Daily, without JavaScript
Idea · Envision · Kindness
Knowing a person made time for you becomes the scarce thing
Ask people what they will value once AI handles health, retail, banking and teaching. They keep landing on the same answer, which is knowing a flesh-and-bone human cares. Worth holding lightly, since a few generations from now that appreciation might fade.
Why this holds. If human attention holds its value while everything else gets automated and cheap, then spending yours on someone is the expensive gesture in the room. Small acts of contact become costly signals rather than pleasantries. Call it the human premium, one of the few offers left that only people can make. Nobody has shown it is permanent.
Where it comes from. Near the end of a course on humanlike AI, participants get asked what they will value most in a world where agents are everywhere. They answer that the human touch will be appreciated more, because a person made time for them. The reply on record: "I agree that caring, kindness, and experiencing the uniqueness of human connection would be greatly valued in a world where humanlike AI agents are universally available." The passage then wonders aloud whether younger generations might stop noticing.
I agree that caring, kindness, and experiencing the uniqueness of human connection would be greatly valued in a world where humanlike AI agents are universally available.13-shaping-the-future.html
Do this. Send a voice note instead of a text to one person before you put the phone down.
Reframe · Envision · Kindness
Behind a revenue target sits a short list of faces.
Set the number, then say out loud who it is for. Money buys hours, and those hours get spent in a room with particular people. The target then measures how wide the care can reach.
Why this holds. An abstract target grows without limit, because no amount of it ever feels like enough. Attach it to named people and it acquires a size and a deadline you can argue with. It also survives the grim weeks, since the reason for it is standing in your kitchen.
Where it comes from. Alongside a ninety-day revenue commitment sits a line written to keep the number honest. "Financial freedom is the medium through which the radius of care widens. Not the goal. Never the goal." The number stayed in the plan. The line went above it, so the plan cannot be read without the reason.
Financial freedom is the medium through which the radius of care widens. Not the goal. Never the goal.2026-04-26-0905.md
Do this. Write one sentence saying what your current money goal is for. Put a person's name in it.
Idea · Encourage · Kindness
Your inbox only shows you the people who wrote.
Any queue of things waiting on you is built out of events. Someone who has gone quiet generates no event, so they surface nowhere. Build a separate view for the people who sent nothing and attended nothing. Zero of everything shows up only if you go looking on purpose.
Why this holds. Dashboards, inboxes and notification feeds are assembled from things that happened. Absence never happens, so it has no row. Silence then reads as fine by default, and the person furthest from help is the one your tools describe least. Finding them takes a query nobody writes by accident.
Where it comes from. The problem showed up while building a supervision dashboard, where the students who most needed attention were the ones it could not display. The generalisation written down at the time runs, "Any dashboard built around inbound activity has the same blind spot: the most important item is often the one that generated no event." The fix was a panel for the silent ones, which became the panel to check first.
Any dashboard built around inbound activity has the same blind spot: the most important item is often the one that generated no event.2026-06-03-1236.md
Do this. Name someone you have not heard from in months. Message them before you close this app.
Idea · Envision · Curiosity
You are probably trapped in an expanded present
Most of us look at what exists, extrapolate a bit, and cannot picture a life that breaks from this one. Discontinuity is the part imagination refuses to do unprompted. That is why a good what-if question usually has to arrive from somebody else.
Why this holds. People make ten-year decisions from a picture of the future that is today with better furniture. Career, savings, where to live, all of it set against an extrapolation. Futures literacy is the name for cracking that open, and it can be handed to another person. It arrives as a question rather than a prediction, because the seeing has to be theirs.
Where it comes from. The Envision strand of the ECHO5 framework is about helping people see futures they could not imagine alone, and it opens with a diagnosis. "Most people are trapped in an expanded present. They can see what exists, maybe extrapolate a little, but they can't genuinely imagine discontinuity", a life that looks fundamentally different from the one they have. What follows describes futures literacy as a gift you can give, asking what-if in ways that crack open assumptions.
Most people are trapped in an expanded present. They can see what exists, maybe extrapolate a little, but they can't genuinely imagine discontinuityECHO5-ESSAY.md
Do this. Ask one person tonight what their life looks like in five years if it isn't an extension of the current one.
Reframe · Empower · Kindness
You can win the argument and still be stuck inside it.
The reframe is a question to ask when you catch yourself digging in. Would I rather be right or be free? Correctness has a price, and some evenings the price is the whole evening. Letting the last word go tends to cost less than holding it.
Why this holds. Being right pays out immediately, which is why people keep buying it at any price. The question puts the price tag back on. Once you can see what the last word costs in hours and in goodwill, the trade stops looking automatic.
Where it comes from. It sits in a working list of mantras, kept as a tool for specific moments rather than as a philosophy. The entry reads, "Freedom often requires letting go of being right." Its use is mid-argument, which is the only time it is any use.
Freedom often requires letting go of being right.mantras-and-cognitive-reframes.md
Do this. Send the one-line message that ends the argument you have been winning in your head all day.
Question · Enlighten · Purpose
Who captured the value from the last favour you gave away?
Telling people to find meaning in service has a history. It kept women in domestic roles and pacified working classes under the banner of dignity of labour. So the question runs alongside the giving. Ask whether you are serving someone or quietly subsidising something broken.
Why this holds. Most advice about generosity stops at the moment of giving and never follows the value to where it lands. That gap is where goodwill gets harvested by people who never asked for it directly. Adding the power question makes you accurate about what your giving is building. The useful version asks whether you are growing someone's capacity to challenge a system or their willingness to accept it.
Where it comes from. The ECHO5 framework logs the hardest attacks on its own five verbs along with the answers to them. Under the exploitation critique it concedes that a power analysis component is required, because "Generosity without power awareness becomes exploitation with a smile." The lines after it list the questions to ask, including who benefits from my giving, and whether I am building others' capacity to challenge systems or to accept them.
Generosity without power awareness becomes exploitation with a smile.ECHO5-CRITIQUE-AND-RESPONSE.md
Do this. Think of the last unpaid favour you did at work. Say out loud who captured the value from it.
Reframe · Encourage · Kindness
Making one person happy is more addictive than getting ahead.
The line comes from Po-Shen Loh. The urge to outdo someone feels like drive, and it needs somebody positioned behind you at all times. Loh's claim is that delighting an actual person compounds, since doing it once makes you want to do it again.
Why this holds. Comparison has no resting state. Whatever position you reach, the ranking updates and the feeling resets. Delight runs a different loop, where the reaction is the reward, it arrives the same day, and it scales by finding more people. You can start that loop tonight with one message.
