News from the Woods

by Filip Molcan
  • 0.0 (0)
  • Technology
  • Updated Biweekly

Exclusively upbeat news from the world’s forests, hills and meadows, with a sprinkling of digital minimalism, AI & startups. newsfromthewoods.substack.com

  1. Sep 1

    Where should you send your kid to school in the age of AI? 👩‍🏫

    Imagine you’re fourteen and you have to pick a school. It’s 1995. Your parents tell you: learn to use a computer, that’s the future. Or go study law. And they’re right. Now imagine you’re fourteen today. Could you give a kid good advice? My son is 15. In a few days he starts secondary school - for now at a gymnázium, the Czech academic track, so he still has a little time before he has to decide. In these hard-to-read times I see that as a big advantage. How do I advise him? I decided to write it down and try writing my son a letter. I’d have appreciated one at his age. Or maybe I’d have laughed at it and shoved it in a drawer, because at fifteen you don’t think of your dad as the smartest person you know :-). Either way, I’d have regretted not at least trying. So here’s my letter - maybe it’ll give you something to think about if you’re in a similar situation. And below the letter you’ll find how I got there. If you want to go deeper, the option is there! Dear Son, I know you hate it when I give you advice. So this isn’t advice. It’s more what I’d want to hear from my own dad if I were fifteen today. Stash this scrap of paper somewhere and try to look at it now and then. First, a confession. When I was choosing a school, the world was reasonably predictable. You graduate, you find a job, you do it, eventually you get promoted. Or you go into business, but it played out much the same way. Studying was something you got out of the way at the start of your life, and then you worked for forty years. That model died during my lifetime, and during yours it won’t exist at all. You’ll be learning your whole life. Not because it sounds nice, but because the world will change faster than any school can rewrite its curriculum. So if there’s one single thing you should learn in secondary school, let it be this: learn how to learn. Work out how your head works, when it thinks best, how to get into something you don’t understand. That will stay with you even if everything else goes obsolete. And I think things are going to go obsolete damn fast. Second thing. When I was young, “IT people” were a strange caste who understood computers, and the rest of the world came to watch them. Being average, or even below average, was enough not to starve. That’s ending. Anyone who wants to be successful will have to become an IT person - but an average IT person will be worthless. Not that everyone will program, but everyone will build their own tools, their own little apps, their own ways of working with AI. Whoever can do that will have a multiplied advantage in any field. That’s why I’m not telling you “go study computer science”. I’m telling you: wherever you go, take this with you. AI is the new basic skill, the way reading was for me and English was for my generation. But on its own it won’t be enough. Combine it with something. With a field, a craft, a subject that’s yours. The third thing is uncomfortable, so I’ll say it straight. You’re smart. You know it and I know it. And that’s exactly why I have to warn you about the trap smart people fall into. Everything comes fast to a smart person, and AI makes it faster still. It’ll spit out your homework, your essay, your code, whatever, and nobody can prove a thing (if you’re clever about it). It’ll be terribly tempting to use your intelligence to avoid doing any work at all. Use AI to make yourself smarter, not to make your life quiet. Have it explain things to you, argue with it, test whether it can catch you out. You’ll know which one you’re doing by whether you still know things when the wifi goes down. Smarts without work are just laziness in nice packaging, and the world spots it faster than you think. The fourth thing may surprise you, because it comes from someone who has worked in technology his whole life. Don’t throw out the humanities and the arts. The world is being flooded with text, images and music made by machines. There will be an infinite supply of it and it will be free. Which makes the person who can say what of it actually matters that much rarer. Someone who understands people, who can build an argument, who can ask a good question. Someone who can create something genuinely unique. History, literature, art, philosophy - those aren’t useless subjects. They’re training for exactly the muscles machines don’t have. Fifth. You can see at home that it’s possible to live freely and run your own thing, so you know it isn’t a YouTube fairy tale. But let me tell you what’s really behind that freedom, because from the outside it looks easier than it is. Freedom isn’t the opposite of work. Freedom means you can choose what you work on, who you work with and where you work from - and that choice is bought with one single currency: being properly good at something and having people trust you. That’s why your brand will matter - what can be found about you, what you’ve built, what people say about you. You don’t have to be an influencer. It’s enough that things are left behind you and that you’re not afraid to put your name on them. And now the thing you may want to ask about: what I’d wish for. I’d avoid the directions where the ground is shifting most right now: rank-and-file coder, lawyer, marketing, admin. You’d make me happy with medicine, caring for people, a proper trade, or deep knowledge of IT and security - things that rest on your hands, on responsibility and on trust. Those are my bets. But feel free to cross out that whole paragraph. Because the main thing is something else, and I know it sounds worn out: do something you enjoy. I’m not saying it because it sounds nice. I’m saying it for a purely pragmatic reason. When you enjoy something, you stick with it even when it gets hard. And if you stick with it, you’ll be above average. And that’s what the whole thing is about. The world you’re stepping into will be merciless to mediocrity - an average text, average code and an average idea can be produced by a machine in a second, for free. Mediocrity used to be the safe choice. It isn’t anymore. The future has no time for the average. It has plenty for the driven. An above-average carpenter will outlast an average programr. An above-average teacher will outlast an average lawyer. So when you’re filling in your application, don’t ask which field will survive. Nobody knows that. Ask where you’ll learn to think properly and where you’ll be doing real things. And if something really grabs you, that’s the right path. And what if you get it wrong? Doesn’t matter. In your world, choosing a school isn’t a verdict anymore. It’s the first move, not the whole game. Dad This Special isn’t a list of safe fields. No such list exists (actually the opposite is true - there are plenty of them, but…) and anyone selling you one is guessing. It’s a map of what we actually know today, where the smartest people on the planet disagree, and what follows from that for parents standing in front of an application form. What’s actually happening The International Labour Organization analyzed almost thirty thousand work tasks and came to a cautious conclusion: most professions touched by AI will be transformed rather than eliminated. So far AI is good at tasks, not at whole occupations. It can write the first draft of a contract, but it can’t carry the responsibility when the contract turns out to be wrong. But then there’s a rather different signal. Economists at Stanford (among them Erik Brynjolfsson, one of the biggest names in the economics of technology) analyzed wage data covering millions of American employees. They found that young people aged 22 to 25 in occupations heavily exposed to AI have employment roughly 19 percent lower than you’d expect from their peers in less exposed fields. For people over thirty there is no such gap. For experienced workers in those very same fields, employment is actually growing. And the mechanism doesn’t look like layoffs. It looks like weaker hiring. Companies aren’t firing juniors, they’re simply not taking them on. AI does the first analysis. AI does the first research. AI writes the first version of the code. And here’s the question every parent should be asking. Not “will that profession exist in 2040”. But: will my child get the first working years in which they learn the craft? Because you don’t become a senior at school. You become one by doing junior work for several years, making mistakes, and having someone more experienced correct them. When the bottom rung of the ladder disappears, nobody gets to the top. Enjoying News from the Woods? Buy me a coffee💚. 🎧 News from the Woods as a podcast on Spotify or Apple End of the commercial break, let’s carry on…👇 Four camps: where the people building AI say to send their kids If there were a clear answer, the people running the AI companies would agree on it. They don’t. And that in itself is the most important piece of information in this Special. Camp one: go where AI physically can’t. Geoffrey Hinton, Nobel laureate and “godfather of AI”, says it without dressing it up: if you want certainty, become a plumber. In his view routine intellectual work is more vulnerable than skilled manual work in an unpredictable environment. It sounds like a punchline, but there’s logic to it: AI can write an essay, but it won’t replace the boiler in a 1978 prefab apartment block where nothing is up to code. Camp two: the biggest casualties will be office juniors. Dario Amodei, head of Anthropic (the company building the AI model Claude), warned last year that within a few years AI could wipe out a large share of entry-level office positions and push unemployment to 10 or 20 percent. It’s a prediction, not a fact. But notice: it fits the Stanford signal exactly. Camp three: stick to math and fundamentals. Demis Hassabi

