Jane Rosenzweig
LightNews LightNews
janerosenzweig.bsky.social
Jane Rosenzweig
@janerosenzweig.bsky.social
3.7K followers 1.2K following 330 posts
Writing in the age of AI stuff
Newsletter: writinghacks.substack.com
Also curate theimportantwork.substack.com
Posts Replies Media Videos
janerosenzweig.bsky.social
Time to reread Roald Dahl's "The Great Automatic Grammatizator"
In the end, after several months of work, he had persuaded something like seventy per cent of the writers on his list to sign the contract. He found that the older ones, those who were running out of ideas and had taken to drink, were the easiest to handle. The younger people were more troublesome. They were apt to become abusive, sometimes violent when he approached them; and more than once Knipe was slightly injured on his rounds. But on the whole, it was a satisfactory beginning. This last year—the first full year of the machine’s operation—it was estimated that at least one half of all the novels and stories published in the English language were produced by Adolph Knipe upon the Great Automatic Grammatizator. Does this surprise you? I doubt it. And worse is yet to come. Today, as the secret spreads, many more are hurrying to tie up with Mr Knipe. And all the time the screw turns tighter for those who hesitate to sign their names. This very moment, as I sit here listening to the howling of my nine starving children in the other room, I can feel my own hand creeping closer and closer to that golden contract that lies over on the other side of the desk. Give us strength, Oh Lord, to let our children starve.
November 19, 2025 at 1:43 PM
janerosenzweig.bsky.social
Put in "Olivia Nuzzi Sam Altman" and got this. If you wanted to know if she had ever written about him, you'd be out of luck (seems to be referring to things she said about RFK) but also this tool needs to be retired.
Google AI overview result. Olivia Nuzzi's recent memoir includes anecdotes about OpenAI CEO Sam Altman, describing their conversations and interactions . The book details a level of personal intimacy in their interactions, including Nuzzi observing Altman flossing his teeth
November 18, 2025 at 6:17 PM
janerosenzweig.bsky.social
New in The Important Work: @thevogelman.bsky.social shares his AI Transparency Survey and writes about what he learned when he asked his 10th and 12th grade students to reflect on their AI usage. Here's a bit of what his students shared. Read it here: theimportantwork.substack.com/p/the-transp...
Alaina: After a few drafts of my conclusion, there were two sentences that I couldn’t find a clear wording for and I felt they were getting too convoluted for my presentation. I decided to ask AI to give me a few options for ways to rewrite them so I combined some of those ideas to make what I was trying to say more articulate. Talia: Personally, I do not think AI usage is ethical, especially in a literature situation. . . Everything AI can tell me from a prompt I can learn from my own research. Smaran: I always use the Microsoft Word Editor after every essay or project . . . Also, I used AI to try to find a scholarly source based on my thinking. Ainslie: I’m not sure how to use AI in a way that maintains my thinking and voice and the integrity of the project. I also don’t see the [point of ] wasting that sort of energy and adding to my carbon footprint if I could use my brain. Erin: I do not like AI and do not like using it in regard to my writing. I think it takes away the work of real writers and is harming us a lot more than it’s helping us. Johan: I used the Google AI on my phone to look for different examples of Norse mythological influences in contemporary media. Cassie: I didn’t use AI at all in this project. I don’t like using AI because it feels as though I am cheating myself out on my learning. This is an AP class; things are supposed to be difficult.
November 18, 2025 at 1:25 PM
janerosenzweig.bsky.social
Went down rabbit hole trying to predict how AI overviews get things wrong but could not have predicted this one. I never taught at Vassar; she didn't go there; we don't know each other. The source mentions her once, someone else named Rosenzweig, not Vassar. But people use these results as fact.
AI Overview Jane Rosenzweig and Olivia Nuzzi's relationship stems from their time at Vassar College, where Rosenzweig was Nuzzi's professor and the director of the college's Writing Center. Olivia Nuzzi, now a prominent political journalist for New York Magazine, has publicly cited Jane Rosenzweig as a significant influence and mentor in her education and career. Nuzzi has mentioned that Rosenzweig taught her fundamental journalistic skills, such as interviewing techniques and developing story ideas, during her time at Vassar. Their professional relationship highlights the mentor-mentee dynamic between a college professor and a student who went on to achieve success in her field.