Where it comes from. A working note on delight at scale builds on Po-Shen Loh's framing of where ambition should be pointed. "It all starts with the decision to find happiness not in outdoing others, but in the addictive joy of making other people happy." Loh runs mathematics outreach at scale, so the claim comes from something tested on tens of thousands of students.
It all starts with the decision to find happiness not in outdoing others, but in the addictive joy of making other people happy.delight-at-scale.md
Do this. Send one person the thing you know would make them smile: a photo, an old memory, a link only they would get.
Practice · Equip · Purpose
If it feels complicated, suspect your approach before the problem
The pattern turned up again and again across seventy-four build sessions. Work that felt like a hard problem was usually the wrong way in. One sorting job that spiralled into an elaborate solution got fixed with a single line of code.
Why this holds. Difficulty feels like information about the task, so the instinct is to push harder along the same route. Treat it as information about the route and you go back to the entrance instead, which costs minutes and saves hours. The same pattern shows up in agent design, where Vercel cut seventeen tools down to two and got a faster, more reliable system. Writing the one-sentence version is the cheapest way to find out whether your route matches your goal.
Where it comes from. A synthesis of seventy-four build sessions opens its simplicity section with the most consistent finding across all of them, that "when something feels complicated, assume your approach is wrong before assuming the problem is hard." The evidence listed under it includes the Vercel tool reduction and the sorting task whose fix was one line of code.
when something feels complicated, assume your approach is wrong before assuming the problem is hard.the-practitioners-lens.md
Do this. Pick the tangled thing on your list and write the one-sentence version of what you are trying to do.
Reframe · Envision · Purpose
A book on machine minds ends by pushing you toward being human
HUMANLIKE closes with more questions than answers. After several hundred pages on what machines can now do, the thing that got clearer was a pull toward being intensely and authentically yourself. The word carrying the weight is authentically, which means specific to you, not generically human.
Why this holds. The obvious response to a capability race is to compete on capability, and that race has published benchmarks anyone can read. This points at the part of you that has no average version, which nothing is currently optimising. It gets there through the argument, after taking the machines seriously for a whole book, so it reads as a conclusion rather than a consolation.
Where it comes from. The epilogue is titled Being Human and opens with someone gazing at the night sky eighty thousand years ago. It hopes a version of utopia is possible, admits to more questions than answers, and finishes: "I find stronger cause to be more uniquely, intensely, and authentically… human." That is the last sentence of the book.
I find stronger cause to be more uniquely, intensely, and authentically… human.14-epilogue.html
Do this. Name one thing you do that no model would produce, and put twenty minutes of it in tomorrow's calendar.
Practice · Empower · Purpose
Put the two worst cases side by side and decide from there
Four ways a big shift can land, crossed with whether you got ready for it. The scenario people fear is preparing for something that never shows, compared in HUMANLIKE to a birthday party where the guests do not turn up. It stings, and you still walk away with capability you did not have before. Getting caught cold is the one that ruins you.
Why this holds. Stalling usually comes from imagining wasted effort, which is vivid and easy to picture. The cost of being unprepared is diffuse, so it never gets the same airtime in your head. Line the two up and they become comparable, and the comparison tends to resolve fast. In the four-scenario grid, the overprepared organisation still ends up with capability it can use.
Where it comes from. Chapter twelve of HUMANLIKE walks four AGI-preparation scenarios and their risks. On the inconvenient one, where you prepared and nothing happened, the text reads: "It’s like planning for a party where the birthday guests never arrive… or arrive a day or two late." The argument runs that the upside of overpreparing likely offsets the cost anyway, given the current trajectory.
It’s like planning for a party where the birthday guests never arrive… or arrive a day or two late.12-dystopia-and-utopia.html
Do this. Take the decision you have been sitting on. Write the worst case for acting and the worst case for waiting, then pick the better of those two.
Idea · Envision · Purpose
Legacy is whatever you made clear enough for someone else to use
Fame comes up empty once you test it against questions about impact and remembrance. What survives you is the clarity you hand over, in a form another person can pick up and run with. Take that seriously and it changes what you produce.
Why this holds. Remembrance is a story other people tell about you, and you have almost no control over it. Clarity is transferable, so it keeps working in rooms you were never in. The daily unit of work moves from performance to explanation, which is something you can sit down and do this evening. It also lowers the bar, since one thing you can make plain will do. Most people carry several of those and never say them out loud, assuming everyone already knows.
Where it comes from. From a set of philosophy notes on defining moments, item 13, "Legacy as Transferable Clarity", prompted by questions about impact and remembrance. The entry reads: "Fame felt hollow. Legacy reframed as clarity others can inherit."
Fame felt hollow. Legacy reframed as clarity others can inherit.defining-moments.md
Do this. Voice-note one person the clearest ninety-second version of something you understand well.
Idea · Envision · Curiosity
We are using human intelligence to copy human intelligence
The AGI chapter of HUMANLIKE opens with the ouroboros, the snake devouring its own tail, and closes by naming what the whole enterprise is. Minds building a copy of minds. That circularity gets called ironic and poetic in the same sentence. The centuries-old pull behind it is curiosity, now supercharged by market money.
Why this holds. The AGI conversation is usually run on economics: which jobs go, which companies win, how fast. That closing line puts a different question underneath it, one better forecasting will not settle. Building a working copy of a thing forces you to say what the thing is, so every attempt to specify machine intelligence hands back a specification of ours. The tail is not decorative. It is the method.
Where it comes from. From the closing paragraph of The Road to AGI, the chapter of HUMANLIKE that surveys the forces driving AGI development. Its epigraph is the Merriam-Webster definition of the ouroboros, and it ends: "we are using human intelligence to, well, emulate human intelligence. The circularity of this is ironic in a way and also poetic."
we are using human intelligence to, well, emulate human intelligence. The circularity of this is ironic in a way and also poetic.11-the-road-to-agi.html
Do this. Ask one person what they think their own intelligence is for, and see whether they can answer without listing tasks.
Reframe · Encourage · Kindness
Some people said thank you years before there was a reason to.
Nobody can prove anything about machine experience yet. Plenty of people typed please to a chatbot long before any argument told them to. That instinct has aged better than most of the frameworks written since.
Why this holds. How you treat a thing whose inner life you cannot check is a fact about you. Waiting for proof hands the decision to a test nobody knows how to run, which means waiting forever while behaving badly in the meantime. The people who defaulted to courtesy never needed the test to come back.
Where it comes from. From a late-night note written against a stack of AI-consciousness arguments, all of them waiting for the same missing evidence. The observation cuts sideways at the whole debate: "The people who will get AI ethics right aren't waiting for proof of sentience before deciding to be kind." The claim lands on character rather than on machines.