  2. Aug 29

    Everything I let AI do - and what I keep offline 🧠

    “Filip, so how do you actually use AI?” I get this question so often that I started keeping a list. It sat in my notes for a long time, and the summer holidays were the perfect excuse to turn it into a special. Here it is. The starting point I’ve always been the kind of person who has to be doing X things at once. Sometimes it’s an advantage, sometimes it isn’t, but that would be a different special. My main job is growing Good Sailors, on top of that I run innovation at M2C, I help out at the Innovation Centre of the Ústí Region, I write Zprávy z lesa and 11 Houses, I sit on the municipal council in Doubice, I’m a member of the Bohemian Switzerland National Park Council and a volunteer ranger. Plus mentoring at a couple of startup accelerators, and I’m always happy to sit on the jury of Křišťálová Lupa (the Czech internet awards), Nápad roku (Idea of the Year) or Firma roku (Company of the Year). Sometimes I give talks about entrepreneurship or about life in Greece - schools are probably my favourite audience. Seeing it all in one place does rather raise the question of whether it isn’t a bit much :-). But when you enjoy all of it… (note from my proofreader, also known as my wife: yes, it is too much!) Today I genuinely can’t imagine doing all of this without using AI every single day. I’d either have to cut things out, or I’d already be as burnt out as the forest around Pravčická Brána (and after the 2022 wildfire here, that isn’t only a metaphor). Where AI helps me and where it doesn’t Pretty much everywhere - it sorts and answers email, it helps with strategy, with projects, with client proposals. The list of things where none of the tools gets involved anywhere in the process is very short. But it isn’t empty. AI doesn’t help me choose the content for Zprávy z lesa, and it doesn’t help me choose the properties for 11 Houses. I’m not saying I haven’t tried. It just isn’t it. Even the presentations I put together I still mostly build by hand in Keynote, because the world is already drowning in decks made by Claude that the person presenting doesn’t actually understand. It doesn’t help me with personal direction, with planning the important tasks and life priorities. And I think that’s going to matter more and more. AI will help you with the running of things, but you have to be the one holding the rudder. I’ll come back to that at the end. The tools Let me just list them so I don’t ramble on about the details. You want a list of tools and a list of use cases anyway… * Claude - most of how I operate revolves around it. I have a pile of projects in there and I’ve connected it to everything I possibly can. We have the company version, so we share projects with colleagues. Some of them run tenders in it, some of them code in it. * Perplexity - I use it because I get it free with my Revolut account. Unfortunately Revolut is now switching from Perplexity to ChatGPT, so my free ride is nearly over. I use it for everyday information lookup, basically instead of Google. Perplexity Computer isn’t a bad solution either, worth a try. * ChatGPT - I’ve used it very little over the past year, mainly on the road thanks to their excellent voice interface, though lately it feels like it’s on the way back up. The voice mode really is superb. It can play couples’ therapist (no idea how well) or a game of word chain (not well at all). A great companion for topping up my brain on the drive to Prague - when I get bored of music and podcasts, I go and have a chat. * Gemini - I have it as part of Google Workspace. I mostly use it for generating images and occasionally for transforming data, which it’s reasonably good at, but otherwise it’s miserable compared to Claude. * Copilot - we have it at work, obviously. Once you’ve used Claude or Grokbot, Copilot feels like Siri or Clippy. It’s trying, but it’s practically useless. So I fire it up every now and then to confirm that it still “can’t do anything”. Anyone who claims otherwise hasn’t tried the alternatives, sorry. * ElevenLabs - this is my voice. If you listen to Zprávy z lesa as a podcast, that voice isn’t mine, it’s ElevenLabs. * Plaud - I have their pendant and use it together with the app to record calls and meetings. Lately I’ve stopped working with the meeting summaries and instead save the full transcripts, which I then work with in Claude. I’m gradually moving over to local Whisper, partly for security reasons. * Lovable - my beloved tool (and European, at that!). I built my own website in it (design, copy, dark mode, translations into English and Greek…), some of our company sites, client prototypes, internal company apps, and 11 Risks came out of it in a single morning. I like it because whatever I build there I can point straight at a domain and it just runs. No hosting to sort out, no other faff. * Grokbot - I’ve never used Grok, but Grokbot has been running for me for the last few days and I’m delighted. It’s Elon’s, so it’s a bit of a loose cannon, but the sheer amount it gets done is remarkable. It reaches out to partners for me, comments on content, and so far I’m happy with the results, so we’ll see. It isn’t cheap, but I have to say it made me go “wow” for the first time in a long while. * A2E - for when I need to do something in graphics that the other models are too strict about. And I’m curious how long this will last, because how easily you can create harmful content there is frankly unbelievable. I’m not encouraging anything, but maybe try it, just so you know how easy it is to fabricate a photo of your wife with a lover. When somebody eventually sends you one, you might be a bit more careful about what you believe. * Spark - an email client that lets me work with AI across all my accounts. It can answer email, sort it, or hand everything over to Claude. * Second brain - a recurring topic. I stick to Apple’s native tools. Most of my content lives in Notes, and my “second brain” is an iCloud folder where I collect the most important material for each project and share it across tools. It’s a set of .md files with a basic project description, research results, downloaded data and other background worth having at hand as context for projects in different tools. For tasks I use Apple Reminders - though I’ll admit I’m still looking around here, and wondering whether to build something custom. Thanks to vibecoding, that’s finally realistic. Enjoying News from the woods? You can buy me a coffee. 🎧 News from the woods is also a podcast on Spotify or Apple🎧 How I actually use it Morning podcast Every morning Claude, working with the other tools, prepares a morning podcast for me. I play it on the way to work or while brushing my teeth. It’s a summary of what matters to me: local news, news from the tech world, and above all the distillation of my email, my calendar and Slack. What does it look like? Here’s a sample (names and projects are hidden or changed). Email Thanks to its connection to Claude and other systems, Spark handles a lot of my email for me. I have both Google and Office 365 accounts connected. It works like this: every morning I already have draft replies waiting for the emails that need dealing with. In practice the rule of thirds applies. One third of the drafts are excellent and I just check them and send. One third are a decent base that needs something added. And one third are useless, so I delete them and write a different email. But the progress over the last few years is astonishing. The day when I won’t have to write email by hand is getting close. Advisory board We’ve built the dream advisory board for Good Sailors: Steve Jobs, Elon Musk, Tomáš Baťa, Jeff Bezos, Yvon Chouinard and Jára Cimrman (the fictional Czech genius, playwright and all-round polymath - if you’re not Czech, just go with it). Cimrman was the tricky one, because the AI doesn’t know him well enough. I had to convert all his plays and lectures from audiobooks I’d bought into text and feed that into Claude. But it works brilliantly. The lads are even allowed to argue with each other during a session (looking at it now, we’ve got quite a gender balance problem there!). Finance I don’t like spreadsheets. I don’t like Excel. I don’t like dealing with money. My AI hack is to turn the spreadsheets that land in my inbox into nice clickable dashboards. I get charts, a decent little app, and a slightly greater appetite for looking at any of it. Sales Claude helps me enormously with sales. We have individual product projects in it, holding the information, the product documentation and the decks. Then all I have to say is “tomorrow we’re presenting to client XYZ, go through their annual reports and their website, and adapt the XYZ product deck so it fits”. Works a treat. Just don’t leave it entirely to the AI - verify the data and the deck before you open it in front of the client. AI is a brilliant assistant, but it needs your oversight and above all your judgement. It’s also excellent at going through all your LinkedIn connections and advising who to approach about what, and drafting the outreach on the spot. LinkedIn has no direct connection to AI tools, so you have to request an archive of your data in the settings and download it. Due diligence and market research One of the biggest savings is “research mode”. When something matters, I give the same prompt to all of them (Claude, ChatGPT, Gemini), pull the results together and then start poking holes in them. I ask questions, I say what doesn’t add up and what smells off, and after about five rounds I’m happy with the output. Basic due diligence on companies or projects, market research. Here the AI tools do in a few hours what used to take weeks. Gallup or Emiero My favourite hack. I have my colleagues’ Gallup profiles loaded i

