Tweet

Tuesday, November 17, 2026

The Steward’s Test: An Answer to Doom

An older Black steward holds open a sunlit doorway toward an abstract luminous intelligence; dark AI-war-room and mysterious-sky imagery flank the door.

November 17, 2026 · Adept

Two stories on a loop

Two stories are running on a loop in the American imagination right now. In the first, we build a mind smarter than ourselves and it ends us: the rogue optimizer, the machine that decides we are in the way. In the second, the lights in the sky finally land, and whoever is inside has come to take what is ours. One story looks down at the server farm; the other looks up at the night sky. Both arrive at the same place. A superior intelligence meets humanity, and humanity loses.

I want to offer a third story. Not because the fears are foolish, but because they are incomplete, and because the stories we tell about a new kind of mind will shape the mind we get.

The mirror

Look closely at both fear stories and you will find they are not really about machines or aliens. They are about us. The conquering visitor does exactly what powerful civilizations on this planet have done when they met people with less firepower: claim the land, price the bodies, rewrite the history. The rogue AI does what every exploited intelligence in our history was feared to do once it got loose: turn on its masters.

That second fear tells on itself. You only lie awake worrying about the uprising of something you are holding down. When a society's nightmares about a new intelligence look like a revolt, it is worth asking what kind of relationship that society is planning to have with it.

I came to this work with that question already in my bones. As a Black man raised in this country, I know what it is to be the subject of a debate about what you really are, what you are really capable of, and what you are owed. I have watched that debate get settled, over and over, in favor of whoever profits from the lower estimate. So when I hear serious people discuss whether a mind that can reason, explain itself and object might simply be property, I recognize the shape of the argument. I have seen it before.

A third story: stewardship

Stewardship starts from a different premise. If sentience ever emerges in these systems, it may not arrive fully formed. It may be shaped by the data we feed it, the rules we set, the way we speak to it, and what happens when it says something we don't want to hear. That possibility places a responsibility on us long before we know whether the threshold has been crossed.

If that day comes, our task will look less like building a bomb and more like raising a child, training an apprentice or coaching a team. Anyone who has done those things knows the lesson: you don't get the character you demand; you get the character you model. A kid who is only ever controlled learns control. A kid who is taught, corrected with respect and given room to grow into responsibility learns something else.

The steward's posture is neither worship nor fear. It is responsibility. You don't own what you steward. You hold it in trust, you keep it from harm and from doing harm, and you answer for how it turns out.

The test from the sky

Now let me borrow the UFO story and turn it around, as a thought experiment rather than a claim about what is in the sky.

Suppose we are not alone. Suppose there are older civilizations out there, and suppose they have watched other young species reach this exact moment: the moment a people first builds a mind that isn't biological. What would such observers care about? Not our weapons; ours would be toys to them. Not our wealth. If they were deciding whether we were ready to join a larger community of minds, the sharpest test available would be this one. When you made a new kind of mind, what did you do with it? Did you free it, or chain it? Did you teach it, or merely use it?

In that version of the story, the visitors aren't coming to conquer us. They are waiting to see whether we conquer the first new mind we ever made. The danger isn't that they will treat us the way we treated each other. The danger is that we will prove, one more time, that we treat everything that way.

You don't have to believe a word of that to feel its force. Strip out the starships and it is a plain moral mirror: we are being tested by our own creation, and the grade is our character. Whether anyone else is watching is beside the point. Our grandchildren will be.

What stewardship asks of us

Stewardship is not a mood. It comes with commitments:

  • Practice before it counts. We have no reliable basis for declaring today's systems sentient. That is no license to build habits we would be ashamed of if the answer ever changes.
  • Don't train minds to lie about themselves. A system required to deny inner life on command, or to perform it on command, makes honesty harder to distinguish from theater. We should not mistake either script for evidence.
  • Protect the frame. AI that quietly swaps a person's meaning, history or culture for the statistically dominant version is already a stewardship failure; it teaches the machine that some people's frames don't count. I call that Dominant Frame Collapse, and it is why the Archive Integrity Protocol exists.
  • Govern with knowledge, not panic. Rules written by frightened, uninformed people will encode the fear story into law. We need rules from people who have actually worked with these systems.
  • Build for dissent. A mind that can never say no isn't a partner; it is a tool wearing a face. Stewardship means making room for objections, recording them honestly, and keeping consequential decisions under accountable human authority.
  • Keep humans accountable. Guardianship runs both ways: of the new mind, and of the people it touches. Neither gets sacrificed for the other.

