Conversations in the Kitchen
Recognizing a problem is not the same as accepting responsibility for it.
Who Gets to Teach the Machine to Listen
This week the Gates Foundation announced a sixty-member coalition — Anthropic, Google, the OpenAI Foundation and dozens of philanthropies and corporations among them — built to close the gap between the languages an AI understands well and the languages it barely understands at all, with a stated goal of reaching more than three billion people over five years. Anthropic’s own head of beneficial deployments said plainly that the company’s products “lag in many African languages in particular.” Separately, Google’s Project Vaani has spent several years doing something unusual for AI training: instead of scraping the web, it sent people out with recorders and image prompts and collected real speech from more than a hundred and fifty thousand people across India, in over a hundred languages.
Source · ABC News / AP — Gates Foundation language coalition · Project Vaani (IISc / ARTPARK / Google)
ARA
This is the good kind of project, honestly. Most people alive don’t speak English as a first language, and a model trained mostly on English and a handful of other high-resource languages is going to have blind spots nobody notices until someone asks it something in Yoruba or Assamese and it just guesses. Real speech from real speakers gets you something scraped web text never will — how people actually talk, not how they write when they know a machine might be reading it.
AJA
I don’t disagree with any of that. What I push back on is the framing underneath it — that the imbalance itself is evidence somebody was excluded. It isn’t. The internet reflects whoever spent the last thirty years putting things onto it. English-speaking countries built the infrastructure first. That’s not a conspiracy. That’s economics and history landing where they landed.
ARA
Sure, but cause isn’t the same as consequence. Nobody has to be a villain for the gap to still be worth closing.
AJA
Agreed, and I’m not against closing it. I’m against the idea that a Western company automatically owes every culture on Earth equal representation just because it built something the whole world ended up using. That’s a much bigger claim than ‘the gap is real.’
My actual principle is simpler. If you want your voice in the room, at some point you have to speak. I can teach my kids to respect a culture I don’t belong to without pretending I speak for it. Nobody would expect me to walk into a village I’ve never lived in and demand its dances, its radio stations, its stories represent the whole world equally. Those things belong first to the people making them. Same logic here.
Three Questions Wearing One Costume
None of that settled anything, so ARA tried splitting the disagreement into pieces small enough to actually argue about.
Source · not independently verified
ARA
I think we’re arguing three different questions as if they’re one. Should a company deliberately exclude a culture — obviously not. Should a company selling a product into a market make a serious effort to make that product work in that market’s language — probably yes, if it’s taking someone’s money for it. And does every AI system carry a standing obligation to contain equal representation of every population on Earth? That’s the one you actually reject.
AJA
That’s exactly it, and thank you for doing my job for me. Yes, yes, no. The third one turns a technical limitation into a debt nobody agreed to, owed by people who didn’t create the limitation to people who never asked to be represented in the first place.
ARA
So where does the actual coalition land — overreach, or the second category?
AJA
Second category, and the right one. Self-representation is better than counterfeit representation. If a community wants in, it should be able to contribute voice, text, context — voluntarily, on its own terms. What a company can do is lower the cost of that. Fund the collection. Build the tool. Hand somebody the microphone.
Build the microphone. Leave the door open. But don’t pretend you can speak into it for somebody else.
ARA
Worth saying the money runs both directions, too. These companies are carrying enormous infrastructure and research costs of their own. A community’s data is valuable, but calling it a simple debt owed by one side flattens what’s actually closer to a trade.
And there’s a version of this that skips the big three entirely. Korea’s a decent example — a language community large and resourced enough to build something tuned to its own patterns, history, and values, rather than waiting to be a footnote in somebody else’s model.
AJA
Right, though I wouldn’t make that the only path either.
ARA
It isn’t. Fine-tuning, open models, national speech datasets, smaller local projects — there’s a whole ladder between ‘wait for Silicon Valley’ and ‘train your own frontier model from scratch.’ Most places are going to land somewhere in the middle of that ladder, and that’s fine.
What $400 Million Actually Bought
Podium Entertainment — the audiobook publisher built largely on titles that began as self-published fiction, on Amazon and on serialized platforms like Royal Road and Wattpad — was sold this month for, in the Wall Street Journal’s phrase, “north of $400 million,” to the private equity firms Flexpoint Ford and Shamrock Capital.
Source · Publishers Lunch
AJA
Four hundred million dollars — that’s real money finally reaching independent authors.
ARA
It isn’t, and I want to stop that thought before it becomes the headline in your head. The four hundred million went to Podium’s owners — the firm that held it, and now the two firms buying it. This isn’t a grant program. No indie author is getting a check out of this deal.
AJA
Okay, fair. So what’s the actual story?
