Jeremy Caplan

Google just launched Study Notebooks, a mode which "turns Gemini into an interactive, adaptive learning platform". Here's the TL;DR: 👉 What it is: A "learning mode" within Gemini that turns your course notes & content into a personalised, quiz-driven curriculum. The product takes the learning content you give it and: → immediately quizzes you, to find what you do & don't know → builds personalised lessons to close your gaps → tracks what you've learned & advises what to study next 👉 My hot take: ✅ Study Notebooks is better than most existing AI learning products, because most of its "friction" has some sort of pedagogical value: → Tailoring instruction to the individual learner is one of the oldest reliable findings we have — Bloom's classic work found it can move an average student's achievement dramatically (Bloom, 1984). → Managing cognitive load — giving learners only what they need — is proven to improve comprehension (Sweller, 1988). → Recall practice, aka the testing effect, is proven to help learners retain information for longer (Roediger & Karpicke, 2006). → Checking understanding and feeding it back into what comes next — formative assessment — is one of the best-evidenced ways to raise achievement (Black & Wiliam, 1998). But… ⛔️ The tool also has some real shortcomings and in many ways doesn't feel innovative at all: → It quizzes before it helps — precisely the same friction that we know makes people abandon LMSs, AI tutors & other AI learn modes. → It buries learners in text. For a tool sold on managing cognitive load it generates a lot to wade through — long lessons, dense explanations, a dashboard of 100+ objectives to track. More reading isn't more learning; past a point, the volume itself becomes the extraneous load Sweller warns against (Sweller, 1988). → It has no mechanism to build understanding. Elaboration, worked examples, analogies, transfer — the generative strategies learners reach for first, and the ones that deliver understanding — all sit outside Study Notebook's content + quiz loop. → It rewards turning objectives green. Once progress is a wall of green ticks, the goal shifts from understanding the concept to clearing the metric — and learners optimise for the measure until it stops tracking the learning (Goodhart's law, in learning form). 👉 TL;DR: → Study Notebooks is the best-built version of a mistake we keep making: optimising learning products for what's easy to measure & sell instead of what learners actually want & need. → It's a step forward — but a very small one in the same direction we've been walking for fifty years, based on the same small sample of research we've focused on for decades. What's your take? Check out Google's launch video & comms using the link in comments. Phil 👋

☕️ Sunday Sharing: OpenAI just launched ChatGPT Work — a "multi-agent workspace" you delegate to rather than chat with. What's the shift, and what does it mean for L&D? On Thursday OpenAI launched ChatGPT Work, a “multi-agent workspace” to help people get work done. The key difference between a “multi-agent workspace” and basic LLM chat is *autonomy & control*: chat responds when you ask a question; a workspace takes a task you give it and completes it while you do something else. Here’s the basic workflow: 1. You give the workspace a goal — e.g. “run quantitative analysis on these survey responses and separate what learners liked from evidence they actually learned anything.” 2. A team of agents work across your docs & apps to gather the data and do the work until the goal is achieved (or, get as far as they can within a set schedule). 〰️〰️〰️〰️〰️ These workspaces are increasingly common. ChatGPT Work is the latest addition to a growing number of multi-agent workspaces built by every major AI vendor alongside chat: → ChatGPT Work (OpenAI) → Claude Cowork (Anthropic) → Copilot agents + Agent 365 (Microsoft) → Gemini Enterprise Agent Platform (Google) 〰️〰️〰️〰️〰️ 🚀 Multi-agent workspaces offer some unique benefits for L&D: → Evidence-gathering without the gathering. The workspace pulls from where your data already lives instead of waiting for you to find, export and paste it. Think: needs analysis grounded in real signals, not recollection. → Evaluation as a standing process, not an annual event. Set up a work flow once once — e.g. “every Friday, flag any course where completion drops 20% week-on-week” — and the signals come to you. → Living deliverables. Dashboards and resource portals that update themselves as new data lands. 〰️〰️〰️〰️〰️ ⚠️ But agentic workspaces also come with risks — because delegation isn’t free: → You lose control of the how. You set the goal; the agents choose the method. If you’d have excluded that survey question or weighted that data differently, you either find out at the end — or you don’t. → The work doesn’t disappear - it moves. The hours saved doing the work largely return as hours verifying it (or we don’t check and compromise quality). → A badly framed goal gets executed, not questioned. These platforms don’t push back on the brief; point it at the wrong problem and it will solve it without question. 〰️〰️〰️〰️〰️ TL;DR: like most AI tools, multi-agent workspaces trade effort for oversight. You’ll produce less and direct, verify and decide more. That could be a genuinely good trade — but only for people who know what good looks like. If this sounds familiar, it should: it’s the same trade off we face when delegating to a talented new hire. You (eventually) gain hours back but also give up some control over the how; your job becomes briefing well, checking the early work, and staying accountable for what ultimately gets produced. Happy experimenting, Phil 👋