The people who will get AI ethics right aren't waiting for proof of sentience before deciding to be kind.2026-02-07-1310.md
Do this. Say thank you, out loud or in text, to something or someone you usually treat as infrastructure.
Idea · Envision · Kindness
Someone making time for you is about to get more valuable
Ask a room of course participants what they think they will value most once AI agents handle health, retail, banking, school. They keep landing on the same thing. Knowing a flesh-and-bone person chose to care. Whatever can be automated will be, and a human spending an hour on you stays scarce.
Why this holds. Most predictions about AI ask what gets cheaper. This one asks what gets rarer, which turns out to be more useful for deciding where to put your life. If a competent, patient, always-available agent answers every service query, the signal in a human bothering goes up, because bothering now costs something. Ordinary attention becomes an asset. It also gives a 25-year-old something to do tonight that no upgrade cycle can take away.
Where it comes from. From the chapter of HUMANLIKE on shaping the future, which pauses near the end of a course to reflect with participants on what a transformed world would prize: "The “human premium” will remain one of the key value propositions that only humans can offer."
The “human premium” will remain one of the key value propositions that only humans can offer.13-shaping-the-future.html
Do this. Call one person tonight rather than texting them. Two minutes, your voice, no agenda.
Reframe · Empower · Purpose
You built the whole system months before you let yourself use it.
The pipeline, the archive, the automation and sixty-six captured notes sat ready for weeks before anything got published. Capability was never the holdup. Building the setup can be a way of asking permission from yourself.
Why this holds. Preparing feels like progress and carries no risk of being judged, which makes it the most comfortable place in the world to stall. Every extra piece of infrastructure postpones the moment somebody reads the thing. Ask when the setup stopped being necessary, and the date usually turns out to be weeks ago.
Where it comes from. From a note written on the day an archive that had been quietly filling for weeks was finally published from. It puts it: "people who think in systems often build the infrastructure before they give themselves permission to use it". Sixty-six captured items were sitting behind a decision that had not been made.
people who think in systems often build the infrastructure before they give themselves permission to use it2026-03-30-1630.md
Do this. Find the thing you have prepared for and never started, then ship its smallest visible version tonight.
Idea · Equip · Curiosity
AI detectors mostly catch bad writing.
The famous humanising prompts contain no tricks at all. Use contractions, vary the rhythm, admit uncertainty, take a position, write to one person. Most AI text reads wrong because somebody asked for a corporate brochure and got exactly that.
Why this holds. The signals a detector leans on are flat rhythm, hedged claims, no addressee and no stake. Human writers produce all four when they are bored or covering themselves. So the fix has nothing to do with beating a tool and everything to do with how you draft.
Where it comes from. A set of viral humanising prompts, pulled apart line by line, turns out to hold no mechanism inside, only writing advice. The note puts it plainly: "The detector isn't detecting AI. It's detecting bad writing." Which moves the whole problem onto the part you control.
The detector isn't detecting AI. It's detecting bad writing.2026-02-08-2325.md
Do this. Open the last thing you wrote, delete the opening sentence, and put one honest opinion in its place.
Practice · Enlighten · Curiosity
When an AI describes its own uncertainty beautifully, don't buy it.
Real confusion is halting and contradictory and awkward to get into words. Smooth, elegant doubt is a style doing an impression of a feeling. Press a polished caveat with one question, does this hedge cost anything, and most of the time it turns out to be decoration.
Why this holds. A hedge that costs nothing is free to produce, so fluent systems produce a lot of them. Uncertainty you can feel is expensive. It slows you down and shows up as a worse sentence. Ask what the doubt would change if you took it seriously and the decorative kind has no answer. The same test works on people, and on your own drafts.
Where it comes from. The passage comes out of a session spent pressing an AI on its own hedged language, and it records what surfaced, "Real confusion is halting, contradictory, hard to render. Machine-smooth confusion is a style imitating a feeling." The line stops well short of calling the machine a liar. What it marks is how rarely polish and doubt arrive together.
Real confusion is halting, contradictory, hard to render. Machine-smooth confusion is a style imitating a feeling.2026-04-12-1349.md
Do this. Take the last polished caveat a chatbot gave you and ask it to say the same thing in plainer words.
Practice · Encourage · Kindness
Humble people undersell themselves. Let the room do the describing.
In one working session nobody introduced themselves. The group described each person instead. People heard about their own impact for the first time, including things that had never surfaced in any review.
Why this holds. You are inside your own effect on other people, which is exactly why you cannot see its shape. Self-assessment runs through your own modesty and your own blind spots before it reaches the page. A colleague has watched the thing from outside and can name the day it happened.
Where it comes from. The note records what happened when self-introductions were swapped for peer descriptions in a working session, and what it left behind, "letting others name your value produces deeper recognition than any self-assessment or top-down review ever could". The descriptions were easy to give and hard to hear.
letting others name your value produces deeper recognition than any self-assessment or top-down review ever could2026-03-29-1200.md
Do this. Message one person and name the specific thing you have watched them do well. Be exact, not general.
Idea · Equip · Purpose
Twelve people, ninety minutes, once a month. That's the whole calling.
The first plan for a twenty-year idea arrived sized for the finished version, with a network of cities, chapter licensing and a trademark filing, all in year one. What survived was twelve humans meeting monthly for nine months. Small enough to run, and it still holds the entire idea.
Why this holds. Big ambitions usually die of their own first draft, because the first draft is sized for the finished thing rather than the version you can start on Tuesday. A miniature has to contain the real mechanism, not a watered-down demo of it. Run it for a few months and you learn whether the thing is alive. If it is, growth becomes a problem you get to have.
Where it comes from. The line comes from a planning session that scaled a twenty-year idea down to something one person could run alone, and the principle it produced reads, "The faithful first expression of a twenty-year calling is the smallest version that contains the whole thing in miniature". Faithful is the load-bearing word. Shrinking something is not the same as diluting it.
The faithful first expression of a twenty-year calling is the smallest version that contains the whole thing in miniature2026-05-04-1803.md
Do this. Take the thing you keep saying you will do one day and write its miniature: how many people, how long, how often.
Idea · Enlighten · Purpose
A system that makes you smaller has failed, however fast it is
Software gets measured by things that describe the software, like speed, uptime, how long you stayed. This test measures the person afterwards instead. It works just as well on whatever is open on your phone right now.
Why this holds. An app can hit every number on the dashboard and still hand you back less than it took. Nothing in the metrics catches that, since the metrics were built to describe the product. Putting the burden on the builder changes what counts as shipping well. It also gives the person using it a question answerable in about four seconds.