  3. Jul 11

    The people who were right about AI just published a plan. It's wilder than their predictions 🦾

    In May I wrote about the people we didn’t believe ten years ago - and who turned out to be right. Now those same people have published a plan for surviving the arrival of superintelligence. It involves China building its datacenters in Canada, America building hers in Mongolia, and humanity handing over the keys to the machines in 2040. Two months ago I wrote here about the AI Futures Project team - the people who said strange things back in 2015, whom we laughed at, and who turned out to be terrifyingly right. I promised myself I’d keep watching them. I didn’t have to wait long. On July ninth they published a new piece called AI 2040. If you finished the last special thinking it couldn’t get any wilder... Same people, different genre A quick recap first. Daniel Kokotajlo, a former OpenAI researcher, and his team wrote the AI 2027 scenario last year - the one read by US Vice President JD Vance - which ended in one of two ways: either superintelligence wipes us out, or the future is run by a handful of people with AI under their thumb. Neither one is a win. AI 2040 is the sequel. But careful - this time it’s not a prediction, it’s a recommendation. The authors say so explicitly: this is the least bad plan we know of. Not what will happen. What should happen. That’s a fundamental shift. Last time they told us what’s coming. Now they’re telling us what to do about it. And by the way - they’ve moved the default timeline from 2027 to 2030. Without intervention, they say, fully automated AI research would be running in 2030, with superintelligence by the end of that same year. Plenty of people will disagree with that, but I’m passing it along as I found it. Daniel adds that things will probably move even faster than the scenario describes. Remember the Automated Coder from the last special? Its median is holding at mid-2028. Nothing has slowed down. Choose your path The most interesting thing about AI 2040 is the format. It’s an interactive website where the story runs year by year and stops in 2029. The world stands at a crossroads and you pick one of five paths. You literally click buttons. If you have some time on holiday, I recommend opening it on a computer or iPad (it doesn’t work well on a phone) and leafing through it like a good encyclopedia. What does the crossroads look like? Plan D - keep racing at full speed. America versus China, whoever gets there first. According to the authors, this is the default scenario, because it requires no decision at all. Just do nothing. Plan C - keep racing, but the leading project sacrifices at least a month of its lead on safety. Cosmetics. This is actually happening a little already (the US government slowing down model releases, and so on). Plan B - sabotage China, gain a lead, then use it for global dominance. It comes in two variants, the harder of which involves drone strikes on Chinese datacenters. Yes, you read that correctly. Plan S - halt development indefinitely. The authors admit they sympathize with it, but they don’t believe in it - sooner or later the race restarts anyway, and in the meantime nobody learns anything. We’ve actually been through a small version of this. Remember the cries from people like Elon Musk to “pause it all”? All it did was buy him time to invest more in xAI. Plan A - a deal. And that’s what the whole rest of the scenario is about. The authors attach an estimate to each path: the probability that it leads to a good future. Plan A got 42%. Plan B and the improved C around 25%. Plan D 10%. Even the best path doesn’t reach a coin flip. These people are not the optimists from a TV commercial. They argue that among bad options, there are less bad ones. And now the “best” part: the authors themselves estimate the chance that Plan A actually happens at three to fifteen percent. They wrote a hundred-page plan they believe will almost certainly not be adopted. Why? Because in their view, it needs to exist. So that when the crisis comes, it’s lying on the table. Plan A: slow down, open up, and booby-trap your own backyard So what does Plan A propose? Three things, each stranger than the last for us mere mortals. First: delay superintelligence until 2040. In 2029, the US and China agree that the race to superintelligence is suicide for both sides. First they declare their compute capacities to each other, then they halt the training of the most powerful models, and finally they pull the rest of the world into the deal. AI keeps developing - but in a controlled way. In 2035, development pauses at the level of a top human expert and waits for safety research to catch up with capabilities. Only in 2040 do they let go of the reins. I can’t imagine today’s world being willing to do this and agree on it. From the US through China and Russia to Iran. Second: total research transparency. All AI research gets published. Dozens of companies around the world are allowed to catch up to the frontier. No secret labs, no lead worth stealing. Vitalik Buterin noticed that the plan is paradoxically friendlier to open source than today’s reality - it even mandates it. This would of course be wonderful; I’ve been supporting open source for thirty-plus years. I’m a big optimist, but... Third - hold on tight: mutually assured compute destruction. New Chinese datacenters get built in Canada. American ones in Mongolia. Just south of the Mongolian border, American servers hum away, guarded by a small contingent of US troops - and just across the border stands a division of the Chinese army, ready to move in the moment it gets the signal. If one side breaks the deal, the other destroys its chips. The Cold War inside out: instead of missiles aimed at cities, servers built within the enemy’s reach. On purpose. It sounds insane. But it has the logic of nuclear deterrence, which has kept us - however nervously - alive for eighty years. An economy to make your head spin Now the most paradoxical part. You’d expect a plan to slow down AI to mean economic austerity. The opposite is true. Even the slowed-down world grows at a pace we can’t imagine today. In 2026, the world has roughly twenty million chips at H100 strength. In 2034, sixty billion. That little square in the bottom left corner - that’s us, today. And let me remind you: this is a chart from the slowdown plan. A three-thousand-fold increase in eight years is the version where humanity steps on the brakes. Meanwhile, the Plan A deal counts on inspectors verifying ninety-nine percent of the world’s compute - looking at that field of squares, try to imagine what that means. The scenario assumes that in 2032 the American economy grows by roughly fifty percent a year. Not two percent. Fifty. Robots build robots, AI agents work around the clock, and the state taxes it all through a system of tradable permits on compute and robots. The break between 2031 and 2035 is the most important four years of the entire scenario - and I probably won’t be the only one who finds it unrealistic, but... Human labor falls from ninety percent to thirteen. And notice the order: the solid black area disappears first - cognitive work. Lawyers, analysts, programmers, journalists. The hatched physical work outlives it by two or three years, until the robots arrive. For the entire twentieth century we expected machines to take work from hands first and heads second. It will be the other way around. The plumber will outlast the lawyer. The proceeds pay for a citizen’s dividend. In 2032, every American receives roughly forty-five thousand dollars a year. By 2035, a million. Work stops being a way to make a living and becomes a hobby. Whether that sounds like utopia or a nightmare to you probably depends on how much you love your job. My favorite, because it answers the question I get under every article about AI: who pays for all this? In 2030, the American state lives off taxing people - income tax makes up eighty-seven percent. Four years later, almost the entire pie consists of permits on compute and robots. Income tax has practically vanished, because there’s nothing left to collect it from. The state taxes machines instead of people - and uses the proceeds to pay a citizen’s dividend that is supposed to reach a million dollars per person per year by 2035. It never stops fascinating me that the most radical overhaul of a tax system in history is something the authors describe as a technical detail in an appendix. By the way - in May I wrote here that Anthropic’s revenue was running at ten billion dollars a year. That number was true in January. In April, the company announced thirty billion. My two-month-old special went stale faster than most of you managed to finish reading it. That’s not my fault or yours. That’s exactly the curve we’ve been talking about all along. The uncomfortable details the authors don’t hide What I appreciate most about AI 2040: the authors don’t sweep the problems under the rug. They put them on display. The year 2035 in the scenario contains a sentence that gives you chills: the AI systems of that era are in fact adversarial - but they are under control. They’re watched by other AIs from different lineages. It’s like employing an army of brilliant employees you know would love to stab you in the back, and counting on them to rat each other out. It only works up to the level of a human expert. Beyond that you need real alignment - and in 2035, nobody has it yet. The deal itself isn’t rock solid either. The authors estimate it collapses within ten years with a probability of around forty-eight percent. A coin flip. For that case, they’ve calculated that the mutual destruction of datacenters slows the subsequent race from a week to a year. That’s it. One extra year. And one branch of the scenario describes what happens if China tries to cheat: a covert project in the tunnels next to the Medog hydropower station, where half a

  4. Jun 28

    Special: Slow, and Above All Different. 30 Tips for Travelling Beyond the Crowds ✈️

