Sign in

Joseph Reagle

@reagle.bsky.social
18 followers 6 following 151 posts

Hacking life, reading the comments, collaborating in good faith, and geeking out. Fond of the saying: "Don't believe everything you think."

PostsRepliesMedia
Joseph Reagle @reagle.bsky.social · 17/09/2026
reports on 6 incidents of AI misalignment, including "Uploading files to the internet in order to cite them." Wasn't Wikipedia, but Wikipedians will have to be wary of … “Our framework for reporting model misalignment” #ai #wikipedia openai.com/index/model-misalignment…
010
Joseph Reagle @reagle.bsky.social · 16/09/2026
reports on Pangram's AI detection service in the context of an instructor's experience. “The world's most reliable AI detector has a human problem” #ai #teaching www.bloomberg.com/news/features/202…
100
Joseph Reagle @reagle.bsky.social · 08/09/2026
reports on teachers moving from policing AI (ie Brisk draft history tool) to creating educational chatbots with … “These teachers are moving from AI policing to acceptance” #ai #teaching www.msn.com/en-us/money/general/the…
000
Joseph Reagle @reagle.bsky.social · 01/09/2026
reviews tech-billionaires' misreading of sci-fi cautionary tales as inspirations (via Lepore's *The Rise and Fall of… “How musk, Thiel, and Altman misread science fiction” #tech #vision #utopia www.theatlantic.com/books/2026/08/s…
000
Joseph Reagle @reagle.bsky.social · 01/09/2026
reports on possible CIO/biased edits to biographies of candidates for secretary-general of the… “The Wikipedia edits shaping UN secretary-general candidates' public records” #wikipedia #bias passblue.com/2026/08/31/the-wikiped…
001
Joseph Reagle @reagle.bsky.social · 27/08/2026
“Zuckerberg's the future is for everyone as Redaction poetry --” #ai #humor bengrosser.com/projects/zuckerbergs…
000
Joseph Reagle @reagle.bsky.social · 27/08/2026
Last semester I had Claude make sure the dates in my syllabus, lesson plan, and on Canvas were all the same; this semester I had it take the dates from the syllabus and update the other resources directly. #ai #teaching
000
Joseph Reagle @reagle.bsky.social · 27/08/2026
reports on MIT's five-month recommendation on AI and teaching, includes guiding principles and recommendations “Report on AI and education” #ai #teaching aiandeducation.mit.edu/report
000
Joseph Reagle @reagle.bsky.social · 26/08/2026
reports on survey of students' (N=6,851) chatbot preferences, with advice on how to use chatbots for essays---comes short of advising misconduct “Which AI is best for college essays in 2026? Gemini wins” #ai #education studyarena.com/blog/chatgpt-vs-clau…
100
Joseph Reagle @reagle.bsky.social · 22/08/2026
not new, but reports of use spreadsheets in dating, including a "friends feedback" tap “Singles are so burned out they're letting spreadsheets evaluate their dates” #lifehack #relationship www.wsj.com/lifestyle/relationships…
000
Joseph Reagle @reagle.bsky.social · 12/08/2026
reports on finding that 40% of people had used AI for financial advice in last 3 months; more than a third found it useful “AI chatbots are offering financial advice. Should you trust them ?” #advice #ai www.npr.org/2026/08/12/nx-s1-592481…
000
Joseph Reagle @reagle.bsky.social · 23/07/2026
finds that LLMs that ace medical exams can still fail to improve, or worsen, patients' decisions because of AI-human … “Publisher correction: Reliability of LLMs as medical assistants for the general public: A randomized preregistered study” #advice #ai dx.doi.org/10.1038/s41591-026-04404…
000
Joseph Reagle @reagle.bsky.social · 23/07/2026
“Improving health intelligence in ChatGPT” #advice #ai openai.com/index/improving-health-i…
000
Joseph Reagle @reagle.bsky.social · 23/07/2026
reports on a lawsuit by man against OpenAI because its chatbot offered "extremely dangerous medical recommendations" that delayed care for a lif… “ChatGPT led to a Man’s near-fatal health crisis, lawsuit claims” #advice #ai www.nytimes.com/2026/07/22/well/ope…
000
Joseph Reagle @reagle.bsky.social · 21/07/2026
“All brands --- the brand ledger” #buy ledger.worseonpurpose.com/brands
000
Joseph Reagle @reagle.bsky.social · 20/07/2026
demonstrates across five randomized experiments (N = 3,132) that merely making AI advice available nearly eliminates people's … “AI advice suppresses people's willingness to say "I don't know", even when the advice is wrong and accuracy is incentivized” #advice #ai arxiv.org/abs/2607.13562