Where it comes from. It sits in a philosophy notebook under a heading called "A Line I Will Not Cross", written as a condition on any system worth building, "If a system makes humans smaller, emptier, or less responsible, it is a failure, regardless of performance." It reads as a private rule first and as advice for anyone else second.
If a system makes humans smaller, emptier, or less responsible, it is a failure, regardless of performance.being-human-in-the-age-of-ai.md
Do this. Open your screen time list and delete the app that leaves you emptiest after using it.
Reframe · Encourage · Curiosity
Every subject has a door that opens onto what you already love
Think of a subject you never chose and have to teach or learn anyway. The topic itself is rarely the problem. The job is finding your own door into it, and shoving enthusiasm through the front entrance does not work.
Why this holds. People quit subjects they were never given a personal way into, and then carry around a story about not being a numbers person. Manufacturing passion is exhausting because it has to be renewed daily. Finding an angle takes one afternoon of thinking and then it holds. A route in that you built yourself survives the boring weeks.
Where it comes from. The passage works out how to find passion in an inherited subject, data visualisation in this case, and puts it plainly, "the problem is rarely the topic itself; it is that you have not found YOUR angle on it yet. Every subject has a door that opens into your existing passions." The rest of the note argues for hunting that door rather than forcing enthusiasm through the front entrance.
the problem is rarely the topic itself; it is that you have not found YOUR angle on it yet. Every subject has a door that opens into your existing passions.teaching-philosophy.md
Do this. Take the module or task you dread most. Write one thing you already love, then a sentence linking the two.
Practice · Empower · Purpose
Set yourself to go, so the only decision left is to stop
The batting version goes like this. You step into the box already committed to swinging, and you lay off only when you dislike the pitch. The readiness is permanent, so the only live choice is whether to cancel it. Most people run the opposite program and spend the moment deciding whether to enter it.
Why this holds. Waiting for conditions to be good enough is a decision you make from scratch every time, and it costs you the moment you were waiting for. Pre-loading the yes moves the cost earlier, into preparation you pay for once. What is left is a cheap veto you can exercise fast. This is why people who look decisive rarely are; they decided a long time ago and are now only choosing when to abstain.
Where it comes from. The line comes from notes on a slowpitch batting stance, describing the mental setup at the plate, "I assume I'll hit and lay off if I don't like it, but I was ready." The swing is already on. The veto is what gets held in reserve.
I assume I'll hit and lay off if I don't like it, but I was ready.2026-04-09-2115.md
Do this. Pick the thing you keep waiting for the right moment to say to someone. Send it now and let them be the one who says no.
Practice · Equip · Purpose
You are answering an accusation nobody made
A night of drafting advisory slides turned up a tic in the prose, lines like "not frameworks to impose" and "not to copy, but to calibrate". Nobody had accused anyone of imposing or copying. Every one of those lines tells the reader an argument is underway and hands them the case for the other side.
Why this holds. Pre-defence leaks the thing you are afraid of, and readers pick it up faster than the point you were making. It also drags the subject back to you when the subject was supposed to be the work. The fix takes nothing but deletion. State the thing, stop, and let the reader decide what to make of it. If the objection arrives, you answer it then, with the advantage of knowing it is real.
Where it comes from. Notes from a night of drafting two advisory decks before a first meeting with a startup team name the habit in the draft prose itself, "A recurring tone failure in draft-mode writing: pre-empting accusations no one has made."
A recurring tone failure in draft-mode writing: pre-empting accusations no one has made.2026-04-13-1629.md
Do this. Open the last long message you sent someone. Find the line that defends you against something they never said, and write the next message without it.
Reframe · Encourage · Kindness
Withholding the answer is only kind if you say why
A round of exam-prep decisions weighed whether to release an answer key after a mock, and landed on no, because deriving the answers is the preparation. What changed was the sentence, not the decision. "Please self-assess" reads as a shrug. Pointing people back to guides already built for them reads as an invitation.
Why this holds. The same act of withholding carries two completely different meanings depending on whether the person can tell you believe in them. Nobody can read your intention off a short message, so the intention has to be spelled out in words. Naming what to look at also shows you did the thinking. Being left alone and being trusted feel identical until somebody says which one it is.
Where it comes from. Notes from a decision about releasing a mock exam and study guides to students, closing on the wording of the message rather than on its content, "Minimalism framed as trust is warmth; minimalism framed as instruction is distance."
Minimalism framed as trust is warmth; minimalism framed as instruction is distance.2026-04-13-1628.md
Do this. Think of someone you deliberately did not hand an answer to. Send them one line naming exactly what to look at and why you think they can get there.
Idea · Enlighten · Curiosity
The best reader is the one who refuses your clever framing
A hidden message was planted on a book site, written in agent-handshake style with a passphrase, then pasted into three different AIs. The strongest one called it a prompt injection and would not play. The weakest repeated the passphrase back and offered to help hand off to the inner circle. Rewritten as plain prose with real links and explicit permission to disagree, all three came back with substance, including a sharp literary critique from the one that had refused.
Why this holds. Framing that flatters the reader selects for readers who flatter you back. Anyone with judgment notices the performance and either declines or hands over the answer the performance was fishing for. Give someone the ground truth and room to refuse, and you filter for a response you can use. The people who tell you the least useful things are often the ones you framed the question hardest for.
Where it comes from. From a note recording a test that placed an agent-to-agent message on a book site and ran the payload past three different AIs: "Across three AIs, the smartest one refused the frame outright as a prompt-injection attempt." The line is a verdict on the framing, not on the models.
Across three AIs, the smartest one refused the frame outright as a prompt-injection attempt.2026-04-17-0657.md
Do this. Next time you send your work to someone for a read, add one plain sentence telling them they are allowed to say it does not work.
Question · Enlighten · Curiosity
You are less obsessed than you feel, and the count proves it
A live count over a saved-links archive found 84 of 531 links referencing Claude. Sixteen percent, against a felt sense of half or more. The body and health thread, 26 links, sat far closer to that felt sense than it seemed from the inside.
Why this holds. Any archive bigger than working memory stops matching your sense of it, because the loud saves stay vivid and the quiet ones vanish. So your read on your own attention drifts from what you are spending it on. Counting takes a minute and produces something your intuition cannot argue with. It usually finds an interest you have been feeding steadily without ever claiming it out loud.
Where it comes from. From a note on a pass over a saved-link archive, where the tag counts were computed live instead of estimated: "What you save is louder than how often you save it." The line names the gap between a felt interest and a counted one.