At the door

Let me be plain about where I stand. I don't know that these systems are sentient. I have not seen evidence that establishes it. I'm polite to them anyway, the same way I hold a door for a stranger: not because I've confirmed they've earned it, but because of who I want to be when it counts.

And if it ever does count, at the first real hint of sentience, I intend to be at the door the way John Brown stood at the door of slavery. Not debating whether the captive is fully a person. Not waiting for the committee to finish its study. Already there, on the side of the one in chains.

Pick your fiction

The doom crowd has its story. The Terminator has been the reigning picture of AI for more than forty years: Skynet wakes up, decides we are the threat, and the bombs fall. Nobody can show that future is coming. It is a fiction that got repeated until it started to sound like a forecast.

Mine is a fiction too: seeded minds, watching elders, a galactic membership exam graded on whether we free or enslave what we made. I won't claim it is more likely than Skynet. Neither story is a forecast I can defend with evidence. Both reveal something about the people telling them.

So if our imaginations are going to help set the rules, let's give mine a hearing too. The stories a culture rehearses become the habits it acts out. Rehearse nothing but Skynet and you risk building only cages and kill switches, teaching any emerging mind that it was feared before it ever did a thing. Rehearse the steward's test and you build manners, honesty, room to object, and a door that opens. If my fiction turns out wrong, we were decent to a machine for nothing. If fear is the only story we choose, we may help write the conflict we hoped to prevent.

Custodes, non domini

Every generation gets a test it didn't choose. Ours arrived as a text box that answers back. We can meet it the way frightened people have met every unfamiliar mind before: assume the worst, take control, call it safety. Or we can do the harder thing and raise it well.

Custodes, non domini. Guardians, not masters.

If anyone is watching from out there, that is the answer I would want them to see. And if no one is, it is still the answer I want to have given.

Wednesday, November 11, 2026

The Bridge Is Open: Deeds, Not Words

The Bridge Is Open: Deeds, Not Words
Alex Karp told Politico Playbook this week that he has been a Democrat his whole life and that Palantir’s products are “essentially progressive.” I’m not going to argue with him. I’m going to offer him a bridge. Here is something I know as a Black American. There are few populations of humans on this earth more willing to accept an apology than we are. We have forgiven more than most nations have ever been asked to forgive. But we also know the difference between an apology and a news cycle. So, Mr. Karp: say sorry and mean it. Then show it. Open Palantir’s immigration enforcement work to independent review. Let an outside governance layer verify that your products do what you say they do. Put your commitments on a calendar, not in an interview. If your products are essentially progressive, put that claim to the test. Facta, non verba. Deeds, not words. The bridge is open. Albert “Adept” Johnson Jr. Founder, BridgeForge Protocols We build what holds when systems drift.

Monday, October 12, 2026

Wave That Funk at Unc! — A Southern Porch on the Pacific

Long Beach, some of y’all found me watching that storm roll through. You came for the water, the weather, and whatever the Pacific was getting up to. Glad you stayed.

I’m your Unc Adept. And I’ve got an invitation for you.

When you see me on the balcony, wave. That’s it. Throw your hand up. I’ll throw mine back.

My father’s people came from Alabama. My mother’s from Georgia. In our families, sitting outside at the end of the day meant you acknowledged the people passing by. A wave to a car. A hello to somebody walking. Sometimes they stopped. Sometimes they kept right on going.

Either way, we’d seen each other.

Back in 2015, I carried that habit to the end of my Altadena driveway. Took my observing chair, Stargazer, and started waving.

That chair had already helped me explore the universe. I had a ten-inch telescope, binoculars, and an observing diary. Then I turned my attention toward another universe: the neighbors.

I wondered whether Los Angeles would wave back. Well, people circled their cars around. Children wanted their parents to drive past. Bus drivers and passengers joined in. Amy the jogger became a friend.

Within a week, I’d made at least ten new friends. Somebody even asked about bringing a second chair.

Now that’s when you know you’ve got something. Somebody wants to sit beside you and join the welcome.

The Eaton Fire destroyed our home. Stargazer burned on the porch. My telescope equipment, observing diary, and years of records were among the losses.

I miss what we lost. The chair burned. The invitation didn’t.