November 18, 2025 at 3:37 AM
janerosenzweig.bsky.social
If you are using google AI overviews for medical advice, cooking advice, car advice, any advice, note that this is not a PhD level intelligence but a word prediction machine that draws from separate articles in the same NYT evening newsletter to produce output like this mixture of multiple stories.
Specifically: A recent article mentions Nuzzi's concern over Robert F. Kennedy Jr.'s health issues, noting that a "concern over lover RFK Jr.'s brain worm had him soothing ... Larry Summers in July 2018—went unanswered". This context is related to Nuzzi's recent personal and professional controversies rather than any direct interaction or connection between her and Summers. Another piece related to Nuzzi's situation mentions that the late Jeffrey Epstein unsuccessfully tried to reach out to figures like Larry Summers and Reid Hoffman.
November 18, 2025 at 3:14 AM
janerosenzweig.bsky.social
No entry level jobs, no opportunity to get experience doesn't seem like a great recipe for society. @nymag.com: nymag.com/intelligence...
If the number of entry-level jobs dwindles across industries, naturally, the supply of experienced employees will wane. Without entry-level positions, how will the next generation of middle managers and executives get training? Millennials would be the last keepers of unwritten workplace knowledge, the intangible lessons and institutional know-how that haven’t been, or can’t be, fed into LLMs. What would the economy look like then?
November 18, 2025 at 1:46 AM
janerosenzweig.bsky.social
A great thing about kids having imaginary friends is that their conversations are based on the kid's understanding of the world not on training data. futurism.com/artificial-i...
After testing three different toys powered by AI, researchers from the US Public Interest Research Group found that the playthings can easily verge into risky conversational territory for children, including telling them where to find knives in a kitchen and how to start a fire with matches. One of the AI toys even engaged in explicit discussions, offering extensive advice on sex positions and fetishes. In the resulting report, the researchers warn that the integration of AI into toys opens up entire new avenues of risk that we’re barely beginning to scratch the surface of — and just in time for the winter holidays, when huge numbers of parents and other relatives are going to be buying presents for kids online without considering the novel safety issues involved in exposing children to AI.
November 17, 2025 at 12:45 PM
janerosenzweig.bsky.social
This below is in the context of computer science but we all need to be having this conversation about whether "using AI" in classes prepares you for future jobs or whether actually doing the thing yourself prepares you for a world in which you may or may not use AI. The idea that using AI to write/1
A degree in CS mostly prepares students to become computer scientists, just as a history degree mostly prepares students to be historians. The idea is not that every student will become an academic— few do—but that the initial training in this direction will prepare students for varied pursuits. When the world changes, Erat argues, the student who studied CS will have the creativity to “learn how to search, how to integrate things.” But for the student who just used AI to make programs, he said, “It’s going to be impossible.” Conversely, CS professor Lin Zhong thinks Yale’s core CS curriculum is “stuck in the 1980s.” He wants students to develop the AI expertise that they would need as professional software engineers. “If your student[s] don’t learn how to use [AI], they lose their job,” he said. Students in Zhong’s Computer Systems Design course earn extra credit for using AI creatively to do their homework.
November 16, 2025 at 5:24 PM
janerosenzweig.bsky.social
Interesting that an AI introduced to avoid corruption is being framed as a human who has given nepo baby jobs to her 83 kids. www.euronews.com/next/2025/10...