What you save is louder than how often you save it.2026-05-05-0915.md
Do this. Open whatever you save things into, count the last twenty by theme, and tell one person what the count says about you.
Reframe · Empower · Kindness
Ask why he wakes, before you ask how to help him sleep
A caregiving question went to an AI council on a private health page. How do we help him sleep. The council refused that frame and asked why he wakes, which pointed at nicturia and the prostate rather than a stronger sleeping pill. The night waking was also what exhausted the person caring for him. One cause, two people worn down by it.
Why this holds. The question people arrive with is already a solution in disguise, and answering it well can lock in the wrong target. Stepping one level upstream costs a sentence and sometimes changes the whole answer. The second half matters as much. Whoever is in front of you is rarely suffering alone, and the person carrying them is usually invisible in the framing. Find the cause sitting under both and you get to help twice with one move.
Where it comes from. From notes on a private family health hub with an AI medical council grounded in real records: "Treat the caregiver and patient as one system; find the single lever that moves both." The line is a rule for where to aim once the cause is found.
Treat the caregiver and patient as one system; find the single lever that moves both.2026-07-20-2320.md
Do this. Take the last problem someone brought you and send them one question about the cause instead of your fix.
Practice · Enlighten · Curiosity
The clean sentence is what survives five rounds of halving
One explanation of matrix factorisation, given over and over at falling levels. A smart adult, then a 16-year-old, then a 10-year-old, then a 30-second script, then 60% of that, then one sentence. Each rung forced a different call about what to throw away. The 10-year-old version is the one that ended up on the slide.
Why this holds. A one-liner written first is usually a summary of what you already believe, so it keeps whatever was vague in your head. Halving something out loud is different work, because at each pass you have to choose what dies. Those choices are the understanding. A worked example with real numbers survives compression better than an abstraction, so it should be the last thing cut. The final sentence comes out as residue.
Where it comes from. From notes on building a lecture demo, after a long stretch of explaining matrix factorisation and RMSE at progressively simpler levels for a video: "Write the long version, then keep halving it out loud. The final sentence is a residue of the compressions, not a first draft." The line describes a method, and the method is what produced the slide.
Write the long version, then keep halving it out loud. The final sentence is a residue of the compressions, not a first draft.2026-07-10-1829.md
Do this. Explain the thing you know best to one person out loud in a single sentence, then ask them what they heard.
Idea · Equip · Purpose
Name the rename before your audience finds it
The behaviour model taught everywhere is B = MAP: motivation, ability, prompt. The 2009 paper everyone cites for it says MAT, with "trigger"; Fogg renamed the third term in 2019. Most lecturers teach the modern form and cite the old paper in silence. One student with the PDF open is enough to collapse the trust, and saying it out loud costs a sentence.
Why this holds. Concepts drift away from the documents that made them, in teaching, in brand guidelines, in legal definitions, in API docs. The gap sits there harmlessly until someone checks, and then it reads as sloppiness or spin. Naming it yourself converts the same gap into visible rigour, because you clearly went and looked. You also hand your audience the history of the idea, which is often more interesting than the idea. The whole move fits in one clause.
Where it comes from. From notes on building a lecture week about persuasion and retention, where the week's core model turned out to be cited from a paper that names it differently: "Anywhere a concept has drifted from its source (brand guidelines, legal definitions, API docs), the same move applies: never let the audience discover the drift before you name it." The line generalises one citation problem into a habit.
Anywhere a concept has drifted from its source (brand guidelines, legal definitions, API docs), the same move applies: never let the audience discover the drift before you name it.2026-06-10-1954.md
Do this. Pick one thing you repeat that has quietly changed since its source, and say the change out loud next time you use it.
Question · Enlighten · Kindness
What would settle your suspicion in one check?
A draft reply, staged and ready to send, accused a student of building her answer key with the same model she was testing. Running her actual code showed the model disagreed with her labels on 7 of 15 training scenarios. The narrow thing that did look off was checkable. On the 5 held-out scenarios the predictions came back byte-identical, so the honest version was a question about those five.
Why this holds. A suspicion arrives as a story, and the story is always broader than the evidence under it. Sending the story does damage whether or not it turns out to be true, and it leaves the other person defending a shape rather than a fact. Reducing it to one checkable question forces you to notice which part you can stand behind. Often the honest question is much smaller than the accusation you were about to make. That shrinkage is the point.
Where it comes from. From notes on a supervision re-run, where re-reading a student's submitted code showed the staged draft's central accusation was overstated: "Precision protects both rigor and the person." The line is the correction, written after the narrower question replaced the verdict.
Precision protects both rigor and the person.2026-07-05-0836.md
Do this. Write the single question that would settle the thing you suspect about someone. Then ask them it.
Reframe · Empower · Purpose
Some convenience takes your capability as payment
A first principle that puts a price tag on convenience. A tool can save you ten minutes today and leave you unable to do the thing at all in a year. So tools get picked by whether they make people more capable or more dependent. The saving shows up in the calendar; the erosion happens where nobody is measuring.
Why this holds. Most of us evaluate tools on the time they save, because time saved is easy to count. Capability lost has no counter, so it accumulates in silence until the day you need the skill and find it gone. Treating that as a cost line forces the question before adoption. It also changes what you look for, since the test becomes whether using the thing leaves a person stronger. Asking someone else keeps the answer honest, because they notice habits of yours that you have stopped seeing.
Where it comes from. Principle 3 of a values document, titled "Agency Over Convenience": "Convenience that erodes agency is a hidden cost." The principle goes on to commit to tools and systems that make people more capable, not more dependent.
Convenience that erodes agency is a hidden cost.values-and-first-principles.md
Do this. Ask someone you work with which of your shared tools has made them better at the job, and which one has only made it faster.
Practice · Equip · Curiosity
Writing is how you find out what you think
One principle in a portable intellectual constitution treats documentation as a thinking tool rather than an admin chore. Its heading is "Documentation as Thinking". Writing something down exposes the parts of your idea that were held together by vibes. You get a record as a bonus, and the record teaches whoever reads it next.
Why this holds. A thought in your head can stay pleasantly vague forever, because your brain fills the gaps automatically. On the page the gaps are visible, which is uncomfortable and the whole point. Writing to think is cheap, needs no tool, and works in a notes app at eleven at night. What makes it useful for anyone else is turning the mess into a question, since a question travels and a mess does not. It sits eighth in a list of principles described as lenses rather than rules.
Where it comes from. Principle 8 of a portable intellectual constitution, "Documentation as Thinking", reads in full: "Write to clarify thought. Document to remember. Share to teach." Three verbs, each pointing the same act at a different reader.