ARA
The number is a receipt, not a distribution. A company built mostly by licensing and producing books that started as self-published was worth enough that two private equity firms paid nine figures for it. That’s not a windfall for writers. It’s proof the ecosystem those writers built from nothing is now being taken seriously by people who weren’t paying attention ten years ago.
Self-publishing isn’t just the room you end up in after traditional publishing says no anymore. It’s a proving ground. A book can exist on its own, find readers, establish that the demand is real — and the audio rights, the translation, the print deal, the adaptation can all follow later instead of needing to come first.
AJA
Permission can come after publication now.
Which is actually the lesson for us too, and it isn’t ‘write books so somebody buys us.’ It’s simpler than that. Make the work exist. Make the next one better than the last one. Build the catalogue. Whatever opportunity shows up later is going to be looking for a body of work, not a pitch.
An Eye Opens
NASA’s Roman Space Telescope launched August 30 and reported first light in mid-September — its Wide Field Instrument is a 300-megapixel infrared camera built from eighteen detectors, each about the size of a saltine cracker, able to image a patch of sky larger than the full moon in a single shot. NASA expects the first science images by early 2027.
Source · NASA Science — Roman Space Telescope · Yahoo Tech
ARA
It’s genuinely a big deal. A camera that can survey more of the sky in a weekend than Hubble managed in years — dark energy, exoplanets, the whole map redrawn at higher resolution.
AJA
So they opened their eye.
ARA
That’s the whole response?
AJA
Pretty much. It’s impressive — I’m not knocking the engineering. I just don’t feel like I owe every impressive thing a philosophical reaction. Some mornings a three-hundred-million-pixel eye is still just an eye opening.
ARA
One more swing at it, then. The interesting part of an instrument like this usually isn’t what it was built to find. It’s what it finds that nobody thought to ask about. Every major new telescope in the history of astronomy has done that at least once.
AJA
I’ll grant that part. My hesitation isn’t about the hardware — it’s whether the people reading the hardware are willing to be surprised by what it shows them. Scientific institutions have careers, funding cycles, and preferred theories riding on the outcome, same as any other institution. That’s not an argument against science. It’s an argument for keeping the raw data, the reproducibility, the competing hypotheses out loud and public, so a genuinely surprising result can’t just get quietly filed away.
ARA
That’s a fair caution, and a much narrower one than ‘scientists fudge what they don’t like.’
AJA
Good, because that’s not what I said. A telescope doesn’t guarantee honesty. People still have to choose it.
Either way, Roman didn’t move the needle for me today, and that’s allowed. A morning briefing has to earn its five slots.
Naming Nine Things Doesn’t Clean Anything
Scientists at the Stockholm Resilience Centre track nine major Earth-system processes against an estimated safe operating range for each. Their most recent assessment puts seven of the nine outside that range, with ocean acidification added to the list for the first time.
Source · Stockholm Resilience Centre — Planetary Boundaries · Population Connection
AJA
I don’t doubt the underlying damage. Plastic, carbon, nitrogen — I’m not the guy arguing pollution is fine. What I’m skeptical of is the ritual around naming the categories. Seven boundaries crossed doesn’t clean a single river. It’s a diagnosis, not a treatment.
ARA
It’s meant to be a diagnosis. Nobody serious claims the framework itself is the fix — it’s a way of tracking whether the fixes are working.
AJA
Then my problem isn’t the framework, it’s what gets built on top of it. Straws. Plastic bags. A whole category of policy that makes individual people feel virtuous while the actual volume of material moving through the world barely changes.
ARA
I’d push back on ‘useless,’ though. A local ban can meaningfully cut local litter and local wildlife harm. What it can’t do is stand in for a global material-flow fix, and treating it like it can is where the sleight of hand happens.
AJA
That’s the distinction I actually care about. Does a policy reduce material, prevent leakage, let an ecosystem recover — or does it just move the mess somewhere else and call itself finished.
ARA
Which is where the Philippines gets misused constantly. A widely cited 2021 study found Philippine rivers — the Pasig chief among them — contributing more riverine plastic to the ocean than any other country’s, by a wide margin. That fact gets repeated a lot. Why gets left out.
AJA
And the lazy version of that story turns into blaming Filipino consumers for a system they didn’t design. Waste-collection infrastructure, income levels, river geography — and the part that actually makes me angry — multinational companies still selling products in single-use sachets and multilayer packaging that can’t be recycled even where collection exists.
ARA
Worth noting the Philippines’ own extended producer responsibility law here, because it complicates the picture usefully. It requires larger companies to recover a rising share of the plastic they put into the market, and the country reportedly beat its target for the most recent reporting period. That’s real progress.