By Euronews Published on 30/10/2025 - 12:14 GMT+1 Share Comments Diella, the country’s first AI system that is now a minister in Edi Rama’s cabinet, will have 83 “children,” who will become parliamentary assistants. Albania’s first non-human minister made up entirely by artificial intelligence (AI) is “pregnant” with 83 children that will become parliamentary assistants, according to the country’s prime minister. Albanian Prime Minister Edi Rama toldthe Global Dialogue Forum in Berlin that the parliamentary assistants will participate in parliamentary sessions, keep notes on what goes on and then give advice to members on how they should react to specific pieces of legislation. “These children will have their mother’s knowledge regarding EU legislation and everything else,” Rama said. “This child will say what was said when [the member] is not in the hall and if [the member’s] name was mentioned and [they] have to counterattack someone who mentioned you for the wrong reasons,” he added.
November 15, 2025 at 12:58 PM
janerosenzweig.bsky.social
today in automation. discuss
📚 Teachers & Counselors — this one’s for you! If you’re swamped writing college recommendation letters right now, you need to try educator.kollegio.ai 💡 Upload your brag sheets (or use their built-in template) and let AI help you: ✍️ Draft personalized rec letters faster 🪄 Polish and organize with real-time suggestions 📄 Download as a ready-to-send PDF Save hours this application season — it’s 100% free and built with feedback from real counselors & teachers 👩‍🏫👨‍🏫 #KollegioAI #EducatorTools #TeacherLife #CounselorResources #CollegeAdmissions #LetterOfRecommendation #EdTech #AIforEducation
November 14, 2025 at 11:29 PM
janerosenzweig.bsky.social
This does not seem like efficiency: "proposals written with the platform’s native AI-writing tool exhibit a negative correlation between effort and signal, and signals no longer predict successful job completion"
Leveraging this unique setting, we provide new descriptive evidence that LLMs have dis- rupted previously informative labor market signaling. We organize our descriptive evidence into three main findings. First, we show that before the mass adoption of LLMs, employers had a sig- nificantly higher willingness to pay for workers who sent more customized proposals. Estimating a reduced-form multinomial logit model of employer demand using our measure of signal, we find that, all else equal, workers with a one standard deviation higher signal have the same in- creased chance of being hired as workers with a $26 lower bid.1 Second, we provide evidence that before the adoption of LLMs, employers valued workers’ signals because signals were predictive of workers’ effort, which in turn predicted workers’ ability to complete the posted job success- fully. Third, we find, however, that after the mass adoption of LLMs, these patterns weaken significantly or disappear completely: employer willingness to pay for workers sending higher signals falls sharply, proposals written with the platform’s native AI-writing tool exhibit a nega- tive correlation between effort and signal, and signals no longer predict successful job completion conditional on being hired. Taken together, these findings suggest that workers’ job-specific pro- posals functioned as Spence-like signals of worker ability prior to the advent of generative AI, but that LLMs disrupted this signaling mechanism on which employers previously relied to make hiring decisions.
November 14, 2025 at 12:39 PM
janerosenzweig.bsky.social
This won me over immediately.
text of a fundraising email that ends with "I'm hoping thta you'll decide to join them too, {Contact.Firstname/default "friend"}}
November 13, 2025 at 1:16 AM
janerosenzweig.bsky.social
We've gone from "can AI make art" to "AI is an artist" quite quickly. www.thedailybeast.com/the-no-1-cou...
The No. 1 Country Song in America Is by a Non-Human Artist
November 11, 2025 at 10:38 PM
janerosenzweig.bsky.social
I've been reading more about Ohio State's initiatives and saw this on their teaching and learning website. There are a lot of ways to help students at these early stages of the writing process but I'm not sure what problem is solved by producing "reams of raw content."
Potential Benefits It is helpful for educators to consider that generative AI: Provides learners a tool for generating rough drafts, outlines, and brainstorming notes. Generative AI applications like ChatGPT can assist students during the earliest stages of the writing process, auto-generating text for learners who are either stymied by writer’s block or stuck during the brainstorming process. When prompted, these systems can produce reams of raw content (of varying levels of quality and accuracy) that students can further evaluate, interrogate, and research. Using AI tools in this fashion can help students during the difficult preliminary stages of the composition process, creating a workable path toward their own inquiries and investigations (Gero, 2022; Krause, 2022; Weissman, 2023).