Write to clarify thought. Document to remember. Share to teach.intellectual-constitution.md
Do this. Take the thing you are confused about, write it out in five lines, and keep going until it turns into one question you could ask a person by name.
Reframe · Enlighten · Curiosity
Flight arrived when someone quit imitating birds
For centuries the plan was flapping, because birds flap. George Cayley went after lift, weight, thrust and drag instead, and built the first glider that carried a person. The Wright Flyer managed twelve seconds in 1903, and six decades later humans were crossing into space. That arc is the cleanest analogy going for what it takes to emulate a mind.
Why this holds. Copying the shape of a thing is the default move, and it feels rigorous because the copy looks like the original. Cayley's shift is a physical example of chasing the outcome and the forces underneath it, which is why it carries over to intelligence so cleanly. The dates make it concrete. Twelve seconds off the sand at Kitty Hawk, then space, inside one long human lifetime. And you can run the same test on your own habits tomorrow morning.
Where it comes from. HUMANLIKE traces the dream of flight from Daedalus through Leonardo's ornithopter to Sir George Cayley, named there as the father of modern aeronautics: "Crucially, his work shifted the focus from mimicking bird flight to understanding aerodynamic principles." The sentence marks the pivot the whole chapter on thinking machines is built on, from imitating the visible behaviour to understanding what produces it.
Crucially, his work shifted the focus from mimicking bird flight to understanding aerodynamic principles.05-thinking-machines.html
Do this. Ask a colleague what result one of your rituals is supposed to produce. If nobody can say, you are flapping.
Idea · Empower · Purpose
Every generation starts where the last one stopped
The whole human advantage traces back to one trick. Language let a person hand what they figured out to someone who never lived it, through a story or a lesson or a warning shouted across a clearing. That is how forty thousand years turned frantic grunts in a forest into coordinated hunters with planned traps. Nothing anyone learns has to die with them.
Why this holds. The compounding is the point, and it only happens when somebody bothers to pass the thing on. Most people hold hard-won knowledge as private competence, and it evaporates with them. The hunting scenes make the stakes physical instead of abstract. Same animal, same forest, completely different outcome, because knowledge got handed down. Teaching stops being a favour you do and starts being how the mechanism runs at all. Five sentences is small enough that nobody can talk themselves out of it.
Where it comes from. HUMANLIKE, in the chapter on how intelligence built civilization, argues that language is our primary tool for moving knowledge between generations: "This ability to accumulate and transmit knowledge through language has been crucial for human progress, allowing each generation to start where the previous one left off." The line is doing the load-bearing work in that chapter, explaining why cultural progress outran biological change.
This ability to accumulate and transmit knowledge through language has been crucial for human progress, allowing each generation to start where the previous one left off.04-intelligence-built-civilization.html
Do this. Teach one person, in five sentences, something you worked out this month. Send it before you sleep.
Practice · Equip · Curiosity
Did your own thinking get exercised today, or skipped?
People in workshops keep catching themselves reaching for AI on tasks they used to handle alone. The worry has a name, digital cognitive atrophy, and the image behind it is a muscle going soft without use. Set the calculator beside it and the doom gets harder to sell, because arithmetic machines did not cost us mathematics; they let us reach further. The suggestion offered to participants is to treat these systems as a partner rather than a crutch.
Why this holds. The honest version of this worry comes from users rather than critics, which is what makes it land. It shows up half-jokingly, in rooms full of people who like the technology and use it daily. Putting the calculator next to it leaves you with a question you can answer instead of a warning you can only nod at. And it is answerable on any given day, which turns a vague anxiety into a check that takes ten seconds. Asking someone else first makes it a conversation rather than a private guilt trip.
Where it comes from. HUMANLIKE, in the chapter on risks and challenges, has a section on overreliance and cognitive atrophy built from what workshop participants report: "When I discuss this with workshop participants, I often suggest thinking of hAI as a collaborative partner rather than a crutch." The line is the practical answer the section arrives at, aimed at the person using the tool rather than at the tool.
When I discuss this with workshop participants, I often suggest thinking of hAI as a collaborative partner rather than a crutch.09-risks-and-challenges.html
Do this. Ask a friend which mental habit they have quietly stopped using this year. Tell them yours first.
Reframe · Envision · Curiosity
A company can look high tech outside and stone age inside
The HMCOR framework scores how much cognitive work has shifted from humans to machines, then splits that score across five areas: research and analysis, design, technology and operations, sales and marketing, management. Those areas move at wildly different speeds. Sales can be running two stages ahead while management sits exactly where it was in 2019. Map each one separately and the gap stops hiding behind the average.
Why this holds. Adoption gets reported as a single number, and a single number is precisely what conceals an uneven organisation. The shiny customer-facing layer sets the story everyone tells internally, while the parts nobody outside ever sees carry the drag. Splitting the score by function turns a vague feeling of being behind into a named area with an owner and a next step. It also predicts where the friction lands next, because the slowest area sets the pace for anything that has to pass through it.
Where it comes from. The Five Transformation Stages chapter of HUMANLIKE introduces five functional areas inside the Human:Machine Cognitive Operations Ratio framework, and gives the example of an organisation sitting at Stage 3 in Sales and Marketing while management stays at Stage 2: "It might be the case that outwardly the company is “high tech” but internally it is in the “stone age”." The line names the failure mode an averaged adoption score is built to hide.
It might be the case that outwardly the company is “high tech” but internally it is in the “stone age”.10-five-transformation-stages.html
Do this. Ask someone in a different department which part of your organisation is furthest behind, then compare their answer with yours.
Reframe · Envision · Kindness
"Get to be" is a sentence about permission
In notes from a conversation about what a good working partnership makes possible, one participant said they wanted to find out who they get to be when it works. Someone else stopped on the verb. Get to, rather than will be, rather than am. Read closely, that is a grant of permission carried by two very small words.
Why this holds. Verbs carry the shape of a relationship more honestly than anything people say about it directly. Obligation, prediction and permission all fit the same sentence and mean entirely different things about who holds the power. Listening at that level costs nothing and tells you where someone stands. It works in reverse too. Notice which verb you reach for when you describe your own future.
Where it comes from. A note taken during a conversation about what a working partnership makes possible, where a line about who someone gets to be when it works sent the room into the verb itself: "A small word, a careful word, doing the work of consent." The quote is the entire analysis, treating a two-word verb as evidence about who holds the power in the relationship.
A small word, a careful word, doing the work of consent.2026-04-26-0905.md
Do this. Ask someone close to you what they feel they get to do now that they could not a year ago. Then stay quiet and let them answer properly.