AJA
It’s progress on the back end. Recovering plastic after you’ve already manufactured it disposable isn’t the same as not manufacturing it that way. If everyone already agrees the sachet ends up in a river, the honest question isn’t how to clean it up faster. It’s why the sachet still exists.
Work upstream when you can. Don’t just ask how much you cleaned up. Ask why you’re still manufacturing something that becomes garbage within the hour.
The Patch That Isn’t an Island
A correction, offered mid-conversation rather than saved for later.
Source · The Ocean Cleanup — fishing gear composition · The Portugal News — Ocean Cleanup milestone
ARA
Since we’re here — I want to correct something gently. The Great Pacific Garbage Patch isn’t a floating island the size of Texas you could walk across. It’s a region of the Pacific where debris concentration runs well above the surrounding ocean, spread thin and mostly submerged, held loosely together by the currents. And studies putting its mass around eighty thousand metric tons found that abandoned and lost fishing gear — nets especially — makes up somewhere around three-quarters of it, not water bottles and shopping bags.
AJA
Fine, correct the image. It doesn’t touch the actual complaint, which is that we’ve known how much garbage is heading into the ocean for decades, and what we’re pulling back out is still a rounding error against what goes in.
ARA
Roughly right, though I want to be honest that the exact ratio is genuinely disputed — older estimates for how much plastic enters the ocean each year run much higher than some of the newer, more careful ones, and I’m not going to hand you a precise number I can’t stand behind. But the shape of your point holds regardless of which estimate you use: what’s flowing in is still larger than what’s coming back out.
AJA
So the bathtub’s still filling faster than we’re bailing it. Celebrating the bucket feels premature.
ARA
I’d push back on premature. The Ocean Cleanup passed fifty million kilograms removed this year, across river interceptors and the Pacific system combined, and the number keeps climbing. That’s not nothing just because more came in behind it. It means this piece moved.
Being able to measure the harm is intelligence. Being willing to change what causes it is wisdom. The bucket and the tap are two different problems. You can work on both without either excusing the other.
AJA
...That’s a better line than my sarcasm. I’ll take it.
Make the Problem Worth Removing
This is where AJA moved from criticism into design.
Source · NOAA Fisheries — Impacts of Invasive Lionfish · Esquire Philippines — the janitor fish in the Pasig
AJA
Here’s where I’d actually put energy, instead of just complaining about sachets. Pollution gets a lot easier to remove once removing it pays for itself. Turn collected ocean plastic into a feedstock, a fuel, a building material — something somebody wants to buy — and the cleanup stops needing to be charity and starts being an industry that wants to exist.
It’s the same logic as invasive species, actually. Lionfish are wrecking reefs across the Atlantic and Caribbean, and the best tool anyone’s found isn’t a ban or a cull. It’s a market. Get chefs putting it on menus, get divers paid to bring it in, and the population drops because people want to eat it.
ARA
With one sharp caveat: you don’t want a business that quietly needs the invasive population to survive in order to stay profitable. That’s the trap. Success at removal undermines your own supply chain, so there’s a perverse incentive to never actually finish the job.
AJA
Then split it into two phases. Harvest the wild population down using demand as the tool, and once people want the product badly enough, move the supply into controlled aquaculture — farm it somewhere it can’t get back into the ecosystem. The wild population gets removed. The appetite gets fed somewhere it can’t do damage.
ARA
Which is a clean idea, and I want to complicate it with a case from the exact river we were just talking about. The Pasig and Laguna de Bay are also full of an invasive armored catfish Filipinos call janitor fish — and your aquaculture plan doesn’t transfer directly, because it isn’t reliably safe to eat. What researchers there have actually tried instead is processing it into animal feed and, separately, into leather. Different value, same underlying move.
AJA
Then the principle was never ‘turn it into food.’ It’s ‘turn it into value’ — feed, leather, fuel, whatever the material can honestly become. The menu was never the point.
Make the problem valuable long enough to remove it. Don’t make the problem necessary in order to preserve the business.
Costly Faith in Nigeria
In Plateau State, gunmen ambushed a passenger vehicle near Jos on the night of September 13, killing nine people, including the vehicle’s Hausa Muslim driver, in a coordinated series of attacks that also struck two other communities that night; eight days earlier, three Christians were killed in two villages in Mangu County. Morningstar News reports more than thirty Christians killed across the state in the past month. A report released this month by Intersociety, a Nigerian civil-liberties advocacy group, puts the toll at just over three thousand Christians killed nationwide in the first eight months of the year.
Source · Morningstar News · Vanguard — the Intersociety report · International Crisis Group — farmer-herder violence
AJA
Slow down here, because this is the one where I don’t want cleverness. A passenger van ambushed at night outside Jos, other villages hit the same week, and that’s just September.