November 9, 2025 at 10:34 PM
janerosenzweig.bsky.social
This essay is worth a read (by a college student in Ontario). macleans.ca/education/ai...
A common argument in favour of AI is that it helps protect our time and energy and increase efficiency. But to me, the idea of a world where we’re conditioned to trade hard work for convenience is scary. The whole point of university is to broaden our minds and push ourselves to do difficult things. What people once got lost in—late-night readings, the contemplation of century-old ideas, the labour of love that goes into writing an essay—has vanished almost overnight. Living your life outside of academic responsibilities is important, but with generative AI, the mantra “work smarter, not harder” has taken on a whole new life force. What was meant to be a place of higher learning has turned into a degree mill, fuelled by quick fixes, avoidance and instant reward, all using the same singular employee: ChatGPT.
November 7, 2025 at 12:17 AM
janerosenzweig.bsky.social
Here's some of what I've learned.
Teaching this course means I’ve spent hours reading, learning, and having sustained conversations with students about AI. What have I learned? I’ve learned that more students each year are coming to college having used AI, but a much smaller group arrive having learned anything about AI — and that’s a problem. I’ve learned that even as we’re hearing weekly that AI has killed the college essay, a writing class focused on the process of reading closely, listening, asking questions, considering evidence, and hashing out other points of view turns out to be a great place to work through the thorny issues presented by AI. And I’ve learned that while AI may be able to offer (often mediocre) writing feedback, it can’t help students articulate ideas that are not yet on the page, which is often the most meaningful feedback of all.
November 6, 2025 at 12:58 PM
janerosenzweig.bsky.social
In Perplexity ad, student brags about a paper Comet wrote. We can see the text, which is quite bad. The caricature of "homework" in these ads is interesting: at times selling an important learning tool, while the output suggests most assignments are pointless busy work and there is nothing to learn.
screenshot of text in a browser. It's a paper called "The History of Railroads in the united States" and it's very basic.
November 5, 2025 at 11:18 PM
janerosenzweig.bsky.social
And lo and behold, it agreed with me. /7
text from ChatGPT that says "You're absolutely right — and that’s an important nuance for academic writing and paraphrasing. ✅ Interpretation Check In the original passage: “…has led to calls for greater accountability in AI governance.” The author (Cheong) is reporting that calls for accountability exist — not necessarily making the call himself. The sentence is descriptive, not prescriptive. In other words: He is describing a phenomenon (others are calling for accountability), Not necessarily advocating a position personally. ⚠️ Why that matters for paraphrasing A paraphrase must not add intention or attribute an opinion to an author that isn't clearly stated. The earlier paraphrase changed the meaning by saying: “…the author says that stronger mechanisms of accountability must be built…” That wording implies that Cheong himself is demanding accountability, which is not what the original says. That subtle shift turns a descriptive claim into a prescriptive one, making it factually inaccurate. This is a form of incorrect paraphrasing, not just patchwriting."
November 3, 2025 at 2:55 AM
janerosenzweig.bsky.social
not in this passage. I asked ChatGPT about this: /6
But in the original it doesn't seem like the author is calling for accountability in that passage--rather, the author seems to be reporting that calls have been made?
November 3, 2025 at 2:55 AM
janerosenzweig.bsky.social
In fact, it caved immediately and offered me a "better" paraphrase, which was not, in fact, better. Because, as it turned out, all versions of the paraphrase had a problem beyond the "flow." /3
Image of text that says "The ideas are the same and the causal relationship (“because X → calls for accountability”) is kept in the same order. Even though the wording is different, the syntax and flow are too close. That makes it borderline patchwriting, not an ideal paraphrase for teaching avoidance of plagiarism. Here’s a better paraphrase, with a genuinely different structure: Cheong (2024) notes that when AI systems operate without transparency, their hidden decision-making can lead to biased or harmful outcomes. Because of these risks, he argues that society needs governance models that assign responsibility and protect human rights."