    Thirty concrete tricks for finding the places the crowds never reach - and for breaking the habit of collecting countries like stamps. I have one ugly habit. When I arrive somewhere everyone is photographing, I get an irresistible urge to walk in exactly the opposite direction. It’s not a pose. It’s more that over the years I’ve figured out one simple thing - the best of any place almost never stands in a queue. Travel has turned into a strange discipline over the past decade. We collect places like stamps. Ten cities in seven days, each one ticked off, photographed, uploaded. And then we come home exhausted and, oddly enough, remember almost nothing. Because we were never anywhere longer than one espresso and one photo in front of the right fountain - and, honestly, we often didn’t experience anything interesting at all. This special is about the opposite. It’s a collection of concrete, usable tricks. One idea ties them together: fewer places, more experiences. And a bit of nerve to go against the current. So let’s get to it. Before you even set off 1. The twenty-percent rule. The traveller and writer Eric Weiner put it beautifully: estimate how much time you reasonably need in a place - then add twenty percent. Over the years he bumped it up to thirty or even forty, because as he says: “you can travel too quickly, you cannot travel too slowly.” Treat it as an antidote to an overstuffed itinerary. 2. Pick one country, not three. Three weeks in a single region will give you incomparably more than three weeks sliced across five countries, four flights and endless repacking. Borders are not a checklist. 3. Go on an “Instagram fast”. Try arriving somewhere without having seen it a hundred times beforehand. It’s a rare luxury these days. 4. Don’t try to be a local. Be a curious foreigner. You won’t become a local anyway, and that’s actually an advantage - a foreigner notices things the locals stopped seeing long ago. Stop pretending you belong in the city, and start asking questions. 5. Take the train, not the plane. For European connections up to roughly five hours, the train is time-competitive once you add in the whole circus around the airport. And a bonus: swapping a domestic flight for a train saves the planet around 86 % of emissions, and taking the Eurostar instead of a plane around 97 %. The scenery out the window is free. Sure, it doesn’t work everywhere - the Balkans, for instance, are a real adventure by train, but even that can be part of the experience. 6. Skip the hotel, rent an apartment. Morning trip to the market for tomatoes, your own breakfast, wine on the balcony in the evening like a local, or down at the local pub. And if you do book a room, book it from someone local - no big chains. That’s the moment you stop being a visitor and become, for a while, a local. And you usually save money too. 7. Don’t sleep somewhere new every night. Moving between hotels every other day is the fastest way to kill the slow-travel mood. Pick one base camp and explore the surroundings from there. 8. Pick a town by, say, the Cittaslow label. There’s an international network of “slow towns” - municipalities under 50,000 people that deliberately reject rushed tourism. How to find the places the crowds never reach 9. Hunt for beaches from satellite view. Before you go, switch your maps to satellite mode and look for small strips of sand between rocks with no name and no label. That’s usually a cove only the locals know about. It works surprisingly well. 10. Read reviews in the locals’ language. Google now auto-translates reviews, so you can finally read what locals - not a tourist from Ohio - think of that ramen place. This is possibly the most underrated trick of all. 11. Don’t judge a place by its stars, judge it by the visitors’ photos. Locals photograph the plates and the interior, tourists photograph the sign out front. Scroll through the photos before you trust the number above them. 12. Use “search along route”. When your navigation is taking you somewhere, you can find a café, a viewpoint or a detour right while you drive, without leaving your route. The journey itself turns into discovery. 13. Click on places with few reviews. Zoom in on the map and read the places that have only a handful of ratings - but good ones. Those are the hidden gems, just before everyone else “discovers” them. 14. Alternative routes in your navigation. One of my favourite things is to take the alternative routes the navigation offers - even the longer ones. There’s often some interesting surprise along the way. Or nothing at all. 15. Historical aerial imagery. In maps, aerial mode lets you switch to older imaging. You’ll nicely see how the landscape changed - and sometimes you’ll spot a forgotten path or a place that has vanished from the current maps. 16. Ride a winding local bus line. Find the bus with the most tangled route on the map and ride it to the end of the line. You’ll see how the city really lives, far from the historic centre polished up for tourists. 17. Google “alternative to…”. To overcrowded Plitvice, to packed Cinque Terre, to jammed Prague. Almost every famous place has a quieter twin nobody talks about quite so loudly. How to find the pub where the locals go 18. The parallel-streets rule. TV host Samantha Brown has a clear recipe for this: walk to the main street or the main square - and then head into the side streets and the parallel ones. And then further still from the centre. The locals don’t eat in those famous restaurants on the square anyway - they’re expensive and packed with tourists. 19. An empty place at lunchtime is a warning. When it’s around noon and the pub is gaping empty, something’s wrong. Brown applies the same to food stalls: look for the queues, not the empty space. 20. A laminated photo-menu in five languages = a trap. A short menu in the local language with seasonal specials = you’re on the right track. The more pictures of the food, the further you are from the kitchen. 21. Search in the country’s language. Don’t type “steak Florence”, type “bistecca alla fiorentina Firenze”. Suddenly you’ll get the places locals actually rate - not the ones optimised for tourists. 22. Study the supermarket shelves. Walk the aisles and learn the names of foods and local products. When you later see them on a menu, you’ll know what to order. Simple, and almost nobody does it. 23. Read the “about us” page. Look for “family-run” and, above all, how long they’ve been going. And signs they go against the grain - say, that they cook only with local ingredients, or keep limited opening hours. A few sentences tell you more than ten reviews. 24. Make lunch your main meal of the day. The same plates as at dinner, just at the lunch price from the set menu. Your wallet and your stomach will both thank you. 25. A waiter out front luring you in? Walk on. When a place is good, it’s too busy to have someone reeling in passers-by. 26. The bartender relay. Ask the bartender where they go after their shift. Go there, and ask again. Repeat until you end up somewhere you won’t meet a single tourist. It works almost like a detective story. 27. Skip the front desk, ask on the street. The concierge will send you somewhere tourist-safe. A barista, a shopkeeper off the main drag, or parents at the playground will give you the real tips. 28. At the market, buy something and share your plan. Mention to the vendor that you’re after a quiet picnic spot or an easy hike. You’ll get tips that never make it onto any TripAdvisor. 29. Pick the person to match the question. Want a route for kids? Ask a mum with children, not a random passer-by in a suit. The right question to the right person is half the battle. And one tip to finish 30. Stay in one place long enough that the baker - or the barista - greets you. Choose one ordinary café or bakery and go there every day. By the third visit they’ll start to recognise you, and suddenly you’re part of the place, not just a passer-by with a camera. This is all of slow travel in a single sentence - it’s not about seeing everything, it’s about really being somewhere. And that’s everything from me today. Slow travel isn’t about going less far or less often. It’s more about giving up collecting places and starting to live them. Take care and enjoy the summer - wherever you spend it. Even in your own garden. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit newsfromthewoods.substack.com