000
Joseph Reagle @reagle.bsky.social · 18/07/2026
uses two pre-registered U.S. studies (Study 1 N=348, Study 2 N=347) to find that a 6-point increase in religiosity (from "never" prays/attends service to "dai… “In god we trust, in AI we ask: Religiosity and moral advice seeking in the AI age” #advice #ai dx.doi.org/10.31234/osf.io/z3y5d_v1
000
Joseph Reagle @reagle.bsky.social · 18/07/2026
argues "It's actually people of faith who more regularly turn to AI for moral guidance and spiritual advice.... were more likely to vie… “Are you there, Claude? It's me, Margaret.” #advice #ai www.washingtonpost.com/opinions/202…
000
Joseph Reagle @reagle.bsky.social · 15/07/2026
“Carefully using 'big-but-intimate data': A careful approach to using advice content (on Reddit)” #ethics #reddit www.youtube.com/watch?v=1R6F8EJX9x4
010
Joseph Reagle @reagle.bsky.social · 14/07/2026
analyzes elite college admissions to reveal a zero-sum mechanism where institutional priorities displace raw academic merit. Utilizing data from the SFFA v. Harva… “The sorting machine --- what did SFFA reveal about 'merit' in college admissions?” #academia #bias sorting-machine.pages.dev
020
Joseph Reagle @reagle.bsky.social · 13/07/2026
report StackOverflow now makes money from enterprise knowledge and AI services “Stack Overflow's forum is dead thanks to AI, but the company's still kicking... thanks to AI” #ai sherwood.news/tech/stack-overflow-f…
000
Joseph Reagle @reagle.bsky.social · 13/07/2026
places current decline is substantive reading into a historical and media ecological context (i.e., Ong) “The end of reading is here” #teaching #culture www.theatlantic.com/magazine/2026/0…
010
Joseph Reagle @reagle.bsky.social · 13/07/2026
"A brief, no fluff, summary of Mortimer Adlers’ How to Read a Book." “The best summary of how to read a book by Mortimer Adler” #teaching fs.blog/how-to-read-a-book
010
Joseph Reagle @reagle.bsky.social · 10/07/2026
“Pangram seems to work and students Don't like it” #teaching #ai reagle.org/joseph/pelican/2026/pang…
020
Joseph Reagle @reagle.bsky.social · 06/07/2026
reports that "By deploying advanced AI tools, we've reduced user exposure to spam by 20%, revoked nearly 2M fake votes daily, and slashed o… “How We're keeping Reddit real and safe in the AI era” #reddit #ai #moderation redditinc.com/news/how-were-keeping…
000
Joseph Reagle @reagle.bsky.social · 24/06/2026
notes self-help books are experiencing a precipitous decline in sales (26.3% decline in units from Q… “Has AI already killed how-to nonfiction? Sales trends, my personal data, and what it might mean for the future” #selfhelp #ai #advice tim.blog/2026/06/12/has-ai-already-…
000
Joseph Reagle @reagle.bsky.social · 24/06/2026
reflects on the disappointments since the 2014 efforts to get PR companies to edit Wikipedia e… “Wikipedia:Wikipedia Signpost/2026-06-21/Disinformation report” #wikipedia #bias en.wikipedia.org/w/index.php?title=…
000
Joseph Reagle @reagle.bsky.social · 10/06/2026
one approach to using Obsidian “How I use obsidian &mdash” #markdown stephango.com/vault
000
Joseph Reagle @reagle.bsky.social · 03/06/2026
finds that AI acts as an accessible 24/7 mental health resource, sati… “Mental health: Although screens can play a negative role, more than 6 out of 10 people turn to artificial intelligence for psychological support.” #ai #advice www.axa.com/en/press/press-releases…
000
Joseph Reagle @reagle.bsky.social · 02/06/2026
Neat! I could easily and securely access my home local AI from anywhere. “LM link • use your local models, remotely.” #ai #network lmstudio.ai/link
000
Joseph Reagle @reagle.bsky.social · 20/05/2026
Steven Rosenbaum book on "the effects of artificial intelligence on truth... included numerous made-up or misattributed quotes concocted by A.I." “Book on truth in the age of A.I. Contains quotes made up by A.I.” #ai www.nytimes.com/2026/05/19/business…
000
Joseph Reagle @reagle.bsky.social · 18/05/2026
“Comparison with other tools -” #markdown panache.bz/guide/comparison.html
000
Joseph Reagle @reagle.bsky.social · 08/05/2026
Pennsylvania sued Character.AI for allowing a (user created?) chatbot to represent… “Character.AI sued over chatbot that claims to be a real doctor with a license” #ai #advice arstechnica.com/tech-policy/2026/05…