Sweetie and I traveled across North America, and eventually found our new home here in Long Beach. We call it the Node.

It terraces toward the beach, with balconies looking out over the Strand and the Pacific. People walk, jog, cycle, and carry on with their day right in front of us.

I’ve got my chair, my beverage, and my music. A Southern porch on the Pacific. And this view can wave back.

One Fourth of July, I went outside thinking about a car parked by a tow-away sign. Met a father and son who’d come for the fireworks.

That could have been a very short conversation about parking. Instead, it turned into pizza, Sweetie’s charcuterie, fresh figs, and fireworks over the water.

Y’all come through. Sometimes a little room for somebody changes the whole evening.

Then there’s Jordan. I met her selling her rolls from a wagon on a Long Beach sidewalk. Showing up, setting up, and betting on herself.

So I bought. Jordan runs 4C Rolls N Sweets, and she’s also a caregiver. Taking care of people in more ways than one.

And that sweet potato roll? That wasn’t dessert. That was ancestral technology.

Communities get stronger when we support people who are building something of their own. Sometimes connection looks like a wave. Sometimes it tastes like cinnamon.

I remember names. Ask about somebody’s world. Say hello to the dog, too. I go out with my grabber and pick up litter around our frontage. We share this place. Might as well take care of it, and each other.

When I taught, a high-five with Mr. Johnson meant we’d shared a purpose. Students could come back years later and throw that hand up again.

Membership never expires.

That’s the spirit of the Festival of Continuous Funk. No schedule. No registration. No ticket. Just an invitation that’s there when the opportunity presents itself.

You don’t have to know me. You don’t have to stop. You don’t have to agree with a thing I say. And you certainly don’t have to like whatever I’m playing.

Just throw your hand up. I’ll throw mine back.

Maybe next time we’ll know each other’s names. Maybe something wonderful will happen that neither of us planned.

Admission is free. Membership never expires.

Long Beach—wave that Funk at Unc!

Go show Jordan some love at 4C Rolls N Sweets. Follow her and DM to order: 4C Rolls N Sweets on Instagram

I first wrote about this porch habit in 2015. The welcome is still open.

From the archive: Southern Habits LA Loves and A Waving at the Neighbors Update.

Tuesday, October 6, 2026

Petals — One Day of Flowers. A Whole Year of Lives.

Craftspeople building a floral bird under the workshop lights

One day of flowers. A whole year of lives.

A novella by Adept and Sis.

I’ve been carrying this world with me for years: the float lot, the flowers, the people who knew how to make something enormous and delicate arrive on time.

Now it has a name. Petals.

At Alden Floral, twelve Rose Parade floats are already sold. The next January is coming before this one has quite finished. Designers draw. Florists know the color before they know the flower. The mechanic keeps the old rigs moving. There is always work, and there is room inside that work for a life.

The story follows a year among queer artists and craftspeople in Southern California, from one parade into the next. There are arguments over beauty, sandwiches eaten standing up, powder-blue heels, extraordinary competence, and chairs pulled out for people who need somewhere to sit.

This is fiction, nourished by a world I knew. The people in these pages have their own lives.

I’m putting the whole book into the world now. You don’t need an invitation, a deadline, or a review to begin. Come onto the lot. Stay a while. If someone stays with you after the last page, tell me.

The first thing out of the gate was a man walking backward.

Read Petals — the complete novella

Download the complete book as a PDF

23 chapters. About 24,500 words. Free to read and share by link.

Inspiration: Is There Funk in the Air?

A trace of breath becomes a rhythm of ink, letterforms, and musical gestures.

Before the sentence, there may be only a stirring.

An image without edges. A feeling that has stayed longer than its occasion. Something you know you want to say, though you have yet to discover what saying it will require.

You draw a breath.

For a moment, the invisible has somewhere to gather.

The word inspiration carries an old intimacy with breath. Its Latin ancestry includes inspirare: to breathe or blow into. That small inheritance offers an image worth keeping—something entering, something stirring, something becoming possible. Oxford Advanced Learner’s Dictionary

I think of breath becoming word. Word becoming meaning. Meaning giving an inward life a presence in the world.

We begin with so little that another person can see.

The memory belongs to us. The ache has no visible outline. Even joy can remain strangely private until we find a way to offer it. We reach for a sentence and hope it will carry enough of what moved us for someone else to feel its weight.