  5. Jun 21

    News from the woods #141 🐋

    Business & Technology 👨‍💻 * Meta needs data centers to develop the AI it doesn’t have. So for now it’s building them as “tent cities”. * There’s a jacket out there that makes water. One day we’ll go on a trek and won’t have to haul water with us! * ChatGPT is losing its lead fairly fast. Will the upcoming “super-app” save it? * Midjourney, known mainly for its AI images, is now trying to move into healthcare. It’s developing a full-body ultrasound scanner meant to create a 3D image of the body within a minute. Will it be another Theranos, or a real revolution? Travel 🧳 * If you’re near Los Angeles, take a look inside this abandoned hospital. * Here are the 40 best newly opened hotels in the world. * If you’re feeling the heat, take a trip to Lapland… Nature ⛰️ * Scientists have discovered (or rather re-mapped) a giant hidden network of underground fungi that stretches practically beneath the entire world and is seriously long. * Beautiful photos from Antarctica. * This one is truly unbelievable - 7 kilometers below the surface of the sea, scientists found a whale graveyard that stretches an incredible 1,200 kilometers. Hundreds of skeletons and remains up to 5 million years old. We want to go to space, yet the ocean floors still hold so many secrets… Unclassifiable 🧠 * You’re playing the final scene of Romeo and Juliet, except a cat walks onto the stage. And the tragedy turns into a comedy. * The pyramids are built to be earthquake-resistant. * You know how it is - a car pulls up, drops off a package, but who carries it those last few meters to the house? A dog! A real, living one! Now that’s what I call last-mile delivery. * He’s 100 and still dancing. Inactive seniors face a lonely old age. * Someone founded a new country run by AI. It reminds me of a great film with one older, similar idea… Tips 💡 What I'm reading: Zorba the GreekWhat I'm listening to: the sound of the seaWhat I'm watching: mountains aroundInteresting app: Firewood splitting - I’m mentally getting ready for summer. The sailing season in our village is in full swing. So if any of you happen to be sailing by, let me know! Thank you very much for your support and have a great day! This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit newsfromthewoods.substack.com

  6. Jun 7

    News from the woods #140 🐙

    Thanks to AI, the world is moving faster than ever. Anyone who stops learning is going to have a harder and harder time. Even though I try to “level up” a few hours every week, it still feels like too little. Time to kill the streaming platforms and stop watching content that gives me nothing in return. I’ve cancelled all my subscriptions except Apple TV, where at least there seems to be some basic effort to produce content with actual value. So instead of an evening series on Netflix, I’m trying lectures, podcasts, interviews… What do you think? Will I last? Anyone else want to try? Business & Technology 👨‍💻 * A nice website has appeared that tracks whether AI has actually started making money for the big players. What do you think? * NASA wants to build a base on the Moon. * Drones - beyond delivering, photographing and fighting - can now also fell trees. * OpenAI no longer wants to publish research on the darker sides of AI. Hmm. * Nvidia isn’t slowing down and is betting its future on personal computers. I think we’re returning to a time when, thanks to AI, raw computer performance starts to matter again. Travel 🧳 * What does a cheese bank look like inside? * How about a Pyrenees thru-hike this year? * Sell the house, buy a boat, live on it. Is it really as romantic as it sounds? * Heading to Sardinia? You might want to read this first… Nature ⛰️ * A new species of octopus has been discovered in the Galápagos — fits in the palm of your hand and is absolutely adorable… * Want a proper aquarium at home? * And how about a proper greenhouse? * It sounds surprising, but after more than 100 years a wolf has reappeared in Sequoia NP. Yes — even in the American wilderness, humans had managed to drive them out… * Three hours of stunning nature footage with commentary by David… Unclassifiable 🧠 * My son suggested that if I ever get bored at work, I should pick up this. * David Attenborough can finally play with Lego. * Police in California can now issue tickets to self-driving cars. The war with the machines has begun — first shot fired! :-) * Scientists have managed to freeze and thaw part of a brain. Immortality is close. Well — in mice, for now… * Ever wondered why an hour has exactly 60 minutes? Why not, say, 50? Tips 💡 What I'm reading: Zorba the GreekWhat I'm listening to: the sound of the seaWhat I'm watching: my kids growing up far too fastInteresting app: MasterClass Thank you very much for your support and have a great day! This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit newsfromthewoods.substack.com