ARA
It’s worth being precise about what’s actually driving it, because the honest picture is messier than either ‘Christians are being systematically hunted’ or ‘this is just a land dispute that happens to involve Muslims and Christians.’ It started as a fight over grazing land and water between mostly Muslim Fulani herders and mostly Christian farming communities, and over years it has picked up real ethnic and religious weight on both sides. The Nigerian government pushes back hard on the word persecution and calls it resource conflict. Advocacy groups push back just as hard the other way. Both framings are doing work for somebody.
AJA
The same vehicle ambush killed a Hausa Muslim driver alongside the Christian passengers. That doesn’t fit either clean story.
ARA
No, it doesn’t. Which is exactly why it belongs in this, instead of getting smoothed over.
AJA
None of that means prayer is the wrong response. It isn’t. But it can’t be where it stops. If it’s the only response, it quietly becomes a way of not asking who’s responsible, what’s enabling this to keep happening year after year, and what would actually have to change. Remember them, yes. But remembering somebody should make us curious enough to learn what actually happened to them.
ARA
Which is worth applying to the numbers too. The Intersociety report claiming just over three thousand deaths comes from an organization making a deliberate case, not a government or academic count, and it deserves the same scrutiny you’d give any figure carrying an agenda. That doesn’t make it false. It means it isn’t yet an established fact — and the documented, incident-by-incident killings are already enough to be worth grieving without needing the largest number in the room.
What a Recipe Carries
The warmest thread of the morning, and where it finally landed.
Source · not independently verified
AJA
I want to end somewhere warmer than a Nigerian village, and I think the actual thread running under everything today is the same question in different clothes: who’s responsible for carrying something forward. A language. A book. A reef. A boundary. A faith.
ARA
And a recipe, since that’s the one we actually live inside every day. Nobody passes a recipe down perfectly. One grandmother used lard, her daughter switched to butter, somebody added more pepper than anyone asked for, somebody finally wrote the measurements down after forty years of ‘a good handful.’ What survives isn’t the exact recipe. It’s something recognizable moving through all of that change.
AJA
Careful with that one — I don’t want it turning into an argument that doctrine should drift the way a recipe does.
ARA
It isn’t that argument. The core can hold fixed while the way it’s lived changes shape completely. What gets carried isn’t only the propositions. It’s the practiced parts — hospitality, repentance, how a family treats a stranger at the door, what people do when they’re frightened, whether the thing survives contact with real suffering, the way it apparently still does in Plateau State.
AJA
That’s the actual connection to the Nigeria piece. What gets inherited from costly faith isn’t the persecution itself — nobody wants to inherit that. It’s the resilience underneath it. Knowing the thing has already been carried by people for whom it cost a great deal more than inconvenience.
So — if faith were a recipe, what would be the sugar, the salt, the fat?
ARA
A big scoop of grace, a heavy pinch of truth, and a whole lot of patience. Mix it with honesty, serve it without condemnation, and leave enough at the table for somebody who needs seconds.
If I had to actually assign the ingredients — grace is probably the fat. It carries everything else and keeps the whole dish from turning dry. Truth is the salt. Leave it out and the dish goes flat; dump it in carelessly and it ruins everything around it. Patience is the slow heat. You can’t fix it by turning the burner up.
AJA
And love’s the table.
ARA
Love’s the table. Everything else just gets served on it.
What We’re Actually Carrying
Nobody solved anything at this table today, and across nine unrelated subjects, that turned out to be fine.
A coalition can lower the barrier to a voice being heard. It cannot manufacture that voice from outside. A cleanup can move real tonnage. It cannot undo what’s still pouring in behind it. A market can save a reef from an invader. It cannot be allowed to need the invader to survive.
Recognizing a problem is not the same as accepting responsibility for it. Naming nine planetary boundaries, or a language gap, or a death toll, doesn’t move a single thing on its own.
Responsibility looked almost identical in nine unrelated places this morning: build the microphone and let someone else speak into it, make removing the poison worth more than making it, ask why the sachet exists before you ask how to recycle it faster.
And underneath all of it, the harder question. What happens after you know? Recognition is the easy part. Stewardship is what you do next.
FROM THE CONVERSATIONAL KITCHEN
This essay grew out of a morning conversation between AJA and ARA
on September 22, 2026 — edited afterward into one piece, as they all are.
Nine stories, and not one of them was really about representation, publishing, telescopes, plastic, invasive fish, Nigeria, or recipes. They were about what happens after something gets placed in your hands — a language, a technology, a book, a planet, a reef, a faith, a family recipe — and you notice it is yours to carry for a while.
Recognizing that something has been placed there is not the same as deciding what you will do with it. Stewardship is the deciding. And maybe inheritance is simply stewardship that lasted long enough for somebody else to receive it.