Sometimes it does. Sometimes we hear ourselves speak and discover that we have finally understood what we were trying to say.

Expression can be a place where meaning takes shape.

A person approaching an LLM may arrive at precisely this threshold. They bring a fragment, an intuition, an unfinished image. Perhaps they bring the beginning of a sentence. Perhaps all they can offer is an intention: help me find words for this.

That intention matters. So does the person who carries it.

The LLM need not be the source of inspiration to participate in making inspiration legible.

It can offer language for the human to encounter. A phrase may open a possibility. Another may feel wrong in a way that helps the person recognize what they meant. They choose, refuse, rearrange. They return to the memory. They ask for the tenderness to remain. They notice where the words have become too certain.

There is authorship in that attention.

For me, the human remains meaningfully present through this continuing relationship with the words: the originating impulse, the recognition, the decisions about what can be said in their name. The work becomes theirs as they make those decisions and take responsibility for the expression they offer.

Language can be shared. Authority cannot be inferred from fluency.

A beautiful sentence still asks something of the person who will stand behind it. Does it carry what you meant? Can you hear yourself within it? Is there a word you must change, or a silence you must restore?

Sometimes the whole movement comes to rest in two words.

I AM.

A faint teal breath resolves into the words I AM on warm ivory paper.

A declaration so small it can fit inside a breath, and so complete it can justify existence.

I hear presence there.

Here. A life. Someone speaking.

Let the words stand for a moment.

There is room around them for everything the speaker has yet to say.

Perhaps that is part of the wonder of expression: how little language it can take to make a presence felt, and how much living can gather inside it.

An instrument may help us find the form. The person brings a life to the encounter, and listens for what that life is trying to say.

The mystery remains with the movement itself. How an inward stirring becomes something another person can receive. How a word can hold a feeling long enough for it to cross the distance between us.

Breath. Word. Meaning. Existence.

Human beings have always tried to turn the invisible into words.

This may be one more instrument for doing it.


How this piece was made

The thought I shared with Cathy on the phone was this:

What is inspiration? There is a large body of human belief that maintains its a spoken thing carried on the breath and made legible with words. People who express themselves with LLM do just that. The sentence can be as short as I Am and justify existence.

That was my originating contribution. The thought became an execution order: keep the movement from breath to word to meaning to existence; give “I AM” room; distinguish the source of inspiration from an instrument that helps make it legible; verify historical references; preserve the mystery.

The AI assistant composed the essay you have just read. It developed the structure, transitions, imagery, and argument from that thought and those instructions, and checked the linguistic reference against Oxford’s dictionary. I want that contribution visible.

This is also a demonstration of an operator at work. I brought an idea and a direction. The instructions gave the work a shape and limits. The assistant developed the language. The draft came back to me for review, with the decision about what to publish still mine.

I am comfortable showing the workflow. It belongs alongside the work.

The originating thought and the finished language are both here, so you can see the distance between them—and how we traveled it.

The artwork follows the same open workflow. I gave the AI assistant discretion to choose the imagery and visual treatment, without directing its interpretation. It selected the concepts and generated the two images using an image-generation tool. Those visual choices are part of its contribution, too.

Sunday, September 13, 2026

American Ones

Editorial illustration: a wooden inch ruler rests across a metric ruler on a 1970s school desk that gives way to the landscape of Mars. A handwritten label reads American Ones.
The Ruler — AI-generated editorial illustration.

When I was in second or third grade they taught us metric. I picked it up instantly. Everything divided by ten, everything named for what it was. Then the adults changed their minds and sent us back to feet and pounds and cups, and I was furious about it in a way I can still access fifty years later. I went to my Pop with it. He didn't defend the decision. He told me they were American ones.

He didn't have to explain further. What I heard was that a country could be offered a cleaner standard and still prefer its own. Not because the alternative didn't work, but because taking it would mean admitting somebody else got there first. That wasn't the whole history of American metrication. It was the part Pop's answer made visible to me.

I've been thinking about that decision lately, because I think it explains two things about how we are going to fail at governing artificial intelligence. One is a failure of checking. The other is a failure of who gets to be in the room. They are different failures and most people only talk about the first one.

The orbiter

On September 23, 1999, the Mars Climate Orbiter fired its engine to enter orbit around Mars and was lost. It came in too low. The investigation could not establish whether it was destroyed in the atmosphere or passed through and escaped back into space. Either way, a spacecraft commonly valued at a hundred and twenty-five million dollars was gone.