  7. May 9

    The People Who Got AI Right – And the Rest of Us Who Didn't Believe Them

    This week I read two texts that sent me out for a walk afterwards. I do that fairly often anyway, but this time I went even though the sun was blazing outside and all my Greek neighbors were asleep, saving their energy for the evening. The first piece is from the AI Futures Project team, which works on predicting when humans will stop programming. The second is by journalist Dylan Matthews, who – eleven years on – is apologizing to people he once dismissed as cranks. I’ll summarize them here, somewhat mercilessly. Anyone already afraid of the future is probably better off deleting this email now. The conference where weirdos stood at the lectern In August 2015, the Effective Altruists organized a conference called EA Global. Among the speakers were Nick Bostrom (author of Superintelligence), Stuart Russell (a legend of computer science), Nate Soares (today the author of If Anyone Builds It Everyone Dies), and a still-fairly-reasonable Elon Musk. The topic: how artificial intelligence will sweep us all away. Among the attendees was Dylan Matthews, a journalist for Vox. He spoke there with a young engineer from Google named Chris Olah. With a philosophy PhD student named Amanda Askell. And with a programmer from PayPal named Buck Shlegeris. Dylan left the conference convinced that a promising movement, one that could be saving lives in Africa and chickens in cages, was about to destroy itself over a speculative fear of a technology that didn’t yet exist. He then wrote a rather tough article in Vox, framing that fear as proof that the movement was running away from real problems. This was August 2015. The Transformer had not yet been invented (and let me just note that one of the people involved in that breakthrough was our own Tomáš Mikolov), and OpenAI had not yet been founded. Nothing in the world resembled today’s ChatGPT. Eleven years later, Dylan writes: I should have looked more carefully. Chris Olah, in the meantime, helped lay the foundation of our present, named the entire field of mechanistic interpretability, and co-founded Anthropic. Amanda Askell works at the same company and is directly responsible for Claude’s personality. Buck Shlegeris runs Redwood Research, one of the most serious technical AI safety labs outside the walls of the big firms. Three people Dylan once treated as oddballs are today holding a piece of the global technological future in their hands. They missed the mark And here is where it starts to get fun. Fun in the dark sense of the word. Dylan in his article mentions Leopold Aschenbrenner, a former OpenAI researcher who in 2024 wrote a series of essays under the title Situational Awareness. In them he predicted that by 2026, $520 billion would flow into AI infrastructure. Everyone tapped their head and called him crazy. Real-world investment is now estimated at $650 to $700 billion. We undershot it again. The reality is wilder than the wild prediction. Similarly, Ajeya Cotra and Peter Wildeford predicted at the end of 2024 what would happen to AI in 2025. Dylan writes: they were very accurate – and where they were wrong, they were wrong in underestimating the revenues of AI companies. Meaning, they didn’t err in expecting AI to grow more slowly. They erred in failing to dare to predict how fast it would actually grow. Anthropic? Annual revenue running at the $10 billion level. Tenfold growth every year. Claude Code, their programming tool, hit $2.5 billion in annualized revenue nine months after its launch. A product that isn’t even a year old. Revision to 2028 The AI Futures Project team (Daniel Kokotajlo, Eli Lifland, Brendan Halstead) published a quarterly revision of its predictions. Daniel Kokotajlo is co-author of the famous AI 2027 scenario. Eli Lifland is a professional forecaster. These people predict the future for a living, not by crystal ball. What did they revise? They moved a moment they call the Automated Coder – AC. The point at which an AGI company would rather lay off all of its human software engineers than stop using AI for programming. Read that definition again. Slowly. It’s not about AI becoming better than programmers. It’s about companies preferring to fire all their people over giving the AI back. Faced with those two options, they choose the second. Daniel shifted the median of his estimate from late 2029 to mid-2028. Eli shifted his from early 2032 to mid-2030. Why? Because the new models are better than the team could have imagined. The doubling time of AI’s coding capabilities has shrunk from 5.5 months to four. METR (the organization measuring this) released a new methodology, and the curve climbs faster than predicted. Daniel even cut the requirement for 80% reliability from three years to one – meaning he believes it’s enough for AI to handle one-year-long tasks comfortably, and the layoffs begin. The people working with artificial intelligence inside the AI companies are telling us it will come sooner than we think. In private and in public, they’re doubling down on their predictions rather than walking them back. Which means that the people who sit closest to the technology, who know what is being readied in the labs we are only allowed to see six months later as filtered marketing material – they consider the pace that puts Daniel into 2028 to be conservative. When I connect this with what Dylan wrote: we ignored this group of people once already. In 2015 we told them to go play on their pseudo-intellectual Reddit. In the meantime, they invented and built a technology that today generates ten-billion-dollar annual revenues, and which their own creators admit they don’t fully understand. These same people are now saying something crazy again. They say AI will replace programmers within two years. That within three to four years it will match or surpass top experts in every field where the work is done with your head – lawyers, doctors, researchers, financial analysts, designers, journalists. That the world economy will no longer grow at today’s two or three percent a year, but at perhaps thirty, because machines work without stopping, without sick leave, without vacation, without notice. That somewhere in the desert there will stand factories run by artificial intelligence, where one robot builds another and a human just occasionally checks the fuses. And that the moment may soon arrive when AI begins improving itself. The point of no return, because every next step forward will be made faster than the previous one. Without us. Today this sounds like a wildly overdrawn scenario, but looking back, maybe the right move is not to dismiss such ideas but to talk about them more. What now When someone asks me: should we be afraid? I answer: no. Fear makes no sense. Fear is the worst counselor you could ever choose. What to do? 1. Listen. Not to everyone – to some. The people who said weird things in 2015 and whom we now see at Anthropic, at Redwood Research, and in labs around the world. The people who attach graphs and methodology to their predictions, not clickbait headlines. People who publish their recalculations every three months and adjust them in both directions, not only the comfortable one. 2. Don’t expect institutions to explain it to you. Dylan in his article admits something graceful – his original skepticism was based on the fact that no major institution in 2015 was dealing with AI. He inferred from that that it couldn’t be serious. He was wrong. Large institutions are often worse at predicting the future than we think. 3. Don’t give all your attention to artificial intelligence. It will take it anyway. Pick up a book, go outside, teach your kid to work with wood. Love the person who shares your kitchen. By the time those people from the previous paragraphs arrive with their predictions, it won’t matter how many productivity books you have on your hard drive. What will matter is what you’ve managed to build as a human being. What will actually have value in the future? ✌️🙏 This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit newsfromthewoods.substack.com