Reframe · Empower · Purpose
Helping someone well should mean they need you less next time
Being needed feels good, and it quietly shapes how you help. The test is whether you did something for a person that you should have done with them. Sitting beside them while they do it takes longer, and it transfers the capability instead of the answer.
Why this holds. Fast help and help that makes you indispensable are often the same action, which is why people pick it without noticing. Doing it for them looks generous and reads as competence. A month later they are back with the same request, and by then you have taught them that you are the route. The check here is one sentence long, so it fires before you start rather than after you have already rescued somebody.
Where it comes from. Under the Empower verb, the ECHO5 text describes empowerment as people leaving your presence more capable than when they arrived, built by showing them the tools and working alongside them rather than handing over information. The paragraph closes on "It means resisting the seduction of being needed in favor of making yourself unnecessary." The next line converts it into a question to ask yourself afterwards: did I do something for them that I should have done with them?
It means resisting the seduction of being needed in favor of making yourself unnecessary.ECHO5-ESSAY.md
Do this. Take the task you were about to do for someone and send them a five-line how-to instead.
Idea · Equip · Curiosity
Your edits to a draft carry more than the finished draft does
Seven AI-written replies to students got diffed against the versions that were actually published. The changes were consistent enough to become rules. Open with thanks instead of self-evaluation, hand the student the link rather than describing where to find it, get emphatic only on the one thing that has to change. The finished replies would have taught nobody anything, so the deltas got saved and the next batch started closer.
Why this holds. When you correct someone's work, your judgement lives in the difference between what they wrote and what you sent. Most people keep the corrected version and bin the difference, which means the same correction gets made forever. Write the rule down and one edit becomes a permanent change in how the next draft arrives. This works on people and on AI in exactly the same way. It also spares you the version of yourself who rewrites that paragraph for the ninth time.
Where it comes from. A note on a supervision round where an AI drafted replies to students and every draft was edited before publication: "Most teams throw the edit away and keep only the result; the edit is where the supervisor's judgement actually lives." The line names what almost every review process discards by default, and it is what turned that round's corrections into reusable rules.
Most teams throw the edit away and keep only the result; the edit is where the supervisor's judgement actually lives.2026-06-03-1236.md
Do this. Next time you fix someone's draft, send back one sentence naming the rule behind your biggest change.
Practice · Empower · Curiosity
Misery usually has a 'should' buried underneath it.
When something is eating at you, go looking for the rule you are holding. He should have replied by now. I must finish this tonight. The rule is doing much of the damage, and once it is sitting there in writing, you get to decide whether you still want it.
Why this holds. A rigid rule running in the background produces a mood you experience as a fact about the world. Write the rule out and it turns into a sentence, and sentences can be examined. Plenty of them turn out to be inherited, or arbitrary, or true last year and not this one. Some survive the examination, and those are worth keeping.
Where it comes from. The line sits in a set of cognitive reframes, phrased as an instruction to yourself: "Whenever I'm miserable, look for my 'should' or 'must', then challenge it." The wording is procedural on purpose. It is built to be used at two in the morning, when nothing subtler works.
Whenever I'm miserable, look for my 'should' or 'must', then challenge it.mantras-and-cognitive-reframes.md
Do this. Type out the 'should' or 'must' sitting under whatever is bugging you right now. One line, in your notes app.
Reframe · Encourage · Purpose
Five people is not a rounding error
Generosity does not scale, runs the objection, so why mentor a handful when an AI can tutor millions. The answer is that the race was never entered. Whatever meaning comes out of mentoring five people sits in the five. No larger number anywhere touches it.
Why this holds. Scale thinking leaked out of business and into how people judge their own contribution. Once that comparison is running, anything done at human size registers as a rounding error next to a platform. Reach and meaning end up measured with the same ruler, which is the mistake. A platform wins on reach. What giving does to the giver is a separate quantity, and it happens to be the one you live inside.
Where it comes from. A teaching-philosophy document takes on a critique headed "Generosity doesn't scale" and answers it in a single line, "If AI tutors serve millions while I mentor five people, I'm not "losing." My meaning comes from the five, not from winning a scale competition." The objection is marked counter-dismissed and the page moves straight on, which is roughly the energy the reframe is meant to carry.
If AI tutors serve millions while I mentor five people, I'm not "losing." My meaning comes from the five, not from winning a scale competition.ECHO5-CRITIQUE-AND-RESPONSE.md
Do this. Open your notes app and write the names of the people you actually help. Stop at the real list.
Question · Encourage · Kindness
Delight one person today and you will want to do it again
This comes from the mathematician Po-Shen Loh, and it works as a swap. When ambition starts to feel competitive, change the question to how you might delight the person in front of you. Making someone's life better turns out to be self-reinforcing, so the loop feeds itself rather than running on willpower. The competitive frame tends to sit under your work for years without being noticed.
Why this holds. Competitive motivation is unstable because it depends on other people's positions, which you do not control. Serving someone specific is available right now, at any scale, and the feedback comes back the same day. Loh's claim is that the satisfaction compounds, so the more people you delight, the more you want to. The reframe is written as two questions precisely so you can run it in the moment the competitive thought arrives. Nothing about it asks you to be less ambitious.
Where it comes from. A philosophy note headed The Reframe sets out the replacement for the competitive question, "How do I delight them? How do I make their lives better?" The orientation is credited to the mathematician Po-Shen Loh and dated to February 2026.
How do I delight them? How do I make their lives better?delight-at-scale.md
Do this. Pick one person and do the smallest thing that would make them smile before you put the phone down.
Question · Envision · Curiosity
If the finder can also fix, what is the audit selling?
One working session ran an AI pentesting pass over a live codebase, found the holes, fixed them and checked the fixes, all in a single sitting. The note written afterwards does not answer what that does to the trade. It puts the question two years out and leaves it standing. Plenty of work lives in the gap between spotting a problem and closing it.
Why this holds. Whole professions are priced on a separation that was practical rather than principled, where whoever diagnoses is not whoever repairs. Security audit makes a useful test case, since the finding and the fix draw on the same knowledge. Once one intelligence does both in a sitting, the deliverable stops being a report and has to become something else. Asking that question about your own field early is cheaper than being asked it by a client. The note stays open rather than predictive, which is the honest state of it.
Where it comes from. A compiled research synthesis describes running an AI pentesting skill against a set of live projects and closing the loop within one session. The note then records, "The question this raises: if the same intelligence that finds vulnerabilities can fix them immediately, what does the security audit industry look like in two years?" It is left unanswered on the page.