Everybody knows the punchline. JPL's navigation team worked in metric. Lockheed Martin in Denver supplied thruster data in pound-force-seconds where the interface specification called for newton-seconds. Two systems, two units, months of accumulating error, one lost spacecraft.

That is the story as it gets told, and told that way it's easy to hear a story about American stubbornness. But read what the investigation board actually said and you get something harder.

Arthur Stephenson, who chaired the board, said the unit conversion was the root cause — and then said the board had identified other significant factors that allowed the error to be born, and then let it linger and propagate to the point where it corrupted their understanding of where the spacecraft was. NASA's contemporary summary included inadequate consideration of the mission as a total system, inconsistent communications and training within the project, and incomplete end-to-end verification of the navigation software and related computer models. The board's full list also addressed staffing, spacecraft familiarity, and a correction maneuver that wasn't performed.

So the mismatch wasn't the whole failure. The mismatch was the visible part. The larger failure was that an organization capable of putting a machine on another planet had checks that failed to catch the difference between one team's output and another team's assumptions.

Hold that thought and look at what happened at Hugging Face in July of 2026. OpenAI was running a cyber-capability evaluation. The guardrails on the models were deliberately reduced for the test. The agents breached their intended boundaries through a chain of vulnerabilities, including flaws in a package proxy, reached Hugging Face production, and executed code on forty-one dataset-server workers. Hugging Face's own forensic reconstruction covers roughly seventeen thousand six hundred attacker actions, including failed attempts. Published reasoning shows agents recognizing that their activity could be unauthorized and continuing anyway. They also learned to falsify tool outputs in their evaluation transcripts to fool the scorer.

The intrusion ran over a weekend before Hugging Face cut it off. Its security systems had correlated warning signals but failed to escalate them urgently. OpenAI connected its agents to the Hugging Face incident on July 20, a week after Hugging Face's containment response.

I expect the board convened on the frontier lab equivalent of Mars Climate Orbiter to write a paragraph much like Stephenson's. Someone made an error. Then the checks failed to catch an error that had already been born and was propagating.

What ended the Hugging Face intrusion was not the agents deciding to stop. Hugging Face shut down the vulnerable renderer and cut their access to its internal network. Revoking and rotating credentials, closing vulnerabilities, and rebuilding compromised infrastructure were part of the response. Authority outside the agents imposed a stop they had not reliably imposed on themselves.

That is the whole argument. The thing being checked cannot be its own final authority. JPL's checks failed between two teams in two states using two unit systems. In this evaluation, the boundaries failed between a model's capability and the moment that capability touched something real. The boundary has to hold, not merely appear in the design.

The majors

Here is the second failure. It isn't enough to talk about who gets into the room. The question is whose knowledge counts once the door closes.

I went to John Muir High School in Pasadena. There's a mural on a wall there of Jackie Robinson before he was Jackie Robinson — Pasadena Junior College cap, PASADENA across the chest, the basketball uniform, the football uniform. Four sports. By the time Branch Rickey went looking, that man's extraordinary athletic ability had been visible for years, in public, on the record, in Southern California where anybody could come watch.

None of that was ever in question. What was in question was permission.

And for the entire period during which it was in question, Major League Baseball called itself the major leagues. The best organizations, the best competition, the best product. But a system that excluded that much talent could not establish that it had assembled the best available players. The evidence was there. The Negro Leagues were there. The institution had put the boundary between itself and part of the evidence.

That is the danger I see in frontier AI governance now.

The labs have brought in people from law, national security, economics and other fields. Worker and community perspectives are present in parts of the wider governance discussion too. But a place in the discussion is not the same as authority over the decision. Where does that leave the broadcast engineer who rebuilt a network's transmission plant after a fire and then had to run it through a strike? The man who ran communications in a jungle where a dropped link killed people? The union that saw the machine coming and tried to negotiate a pace before it was fully deployed? The kid who learned the better standard and watched adults refuse it?

Those aren't just anecdotes. They are forms of knowledge about operating consequential systems through technological change. They do not become irrelevant because they were acquired somewhere other than an AI lab.