Hey all. Angelica Quicksey and I are pulling together a "reading list" of great articles that illuminate how AI will shape the information environment. The first draft is here (with some brilliant articles from, e.g., Gina Chua, Rasmus Kleis Nielsen, Renee DiResta, Tony Haile, Shuwei Fang, Shirish Kulkarni, David Caswell and many more), but I'm sure we're missing some really good ones. What has helped clarify and illuminate this topic for you? Feel free to either add directly in the doc or suggest articles (or videos, or podcasts! hey, we're in a post-format world!) in the comments below. https://lnkd.in/gZibBFmB

I once came across a PDF table of potential funders for journalism organizations. There were 40 names in it, and it said "please don't share this around", as if names and websites of organizations were private information. For the past 6 months I've been assembling a list of organizations from all over the world that might have something to do with journalism, and I've been adding a few almost every week. This list is open and free and can be found at Journalism Relay Project, a knowledge-sharing project I created during my time at John S. Knight Journalism Fellowships at Stanford. There are other good lists around, like the Journalism Support Exchange, but mine goes beyond the US. If you have tips on organizations to include, please leave a comment with their name or LinkedIn page. Link in the comments.

Your day, measured in inches: Wake up staring at a 6.7" screen. Work on a 16" screen. Relax with a 55" screen. Fall asleep staring at a 6.7" screen. The average person now checks their phone 96 times a day. Once every 10 minutes. Here's how to take your day back, according to the research. First, drop the self-blame. A 2020 USC study found the biggest predictor of lasting behavior change wasn't motivation or willpower. It was repetition and environment. Thousands of engineers spent years making that screen hard to put down. So redesign your environment instead. 1. Name the feeling. Reaching for your phone is rarely about the phone. It's emotional first aid for boredom, stress, or loneliness. Before you unlock it, ask: "What am I avoiding right now?" Research on affect labeling shows that naming an emotion weakens its grip. 2. Make your usage impossible to miss. Harvard's Teresa Amabile found that what fuels daily motivation is visible progress. The reverse is also true: invisible costs undermine your drive. Put a Screen Time widget on your home screen. See the number every time you unlock. 3. Convert your phone from a casino to a toolbox. Delete the apps built for endless scrolling. Keep the ones built for finishing a task: maps, messages, banking, calendar, notes. A tool helps you complete something. A casino keeps you playing. 4. Add 20 seconds of friction. Charge it in another room. Use a timed lockbox during deep work. Twenty seconds of inconvenience beats two hours of resolve. Every time. 5. Replace the break. In my book "When," I found the most restorative breaks share four common elements: moving beats sitting, outside beats inside, with someone beats alone, and fully detached beats half-checking. A walk with a friend is the anti-scroll. 6. Recruit a witness. Pick one person. Share your weekly screen time every Sunday. Discuss what triggered you and what you'll adjust. Behavior watched is behavior changed. One last thing. For my book "The Power of Regret," I collected more than 26,000 regrets from people in 134 countries. Not one person said, "I regret not spending enough time on my phone." Your attention is the raw material of your life. Spend it like it's scarce. Because it is.

💥 Muy emocionada por esta conversación en unas semanas como parte de la conferencia de NAHJ (National Association of Hispanic Journalists) y de compartir todo lo que hemos logrado y seguimos avanzando en el programa de Periodismo Bilingüe de la Craig Newmark Graduate School of Journalism at CUNY. Siempre honrada de compartir espacio y discutir nuestra apuesta por el periodismo en español, riguroso y de excelencia que pone a las comunidades latinas en el centro de las historias. 🚀 Estoy especialmente ilusionada por este panel que coincide con los 10 años de nuestro amado programa de Periodismo de Periodismo Bilingüe en Cuny-Newmark. La cita es el miércoles 22 de julio, tempranito a las 9:45am. ¡Les espero saludar en New Orleans! Más aquí ➡️ https://lnkd.in/gzHN6WQF