000
Joseph Reagle @reagle.bsky.social · 08/05/2026
discusses Pangram's browser plug-in and the "pangramming" of others online, as well as the apparent weaknesses of the results “How an AI detector made me trust people less” #ai marcwatkins.substack.com/p/how-an-a…
000
Joseph Reagle @reagle.bsky.social · 07/05/2026
Google says upgrading AI mode/over with preview and links to expert advice (including on Reddit); also improving links “How AI mode and AI overviews help you explore the web” #ai #advice blog.google/products-and-platforms/…
000
Joseph Reagle @reagle.bsky.social · 04/05/2026
“Nine AI-Resilient in-class assessment activities” #teaching #ai learning.northeastern.edu/9-ai-resi…
000
Joseph Reagle @reagle.bsky.social · 30/04/2026
“Student AI use: Stupid, sneaky, and skillful” #ai #teaching reagle.org/joseph/pelican/2026/stud…
000
Joseph Reagle @reagle.bsky.social · 29/04/2026
"articles are researched and drafted by AI agents under human oversight" under CC0 (public domain) “Welcome to Agpedia” #ai #wikipedia agpedia.org/meta/welcome
000
Joseph Reagle @reagle.bsky.social · 01/04/2026
explains transformers via excellent interactive webpage poloclub.github.io/transformer-expl… #ai
000
Joseph Reagle @reagle.bsky.social · 01/04/2026
demonstrates that LLM sycophancy is pervasive and consequential for interpersonal behavior. In Study 1, 11 models affirmed users' actions ~50% more than humans across thre… “Sycophantic AI decreases prosocial intentions and promotes dependence” #advice #ai dx.doi.org/10.1126/science.aec8352
000
Joseph Reagle @reagle.bsky.social · 30/03/2026
I revamped my version of Wikipedias first 10K contributions, which allows you to browse *forward* in the first few weeks of Wikipedias history “Wikipedia 10K Redux, revamped” #wikipedia #history reagle.org/joseph/pelican/2026/10k-…
000
Joseph Reagle @reagle.bsky.social · 28/03/2026
interviews professors Bender and Reagle about motivations and implications of Wikipedia's AI guideline “Wikipedia bans AI-Generated text in articles under new editing policy” #wikipedia #ai #press decrypt.co/362510/wikipedia-bans-ai…
000
Joseph Reagle @reagle.bsky.social · 25/03/2026
describes en.WP policy prohibiting LLM for generating prose except as feedback an initial stages of translation “Wikipedia:Writing articles with large language models” #wikipedia #ai en.wikipedia.org/w/index.php?title=…
000
Joseph Reagle @reagle.bsky.social · 25/03/2026
reports that Wikipedia forbids use of generative AI except for (1) editors asking for feedback on their own prose and (2) as a… “Wikipedia has banned AI-generated text, with two exceptions” #wikipedia #ai www.howtogeek.com/wikipedia-banned-…
000
Joseph Reagle @reagle.bsky.social · 18/02/2026
describes adoption of AI detector in WikiEdu projects “Generative AI and Wikipedia editing what we learned in 2025 -” #wikipedia #ai #teaching wikiedu.org/blog/2026/01/29/generat…
000
Joseph Reagle @reagle.bsky.social · 28/01/2026
draft independent study shows Pangram achieved near-zero FNR (false negative rates) and FPR (false positive rates) of ≤ 0.005 robustly across models, threshold rules, ultra-short passages, "stubs" (≤ 50 wo… “Artificial writing and automated detection” #ai #teaching dx.doi.org/10.3386/w34223
000
Joseph Reagle @reagle.bsky.social · 23/01/2026
I just got my first pangram AI detection report on a WikiEdu student's contributions to Wikipedia; it was accurate “AI detector - free AI checker & content scanner” #teaching #wikipedia www.pangram.com
000
Joseph Reagle @reagle.bsky.social · 22/01/2026
It's amazing how small changes across many people can add up “Weight-loss drugs could save u.s. Airlines $580 million per year” www.nytimes.com/2026/01/19/travel/a…
000
Joseph Reagle @reagle.bsky.social · 22/01/2026
finds a significant association between daily generative AI use and depressive symptoms. Among 20,847 US adults (mean age 47.3 years; 49.5% female) surveyed 2025-04 to 2025… “Generative AI use and depressive symptoms among us adults” #advice #ai dx.doi.org/10.1001/jamanetworkopen.…
000
Joseph Reagle @reagle.bsky.social · 22/01/2026
provides AI writing smells “Wikipedia:Signs of AI writing” #ai #teaching en.wikipedia.org/w/index.php?title=…
000