responsibility three ways. Publishers, institutions and AI providers. I love this split.
www.researchinformation.info/analysis-opi...
www.researchinformation.info/analysis-opi...
Scientific literacy must keep pace with artificial intelligence - Research Information
AI is making science easier to access but harder to trust. Scientific literacy must catch up, writes Giulia Stefenelli
www.researchinformation.info
September 18, 2026 at 3:02 PM
Generation has become cheap and verification has not, so this piece splits the responsibility three ways. Publishers, institutions and AI providers. I love this split.
www.researchinformation.info/analysis-opi...
www.researchinformation.info/analysis-opi...
AI and psychology, a fascinating mix. For introverts like me, AI is a bridge from private thought to outward expression. Extraverts get feedback, so they can use it for introspection.
theconversation.com/whether-your...
theconversation.com/whether-your...
Whether you’re an introvert or extravert may decide how you should use AI
We should adapt AI to how extraverts and introverts think.
theconversation.com
September 17, 2026 at 4:23 PM
AI and psychology, a fascinating mix. For introverts like me, AI is a bridge from private thought to outward expression. Extraverts get feedback, so they can use it for introspection.
theconversation.com/whether-your...
theconversation.com/whether-your...
Just about everywhere but the US is incentivizing researchers to... well... do research. India joins the party! Its Prime Minister Research Chair funds 120 returning researchers of Indian origin at up to 50 million rupees each over five years.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
India has an ambitious plan to lure scientists back — will it work?
Initiatives to attract Indian researchers who are working abroad do not offer long-term income security, some scientists say.
www.nature.com
September 16, 2026 at 3:56 PM
Just about everywhere but the US is incentivizing researchers to... well... do research. India joins the party! Its Prime Minister Research Chair funds 120 returning researchers of Indian origin at up to 50 million rupees each over five years.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
Just over half of 1,645 researchers used AI for review tasks last year, and most journals ban it. I'm still going out on a limb and saying AI assistance will only enhance human review.
www.science.org/content/arti...
www.science.org/content/arti...
Can AI help solve the peer-review crisis? Here are its promises and pitfalls
Amid a shortage of human reviewers, tests indicate AI can assist—but is not a savior
www.science.org
September 15, 2026 at 3:37 PM
Just over half of 1,645 researchers used AI for review tasks last year, and most journals ban it. I'm still going out on a limb and saying AI assistance will only enhance human review.
www.science.org/content/arti...
www.science.org/content/arti...
OpenAI hit its automated research intern goal. Wielded well, it could only benefit research. Its Astra model loops on queries, not readable steps, and I don't want to lose track of why AI does what it does.
techcrunch.com/2026/09/02/o...
techcrunch.com/2026/09/02/o...
OpenAI’s new reasoning technique alarms AI safety experts | TechCrunch
OpenAI’s new Astra model will use “recurrent depth,” a technique that allows the model to operate outside of the sequential thinking that characterizes most reasoning models.
techcrunch.com
September 14, 2026 at 12:55 PM
OpenAI hit its automated research intern goal. Wielded well, it could only benefit research. Its Astra model loops on queries, not readable steps, and I don't want to lose track of why AI does what it does.
techcrunch.com/2026/09/02/o...
techcrunch.com/2026/09/02/o...
An 89,000-word novel written entirely by AI agents, five hours and $209.55 in tokens. This is the type of stuff you may find on Kindle Unlimited, something I posted about recently.
antipodes.substack.com/p/superslop-...
antipodes.substack.com/p/superslop-...
Superslop, and How It Will Kill Your Book Deal
This is the opening of a novel that led to a four-way bidding war and a $3.5 million, two-book deal.
antipodes.substack.com
September 13, 2026 at 12:55 PM
An 89,000-word novel written entirely by AI agents, five hours and $209.55 in tokens. This is the type of stuff you may find on Kindle Unlimited, something I posted about recently.
antipodes.substack.com/p/superslop-...
antipodes.substack.com/p/superslop-...
A preprint found AI writing in 14,419 self-published titles on Amazon, none disclosing it. The rebuttal says you cannot flood an already saturated market. I'm happy to see the counter point, but a saturated market with poor quality can still get worse.
arxiv.org/abs/2607.20349
arxiv.org/abs/2607.20349
Generative AI floods and dilutes the market for books
Generative AI can produce book-length works of fiction at near-zero cost. These books are often dismissed as low-quality ``slop'' that buyers will ignore, and are assumed to carry little commercial…
arxiv.org
September 12, 2026 at 12:55 PM
A preprint found AI writing in 14,419 self-published titles on Amazon, none disclosing it. The rebuttal says you cannot flood an already saturated market. I'm happy to see the counter point, but a saturated market with poor quality can still get worse.
arxiv.org/abs/2607.20349
arxiv.org/abs/2607.20349
The best thing I got out of this interview is on textbooks. Publishers cannot compete with ChatGPT on breadth and should not try. A well-made textbook defines the field.
www.ludenso.com/blog/thought...
www.ludenso.com/blog/thought...