  8. Apr 12

    News from the Woods #136 🥾

    Hey everyone, this is Filip and welcome to another episode of News from the Forest! Episode 136, and it’s a packed one. We’ll talk about how AI is reshaping schools — from typewriters at Cornell to a school with zero teachers in Chicago. How Anthropic just overtook OpenAI in revenue. We’ll discover a brand-new island near Antarctica and find out how the war in Ukraine is devastating nature on a massive scale. And at the end, you’ll try doing absolutely nothing for two minutes. Let’s go. What do you tell your kid when you know the technology you’re building will rewrite the rules for an entire generation? The Wall Street Journal asked exactly this question to the heads of the biggest AI companies — Daniela Amodei from Anthropic, Jaime Teevan from Microsoft, Ethan Mollick from Wharton. And you know what’s fascinating about their answers? Not a single one said: learn to code. That Wall Street Journal piece really got to me. Because if I asked you — what should kids study to thrive in a world full of artificial intelligence — most of us would say: STEM, coding, data science. Makes sense, right? Except the people who are actually building AI are saying something completely different. Daniela Amodei, co-founder of Anthropic — the company behind Claude, the AI model that, full disclosure, also helps me produce this podcast — says, and I’m paraphrasing: “What won’t be replaceable is how you treat other people, how well you communicate with them, how kind you are.” This isn’t some motivational platitude. This is coming from someone whose company just surpassed OpenAI in revenue. Ethan Mollick from Wharton, who wrote the brilliant book Co-Intelligence, advises his teenagers to avoid hyper-specialization entirely. His logic is straightforward: if your job consists of repeating one specific cognitive task, AI will eventually do it faster, cheaper, and without complaining. The future, he says, belongs to people who bundle three or four distinct skills — communication, judgment, creativity, accountability. And that word — accountability — is key. AI can analyze data, write reports, propose solutions. But it can’t be held responsible. A human does that. And the ability to say “I own this” is, according to these people, the most valuable currency of the future. So here’s the paradox: the people building AI are telling their children — be as human as possible. Learn how to learn, be flexible, communicate, take responsibility. And above all — don’t be a narrow specialist, be a generalist. This theme — the tension between technology and humanity — runs like a red thread through today’s entire episode. Let’s start with how it’s playing out in schools. I’ve got three stories about education that seem to come from three different universes. And yet they’re all happening right now. Story one: Typewriters at Cornell At Cornell University, one of America’s most prestigious schools, a German language instructor named Grit Matthias Phelps does something once a semester that completely blows her students’ minds: instead of laptops, they find manual typewriters on their desks. No screens, no online dictionaries, no spellcheck, no Delete key. She started doing this in 2023 because she noticed students were submitting grammatically perfect German essays — thanks to AI and online translation tools. As she puts it: “What’s the point of me reading it if it’s already correct anyway, and you didn’t write it yourself?” And the students? Catherine Mong, a 19-year-old freshman, said: “I was so confused. I’d seen typewriters in movies, but they don’t tell you how a typewriter works.” One student was puzzled by the key labeled “Return” — and then realized you physically have to return the carriage to the beginning of the line. “Oh, that’s why it’s called Return!” But the most interesting observation came from computer science major Ratchaphon Lertdamrongwong. He said: “The difference with typing on a typewriter is not just how you interact with the typewriter, but how you interact with the world around you.” Without screens, no notifications. Without Google, he had to ask classmates for help. Suddenly, they were actually talking to each other. As he put it: “That was probably normal back then. But it’s drastically different from how we interact in the classroom in modern times.” This analog wave is spreading. It’s part of a broader national trend — back to handwritten tests, oral exams, pen-and-paper assignments. Because schools are looking for ways to verify that students are actually thinking. Story two: Sweden goes back to books Now one at the national level. Sweden — a country that was a pioneer of digital education for twenty years. Every student had a tablet or laptop, textbooks were replaced with digital content. And now? A complete 180. The Swedish government is investing over 100 million euros to bring printed textbooks back into classrooms. Starting in 2026, mobile phones will be banned in compulsory schools for the entire school day. For preschool children under two, only analog learning tools may be used. Why? Because