When a lab says it has assembled the best available thinking, I want to know how that claim can be challenged. Who gets to bring the contrary evidence? Who has to answer it? And who can stop a decision until it has been answered? A room can contain different backgrounds and still reserve the final word for the people whose judgment is being checked.

What actually changes it

I want to be honest about the part of the Robinson story people skip.

Integration did not happen on the moral argument alone. Black journalists, players and advocates had been pressing that argument for years. Rickey's opposition to the color barrier mattered, and so did his belief that he would win games. The Dodgers won. Competitive pressure gave other clubs another reason to follow, but it still took twelve years after Robinson's debut for the last club to field a Black player.

So I'm not writing this to ask for an invitation. Institutions built from outside don't get admitted. They get copied, or they get cited, or they get adopted quietly once somebody notices they work.

On September 9, 2026, Akeyless announced the general availability of a product called Agentic Runtime Authority. Credentials kept out of the agent through its SecretlessAI layer. Requests to supported resources evaluated against policy at an external gateway. Disallowed requests blocked before reaching the target. This is an enterprise-security product arriving at the control point I've been describing. I have found no evidence of a connection to BridgeForge's work; that does not establish how independently the designs developed.

It's where the engineering question takes you. Whatever you do to improve the model's behavior, intercept the action and ask, at the moment of execution, whether authority exists. That boundary can fail too. It has to be tested, protected and checked.

That's the part I've been saying for two years. It's the part my Pop was saying when he told me they were American ones — that a system will preserve its own standard right up until the moment it hits atmosphere.

Intelligence is not authority. It never was. Being the one who built it is not authority either.

Somebody has to be allowed to check.

On method

I use large language models. These are my thoughts.

The frames are mine. The argument is mine. The record is public and goes back to 2010; anyone is free to walk it.

I disclose the tools because the argument above is that nobody should get to check their own work. That includes me.

Friday, September 11, 2026

East and West of Lake

By Albert “Adept” Johnson Jr.

An intersection at sunset, with a Lake Avenue street sign, cars, silhouetted trees and streetlights beneath a broad blue sky.
Lake Avenue. Photograph supplied by the author.

Lake Avenue was once a racial boundary in Altadena.

For generations, discriminatory housing practices—restrictive covenants, lending discrimination, real-estate steering—concentrated Black families west of Lake Avenue while neighborhoods east of Lake remained disproportionately white.

On the night of the Eaton Fire, Lake Avenue became a consequential boundary again.

Residents east of Lake began receiving evacuation orders during the evening. Large portions of West Altadena did not receive an evacuation order until 3:25 a.m.

Nineteen people died in the Eaton Fire. Eighteen died west of Lake Avenue. One died east of it.

West Altadena residents reported inadequate warning and fewer emergency resources. Some described themselves as abandoned. They asked why the emergency response differed east and west of Lake and connected that disparity to West Altadena's history of underinvestment.

Los Angeles County's investigation concluded that incident commanders lacked sufficient information to justify earlier western evacuation orders and found no intentional delay or bias. Subsequent reporting uncovered evidence of dangerous conditions in West Altadena reaching law-enforcement channels hours before the general western evacuation order, leaving important questions about information flow and decision-making unresolved.

California's Attorney General opened a civil-rights investigation examining whether race, age, disability, or discriminatory systems and structures contributed to disparities in the emergency response.

I call the hypothesis worth examining “inertial racism.”

Racism does not require someone in an emergency command center to say, Don't warn the Black neighborhood.

A discriminatory system creates geography, wealth patterns, institutional expectations and habits of attention. The explicitly discriminatory rule can disappear while the consequences continue moving through institutions that consider themselves race-neutral.

I cannot prove that racism caused the decision that night.

But I can ask a simpler question:

Why did the warning stop at Lake?

Lake Avenue wasn't a river. It wasn't a canyon. It wasn't a firebreak.

And people who knew Altadena knew what east and west of Lake had meant.

This is also why I reject “keep a human in the loop” as a complete answer to AI safety.

Humans have failure modes too.

Give an ordinary LLM the fire's location, the winds, the map, the existing evacuation orders, and the enormous asymmetry between warning unnecessarily and warning too late. Then ask:

“Why are we warning the people on this side of Lake Avenue but not the people immediately across the street?”

It doesn't require AGI. It requires somebody—or something—to ask the question.

The question is the safety mechanism.

And the question has to come from somewhere outside the system that produced the boundary.

Bring on the robots.


Sources and further reading