Thought Leaders in Publishing: Henriette Thiesen on AI and the future of the textbook
An interview with Henriette Thiesen, Publishing Director at Gyldendal and a Ludenso partner, part of Ludenso's series with leaders shaping educational publishing. The conversation explores what's at…
www.ludenso.com
September 11, 2026 at 3:02 PM
The best thing I got out of this interview is on textbooks. Publishers cannot compete with ChatGPT on breadth and should not try. A well-made textbook defines the field.
www.ludenso.com/blog/thought...
www.ludenso.com/blog/thought...
Clay published its internal AI writing policy. Stand behind every sentence. Writing is thinking. Longer is not better. I really like it.
www.linkedin.com/posts/vaanan...
www.linkedin.com/posts/vaanan...
https://www.linkedin.com/posts/vaanand_we-just-instituted-an-official-ai-writing-share-7492584308881330176-rcmv/
www.linkedin.com
September 10, 2026 at 4:23 PM
Clay published its internal AI writing policy. Stand behind every sentence. Writing is thinking. Longer is not better. I really like it.
www.linkedin.com/posts/vaanan...
www.linkedin.com/posts/vaanan...
Undoubtedly, this study highlights/demonstrates that we are not being humble and hedgy anymore. Maybe this is a sign that AI is writing for us? Could it be? It might very well be, perhaps.
PS: I'm definitely NOT losing my maybes.
mahmudomar.com/essays/are-w...
PS: I'm definitely NOT losing my maybes.
mahmudomar.com/essays/are-w...
Are we losing our maybes?
A personal note on hedging, boosters, and a change in the language of scientific writing.
mahmudomar.com
September 9, 2026 at 3:56 PM
Undoubtedly, this study highlights/demonstrates that we are not being humble and hedgy anymore. Maybe this is a sign that AI is writing for us? Could it be? It might very well be, perhaps.
PS: I'm definitely NOT losing my maybes.
mahmudomar.com/essays/are-w...
PS: I'm definitely NOT losing my maybes.
mahmudomar.com/essays/are-w...
If you work with Claude Code or Codex, you will have been scalded enough to know that agentic systems fail on stale data, not weak reasoning. Context engineering is the differentiator now.
thenextweb.com/news/whats-n...
thenextweb.com/news/whats-n...
What's next for AI and how to get there
Frontier models have converged. Stanford data shows top systems now score within 5% of each other. Oxylabs SVP Gediminas Rickevičius argues the competitive edge has shifted from model selection to…
thenextweb.com
September 8, 2026 at 3:37 PM
If you work with Claude Code or Codex, you will have been scalded enough to know that agentic systems fail on stale data, not weak reasoning. Context engineering is the differentiator now.
thenextweb.com/news/whats-n...
thenextweb.com/news/whats-n...
The Scholarly Kitchen now enforces disclosure of AI-assisted writing and treats writing as thinking. I do see the point, but I have also enjoyed using AI to improve my writing while keeping my voice.
scholarlykitchen.sspnet.org/2026/08/27/a...
scholarlykitchen.sspnet.org/2026/08/27/a...
AI Use in The Scholarly Kitchen - The Scholarly Kitchen
Some thoughts on the reader response the Scholarly Kitchen has seen regarding AI use by authors, a recent Nature News article on the efficacy of AI detection tools, and why it's important to keep…
scholarlykitchen.sspnet.org
September 7, 2026 at 12:55 PM
The Scholarly Kitchen now enforces disclosure of AI-assisted writing and treats writing as thinking. I do see the point, but I have also enjoyed using AI to improve my writing while keeping my voice.
scholarlykitchen.sspnet.org/2026/08/27/a...
scholarlykitchen.sspnet.org/2026/08/27/a...
64% of people now feel more confident challenging professional advice after checking it with AI. I am AI-forward and I see the point. But this survey shows the bad side clearly.
www.techradar.com/pro/ai-is-ma...
www.techradar.com/pro/ai-is-ma...
AI is making people less confident in professionals such as doctors and teachers
AI is empowering people to ask questions - but the wrong kind
www.techradar.com
September 4, 2026 at 3:01 PM
64% of people now feel more confident challenging professional advice after checking it with AI. I am AI-forward and I see the point. But this survey shows the bad side clearly.
www.techradar.com/pro/ai-is-ma...
www.techradar.com/pro/ai-is-ma...