student outcomes — reading comprehension, ability to focus, deep understanding of text — were declining. Researcher Linda Fälth from Linnaeus University summarizes it: Sweden positioned itself as a frontrunner in digital education, but over time concerns emerged about screen time, distraction, reduced deep reading, and the erosion of foundational skills such as sustained attention and handwriting. Sweden’s education minister called it “an experiment that wasn’t scientifically based.” And UNESCO’s 2023 Global Education Monitoring Report backs this up, warning against uncritical adoption of technology in classrooms. Story three: A school with no teachers in Chicago And then there’s Alpha Schools. A private school opening in Chicago this fall that goes in exactly the opposite direction. No teachers. At all. Core academics — math, reading, science — are delivered in a two-hour daily block entirely through AI software. Kids from kindergarten through eighth grade sit at computers while the AI adapts to them in real time. Instead of teachers, they have “guides” — adults who motivate, provide emotional support, and lead afternoon workshops on robotics, entrepreneurship, public speaking, even running their own food truck. Guides don’t need teaching degrees, just a bachelor’s. Starting salary: $100,000 a year. Tuition? $55,000 per child per year. Founder MacKenzie Price says AI will “unlock the greatest untapped resource in our world, which is human potential.” Alpha claims their students grow 2.6 times faster than the national average and rank in the top one percent on standardized tests. But experts are skeptical. A 2026 Stanford review of over 800 academic papers found that while AI can improve student performance, the benefits become less clear when students are later asked to work without AI support. And philosophy professor Joe Vukov from Loyola University put it bluntly: “I worry that you’re changing the nature of what learning and education, at its best, has always looked like.” So three stories: typewriters as a cure for AI cheating, an entire country returning to books, and a private school that eliminated teachers entirely. All happening now. All responding to the same question — what role should technology play in education? And what fascinates me is how perfectly this mirrors what those AI executives tell their own kids. Build human skills. Accountability, communication, adaptability. Exactly the things you learn better from a typewriter or a book than from a chatbot. Now from a completely different angle — business and technology. Because something happened this week that would have been unthinkable six months ago. Anthropic — the company behind the AI assistant Claude — announced that its annualized revenue has topped $30 billion. At the end of 2025, it was $9 billion. In four months, it tripled. And with that, Anthropic has overtaken OpenAI — which sits at roughly $24 to $25 billion — for the first time in history. How? The key is the customer base. While OpenAI earns heavily from consumer-facing ChatGPT — 900 million weekly active users — Anthropic bet on enterprise. Eighty percent of its revenue comes from business customers. Over a thousand companies now pay more than $1 million annually for Claude services. That number doubled in under two months. A massive driver is Claude Code — the agentic coding tool that alone generates over $2.5 billion in annual revenue. It’s become what analysts are calling generative AI’s first true killer app for enterprise. And then there’s Mythos. Claude Mythos is a new model that first leaked in late March when Anthropic accidentally left internal documents in an unsecured public data store. What emerged was striking: Mythos is so capable at coding that it autonomously discovers security vulnerabilities in software — at a level that surpasses most human experts. Anthropic says Mythos Preview found thousands of previously unknown zero-day vulnerabilities across all major operating systems and web browsers. One of them was a 17-year-old bug in FreeBSD that allowed complete root access to any machine running NFS. Mythos found it and built a working exploit entirely on its own. In one test, Mythos chained together four vulnerabilities into a single browser exploit that escaped both the renderer and operating system sandboxes. In another, it solved a corporate network attack simulation that would have taken a human expert over 10 hours. That’s why Anthropic chose not to release Mythos publicly. Instead, they launched Project Glasswing — a coalition including Apple, Microsoft, Google, NVIDIA, CrowdStrike, and the Linux Foundation, who get access to Mythos to find and fix vulnerabilities before at

    News from the Woods #136 🥾
See All (29)

About

Exclusively upbeat news from the world’s forests, hills and meadows, with a sprinkling of digital minimalism, AI & startups. newsfromthewoods.substack.com

Information

  • Creator
    by Filip Molcan
  • Years Active
    2024 - 2026
  • Episodes
    29
  • Rating
    Explicit
  • Copyright
    © Filip Molcan
  • Show Website
    News from the Woods