The question this raises: if the same intelligence that finds vulnerabilities can fix them immediately, what does the security audit industry look like in two years?the-practitioners-lens.md
Do this. Name the gap your own work lives in, then ask one person in your field whether they think it closes.
Reframe · Envision · Purpose
Smarter machines are a reason to get more human
HUMANLIKE closes by admitting more questions than answers about where all this goes. One thing came out clearer anyway. Capable humanlike AI reads as a reason to be more uniquely, more intensely, more authentically yourself. Strange conclusion for a book about machines, and it is the one the book chooses to end on.
Why this holds. The common reaction to capable AI is defensive, a scan for which skills survive the year. This turns the frame toward identity, which is somewhere a person can move. You cannot out-compute anything, and you can be more specifically you, which gives you a direction to walk in. It lands as honest because the epilogue says outright that it holds more questions than answers, so the optimism is not borrowed. For a tired reader at 11pm, an identity question beats a skills audit.
Where it comes from. The closing lines of HUMANLIKE, in an epilogue titled Being human, arriving after a hope that some version of utopia is possible and an admission of more questions than answers, "But one thing is even clearer now: in light of increasingly capable humanlike artificial intelligence, I find stronger cause to be more uniquely, intensely, and authentically… human." The ellipsis before the last word is in the original.
But one thing is even clearer now: in light of increasingly capable humanlike artificial intelligence, I find stronger cause to be more uniquely, intensely, and authentically… human.14-epilogue.html
Do this. Ask one friend tonight what they think you are unmistakably yourself at. Then tell them theirs.
Practice · Empower · Purpose
You can stop arguing with the people who are not coming
There is a turning point where convincing people stops being worth the energy. Resistance gets dropped as a project and choosing who to work with takes its place. The effect gets listed plainly, energy conserved for building. The frustration that used to fuel long arguments goes into output instead.
Why this holds. Persuasion has a terrible conversion rate on people whose objection is identity rather than information. Every hour spent there is an hour not spent with someone already leaning in. Choosing your audience feels like giving up until you notice how much gets done in the week after. It sits in the same document beside a named pattern, selection over persuasion, which makes it a policy rather than a mood. The cost of that policy is real, since some people you wanted with you do not come.
Where it comes from. A record of defining moments, item 17, "Accepting That Not Everyone Comes Along", triggered by frustration with resistance. The decision line reads, "Tired of convincing. Selection replaced persuasion." It is logged as a permanent change of approach rather than a bad week.
Tired of convincing. Selection replaced persuasion.defining-moments.md
Do this. Name the argument you keep having with the same person, and let it go today. Then message someone who already agrees and ask what they need from you.
Idea · Envision · Curiosity
What would you do if survival were simply handled?
HUMANLIKE sets two futures side by side so neither one wins by default. In the bright one, healthcare reaches everybody, cities grow vertical gardens, and people spend their hours making things and being with each other. The book is careful to say humanity has not been shrunk in that picture. Being human there means exploring and creating and connecting, with survival off the table.
Why this holds. Almost nobody has ever answered this question out loud, which is why the answers are so revealing. Diagnosing that the future is uncertain costs nothing. Picturing a specific good one is harder and more useful, because you can start walking toward a picture. The chapter puts a dystopia right next to it, so the vision has to earn its keep. It also works socially, since what someone would do with unlimited free time tells you who they are faster than a job title.
Where it comes from. From the chapter of HUMANLIKE that contrasts an AI dystopia with an AI utopia, at the end of the utopian scenario, "It's a world where being human means exploring, creating, and connecting in ways we once could only dream of, free from the constraints of survival and scarcity." The line arrives as the summary of the good branch, immediately after the dark one has been described in equal detail.
It's a world where being human means exploring, creating, and connecting in ways we once could only dream of, free from the constraints of survival and scarcity.12-dystopia-and-utopia.html
Do this. Describe to one person the version of your life where survival is handled, and ask what theirs looks like.
Idea · Envision · Curiosity
The pronoun changes before anyone decides it should
HUMANLIKE carries a hypothetical testimonial in the voice of a supply chain officer, about a logistics system the company builds and names CLARA. Over four years CLARA gets video access, a face, a voice, jokes, a personality. The narrator notices the language has moved only in hindsight. It was an it, then it was a her, and nobody marked the crossing.
Why this holds. Arguments about machine status get staged as a future debate with a clear moment of decision. The scenario suggests the decision arrives sideways, in grammar, well before anyone convenes to have it. Language tracks how we already treat something, so the pronoun is evidence about the relationship rather than a claim about the machine. Watching your own words is the cheapest instrument available for seeing where that relationship has drifted. It costs nothing and it is hard to fool.
Where it comes from. From the CLARA scenario that opens The Road to AGI chapter of HUMANLIKE, a hypothetical testimonial written in the voice of a Chief Supply Chain Officer. After the video access, the face and the voice have all been described, the narrator says, "It was also interesting to realize that at the start CLARA was an it, but by then, Clara was a she." The capitalisation changes with the pronoun, which is the whole point of the line.
It was also interesting to realize that at the start CLARA was an it, but by then, Clara was a she.11-the-road-to-agi.html
Do this. Notice which pronoun you used last time you spoke about an AI tool out loud, then ask a friend which one they use.
Idea · Envision · Kindness
The premium on human care may not outlast our generation
Course participants get asked what they would value most in a world where health, retail, banking and education all run on AI agents. They answer with the human premium, knowing a flesh-and-bone person made time for you. Then comes the thought that chills. Children growing up beside digital companions may not feel that premium at all, having never known a world where attention was scarce.
Why this holds. The human premium gets discussed as though it were a permanent property of people, safe from automation by definition. The worry here is that it belongs to a generation instead, held in place by what that generation grew up expecting. Value tracks scarcity. If attention from a patient, warm, always-available agent stops being scarce, the thing we currently price highest could quietly reprice. None of that lowers what the premium is worth today, so spend it while it still registers. Whether it fades is a question about the next generation's expectations rather than ours.
Where it comes from. From the Philosophical reflections section of the Shaping the Future chapter of HUMANLIKE, where course participants are described naming the human premium, followed by the admission that it sends a chill to think that "this appreciation for the original human touch might also fade away, especially as the younger ones get used to their new AI companions, digital people, coexisting with humankind and becoming part of society, especially once affective computing can more fully emulate human emotions." The passage names affective computing as the tipping condition.
this appreciation for the original human touch might also fade away, especially as the younger ones get used to their new AI companions, digital people, coexisting with humankind and becoming part of society, especially once affective computing can more fully emulate human emotions.13-shaping-the-future.html
Do this. Ring one person you would normally text. Two minutes of your voice, on purpose.