Some more news! I’m very happy to share that I will be joining WAN-IFRA as Director, AI in Media, starting 1 October. After three meaningful years as Director of Strategy & Innovation at NPO, this feels like a natural next step: still very close to journalism, media and public value, but now with a broader international focus at a moment when our industry is being reshaped at remarkable speed. AI is no longer a distant strategic question for publishers. It is already changing how news organisations work, how audiences find and experience information, how value is created, and how the relationship between journalism, platforms and technology companies is being redefined. That makes this both an exciting and a difficult transition. And I believe the industry needs more than hype, fear or surface-level answers. We need to understand what is really moving underneath: changing audience needs, new forms of intermediation, shifting business models, newsroom capabilities, trust, public value and the role journalism wants to play in an increasingly AI-driven information ecosystem. That is exactly the kind of work I look forward to doing with WAN-IFRA: helping publishers learn from each other, build practical capabilities, make better strategic choices and navigate this next phase with confidence, responsibility and purpose. I’m grateful to Stig Kirk Ørskov, Thomas Jacob and the WAN-IFRA team for their trust. I’m also excited to build on the work already happening across the AI in Media community, Newsroom AI Catalyst, AI Futures Lab, executive programmes, research initiatives and global forums. Alongside this new role, I will also create more deliberate space for advisory work and selected assignments through WAYFINDER. Until now, that work often had to happen in the margins. From October, I want to give it more focus: working with media organisations, leadership teams and industry partners that are trying to make sense of this transition and translate it into strategy, products, capabilities and decisions. As always, my starting point will not be blind optimism or reflexive resistance. It will be curiosity, craft, listening carefully, asking better questions, and looking beyond the noise to the deeper movements shaping media, journalism and audiences. A new chapter, but very much the same mission: helping journalism and media organisations navigate change with more clarity, nuance and purpose. https://wan-ifra.org/news/

You might know Martin Schori as the joyful AI guy from Sweden’s Aftonbladet, pushing ideas for how AI could be used in the newsroom. And yes, that includes AI rapping the news to younger audiences. Since then, he’s become co-founder of a newsletter startup called Hint, and written a practical book about his AI journey. The Swedish library service calls it “som helhet en oumbärlig bok och en ledsagare i AI-eran,” which sounds really nice. Thankfully, Martin has translated his book into English. → Martin, who should read your book, "AI in the Newsroom" and why is it not a custom GPT but words on pages? If you're a journalist curious about AI, or an editor or senior manager who has to lead the transformation, I hope you'll find this book useful. If you've been coding since you were 13 and already used Claude Code on day one…maybe not. (Unless you came for the bad jokes.) Why not a GPT? I guess because a book forces me to actually hold a position, not generate a new answer every time someone phrases the question differently. → What's one surprising thing AI does better than us and one thing it should never do for us? Better: persistence without ego. AI doesn't mind doing version 47 of a summary. We humans sometimes cut corners because we're proud, or just tired. Never: ask for forgiveness. We will make mistakes, with or without AI's help. But only you can take responsibility and apologize. We need that, always. → How did you make sure the book survives the next months or the next year? I tried not to write a book about tools. What I wrote about instead is the structures and principles underneath — who owns the decisions, how you build trust, what happens to a newsroom culture once persistence becomes cheap. Those are the same questions in 2029 as in 2026. I hope! → "AI for research, never for a single written word", where do you stand? It depends on what you're trying to achieve and what user need you're trying to fulfill. Updating someone on the interest rate? AI can do that faster, and maybe more concisely. But I believe in more humans in journalism, because that's where we make a difference, connect with the audience, and build communities. Which I do believe is the future.

You work hard for your money. You negotiate your salary. You compare interest rates. And then you spend $1,200 on something that makes you happy for about 11 days. The real problem: we aren't bad at earning money. We're bad at spending it. Over a lifetime, that mistake costs you more happiness than a lower salary ever will. The research is clear and a little uncomfortable. What determines your happiness isn't how much you spend. It's how you spend it. Social scientists Elizabeth Dunn and Michael Norton call it "happy money." Here are 5 science-backed questions to ask before any purchase. One of them is deeply counterintuitive. Question 1: Is this buying me time or stealing it? Someone proudly purchases a ginormous house in the suburbs to upgrade their life. What they really upgrade is their commute. 90 minutes a day. Harvard researchers tracked thousands of people across countries: those who spent money to save time reported higher life satisfaction than those who bought material goods. Question 2: Is this a story or just a thing? Across dozens of studies, experiences beat possessions. Three reasons: 1. Experiences become part of your identity. ("I'm someone who hiked the Grand Canyon.") 2. They're harder to compare. No one ever says, "Your memory is nicer than mine." 3. They improve with age. Things depreciate. Stories appreciate. Question 3: Does this bring me closer to other people? We believe happiness is individual. Turns out it's relational. Spending on others makes us happier than spending on ourselves as shown with toddlers, with people in Uganda, with office workers in Canada. Question 4: Can I make this a treat instead of a baseline? The first time you drive the fancy car: incredible. The 10th time: normal. The 20th: expected. Researchers call it hedonic adaptation. The same pleasure bought too often becomes the new normal. So put guardrails around it. Fancy coffee, Fridays only. One luxury hotel per year. Question 5 (the counterintuitive one): Can I pay now and enjoy later? Pay for the vacation months in advance and you get something powerful: anticipation. Anticipation is a form of happiness. It turns one moment into many. And when the experience arrives, you're not thinking about the cost. Don't just buy experiences. Buy anticipation. 1. Does this buy me time or steal it? 2. Is this a story or just a thing? 3. Does this bring me closer to people? 4. Can I make it a treat rather than a standard? 5. Can I pay now and enjoy later? Money doesn't buy happiness. But smarter spending absolutely can.