A DOI tells you which article. It does not tell you which version an AI actually read. The metadata around the identifier is where the work is.
scholarlykitchen.sspnet.org/2026/08/19/g...
scholarlykitchen.sspnet.org/2026/08/19/g...
Guest Post — AI Can Find the Article. Can It Tell Which Version It Used? - The Scholarly Kitchen
In this AI era, establishing trust and validating article versions with persistent identifiers (PIDs) is more important than ever.
scholarlykitchen.sspnet.org
September 3, 2026 at 4:23 PM
A DOI tells you which article. It does not tell you which version an AI actually read. The metadata around the identifier is where the work is.
scholarlykitchen.sspnet.org/2026/08/19/g...
scholarlykitchen.sspnet.org/2026/08/19/g...
One paper mill in a dataset of seven supplies thirteen of the fourteen product categories on sale. Textbooks, articles, patents, design registrations, a national award. It operates out of India. Worrying.
arxiv.org/abs/2604.24576
arxiv.org/abs/2604.24576
BuyTheBy: A dataset of 18,710 text-based paper mill advertisements with 51,812 timestamped prices
The study of paper mills and similar businesses operating in the market for academic and education fraud services is frustrated by the lack of market price data on their various offerings. Here, we…
arxiv.org
September 2, 2026 at 3:56 PM
One paper mill in a dataset of seven supplies thirteen of the fourteen product categories on sale. Textbooks, articles, patents, design registrations, a national award. It operates out of India. Worrying.
arxiv.org/abs/2604.24576
arxiv.org/abs/2604.24576
Owner slots on Australian design registrations, $95 a head. A real government record that reads as a patent on a CV. Worrying.
www.theguardian.com/science/2026...
www.theguardian.com/science/2026...
Science or fiction? Shadowy ‘paper mills’ let you pay to be a published author – of fraudulent research
Paper mill activity has exploded in recent years, polluting the scientific ecosystem with falsified data and compromising medical research
www.theguardian.com
September 1, 2026 at 11:45 AM
Owner slots on Australian design registrations, $95 a head. A real government record that reads as a patent on a CV. Worrying.
www.theguardian.com/science/2026...
www.theguardian.com/science/2026...
120+ retractions over manipulated peer review in guest-edited issues. The guidance for guest editor collections exists. Put it into practice and automate.
retractionwatch.com/2026/08/12/b...
retractionwatch.com/2026/08/12/b...
Biology journal pulls 100+ articles and counting from special issues for peer review manipulation
Elsevier’s International Journal of Biological Macromolecules has issued more than 100 retractions and counting for manipulated peer review in guest-edited special issues. The publisher has said on…
retractionwatch.com
August 31, 2026 at 12:55 PM
120+ retractions over manipulated peer review in guest-edited issues. The guidance for guest editor collections exists. Put it into practice and automate.
retractionwatch.com/2026/08/12/b...
retractionwatch.com/2026/08/12/b...
You can forge a painting but not its paper trail. Generator against detector is a never-ending game. Base the judgment on the coauthors, the funding and the affiliations, not just the prose.
www.jmir.org/2026/1/e109033
www.jmir.org/2026/1/e109033
https://www.jmir.org/2026/1/e109033
www.jmir.org
August 28, 2026 at 3:01 PM
You can forge a painting but not its paper trail. Generator against detector is a never-ending game. Base the judgment on the coauthors, the funding and the affiliations, not just the prose.
www.jmir.org/2026/1/e109033
www.jmir.org/2026/1/e109033
About 90% of biomedical papers from December 2025 show signs of AI-assisted writing, against a last published estimate of 13.5%. Some of that is a better estimator. The evidence still points one way. AI assistance is standard.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
Staggering 90% of biomedical papers now show signs of AI help
The figure, which is much higher than previous estimates of large-language-model use, is for papers published in December and archived in PubMed.
www.nature.com
August 27, 2026 at 4:23 PM
About 90% of biomedical papers from December 2025 show signs of AI-assisted writing, against a last published estimate of 13.5%. Some of that is a better estimator. The evidence still points one way. AI assistance is standard.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
A review scores journal AI policies on a five step readiness ladder. Most sit at the declarative level, disclosure rules and nothing behind them. I doubt we will ever reach the top two.
sciencepolicyreview.org/wp-content/u...
sciencepolicyreview.org/wp-content/u...
https://sciencepolicyreview.org/wp-content/uploads/securepdfs/2026/07/v7_01_ai_education.pdf
sciencepolicyreview.org
August 26, 2026 at 3:56 PM
A review scores journal AI policies on a five step readiness ladder. Most sit at the declarative level, disclosure rules and nothing behind them. I doubt we will ever reach the top two.
sciencepolicyreview.org/wp-content/u...
sciencepolicyreview.org/wp-content/u...