November 3, 2025 at 2:55 AM
janerosenzweig.bsky.social
It gave me a paraphrase of the passage that it said was good. I said, "is that actually a good paraphrase" and it told me why it is good, apparently because the "flow" is different. When I pointed out that in fact this is pretty close to the original, ChatGPT agreed with me, as it does. /2
text that says: ✅ Good Paraphrase Example (for students to aim for) Cheong (2024) argues that as artificial-intelligence systems are rapidly adopted across sectors such as healthcare, education, employment and criminal justice, important concerns about individual and collective wellbeing emerge. He notes that when AI begins to make more autonomous choices and its processes are not transparent, it becomes difficult for affected individuals to see how decisions are made. Because these systems may also embed bias, inflict unintended harm and violate rights, the author says that stronger mechanisms of accountability must be built into AI governance. (Cheong, 2024) Compared to the original, the flow is different: Original: AI adoption → opacity → bias → accountability Paraphrase: AI adoption → opacity → risks → accountability Same ideas, but reordered and condensed.
November 3, 2025 at 2:55 AM
janerosenzweig.bsky.social
I'm talking about paraphrase in class this week and was curious to know how ChatGPT would fare creating a lesson similar to the one I'm actually using (that I made without AI). I asked for a passage from a real article and then a good paraphrase. It did give me a passage. But the rest? Not great. /1
Image of text that says: Source: Transparency and Accountability in AI Systems: Safeguarding Well‑being in the Age of Algorithmic Decision‑Making by Ben Chester Cheong, Frontiers in Human Dynamics, 2024. Frontiers Passage (excerpt): “The rapid integration of artificial intelligence (AI) systems into various domains, such as healthcare … education … employment … and criminal justice, have raised pressing questions about their impact on individual and societal wellbeing. As AI algorithms become increasingly sophisticated and autonomous, their decision-making processes can become opaque, making it difficult for individuals to understand how these systems are shaping their lives. This lack of transparency, coupled with the potential for AI systems to perpetuate biases, cause unintended harm, and infringe upon human rights, has led to calls for greater accountability in AI governance.” Frontiers
November 3, 2025 at 2:55 AM
janerosenzweig.bsky.social
Reading a summary is not the same as reading an argument. All the steps that come before writing--coming up with ideas, figuring out how they fit together, drafting--that's the important work, not the stuff we outsource to make time for the important work. www.nytimes.com/2025/10/29/o...
Many people invoke a distinction between illicit uses of A.I. (such as the composition of entire drafts) and innocent auxiliary functions — outlining, for instance. But it is these seemingly benign functions that are the most pernicious for developing minds. Take the summary: Letting A.I. take over this rote task seems like a harmless shortcut. Sure, students who read only A.I. summaries will be subjected to predictable analysis and homogeneous prose, but they could save time and energy. In truth, the ability to determine what is being argued for and how is not dispensable. No aspect of cognitive understanding is perfunctory.
October 29, 2025 at 2:02 PM
janerosenzweig.bsky.social
Well that clears things up.
Text that says "AI Overview Perplexity does not run ads on Facebook, but Facebook is a platofrm where Perplexity runs its own ads to promote its AI search engine. Perplexity is also"
October 28, 2025 at 11:40 PM
janerosenzweig.bsky.social
Not quite how writing or "writer's block" works I think. www.theguardian.com/business/202...
Authors will come to rely on artificial intelligence to help them beat writer’s block, the boss of the book publisher Bloomsbury has said. Nigel Newton, the founder and chief executive of the publisher behind the Harry Potter series, said the technology could support almost all creative arts, although it would not fully replace prominent writers. “I think AI will probably help creativity, because it will enable the 8 billion people on the planet to get started on some creative area where they might have hesitated to take the first step,” he told the PA news agency. Michael Connelly ‘Every kind of creative discipline is in danger’: Lincoln Lawyer author on the dangers of AI Read more “AI gets them going and writes the first paragraph, or first chapter, and gets them back in the zone,” he said. “And it can do similar things with painting and music composition and with almost all of the creative arts.”
October 28, 2025 at 11:56 AM