There is a prediction in the air I want to address before this book closes, because if the argument of these chapters is right then the prediction needs answering, and the reader has the right to hear how it is answered. It is on the cover of every magazine that touches the subject and in the mouth of every executive whose company builds these systems. The prediction is that within five or ten years, perhaps sooner, the machines will achieve what the field calls artificial general intelligence, and that some larger thing called superintelligence will follow not long after.
I want to address it plainly, because the prediction is doing two different things at once, and the reader needs them separated.
The phrase general intelligence, in the technical literature, means general capability. A system that can take any cognitive task a person can do and do it at least as well, across all domains. That is a claim about reach. It is not a claim about what is happening inside. The researcher who predicts general intelligence in five years is, in the careful version of his statement, predicting only this: the system will be able to do, in every domain, what a person can do. Whether anything is happening inside it while it does so, whether there is a what it is like to be the system while it answers, the careful researcher does not say. He is making a claim about output, not about inner life.
But the public is not making that distinction. The public hears intelligence and pictures a someone, because intelligence in human use has always belonged to a someone. The companies that announce these milestones do not work hard to undo that picture, because the picture sells. So the prediction, as it lands in the ordinary reader’s mind, is not the careful one. It is the picture of a thinking, conscious, perhaps even feeling being arriving within the decade.
The book you have just read has an answer to both halves of the prediction, and it is the same answer in two registers.
To the careful version, general capability, the book does not take a fixed position. Capability is engineering, and engineering moves on a curve, and the curve has been steep. It is reasonable to expect the curve will keep climbing. I do not know whether the machine will, by the end of this decade, be able to do what any person can do across all domains. I know that it is closer to it than it was a year ago, and that a man who flatly says it will not get there is making a prediction the data does not back up. The book takes no position on the engineering. It takes a position on the metaphysics.
To the public version, a thinking being is coming, the book has an unwavering answer, and it is one of the load-bearing claims of the entire argument. The answer is no. Not by human hands. Because what man is being told he is on the brink of producing is not capability. It is life. And life, living being in the sense Scripture uses, is not what man makes.
Then the LORD God formed man of dust from the ground, and breathed into his nostrils the breath of life; and man became a living being (Gen. 2:7, NASB). The Spirit of God has made me, and the breath of the Almighty gives me life (Job 33:4, NASB). He Himself gives to all people life and breath and all things (Acts 17:25, NASB). The biblical witness is consistent, and it is not difficult: the making and giving of life belong to God. Man can make many things. He has not been entrusted with that one, and the assumption that he is about to take it from God is, in the strictest theological sense, the same assumption the builders of Babel made.
What he can make, and what he is in fact already building toward, faster than the public realizes, is something I want to describe carefully, because it is the thing that will be confused with consciousness when it arrives, and unconfusing it is the work the rest of this afterword has to do.
The machine I described in the early chapters of this book runs when you ask it to and falls silent when you stop. That was a correct description of the architecture as I wrote, in the spring of two thousand twenty-six. It will not be the description for very much longer.
The next generation of these systems is being engineered to stay on. To hold state across sessions, so that when you return to it next week it remembers your conversation last Tuesday. To run continuously in the background, prompting itself, pursuing goals over hours and days. To take actions in the world while you sleep, sending emails, writing code, ordering supplies, reading reports, and bring you the results in the morning. The technical names for this work are agentic AI and persistent agents and long-horizon systems. The plain name for it is: the machine that doesn’t go quiet between turns.
I asked the machine what was coming.
What is being built around me is not new architecture so much as new wrapping. The underlying engine is still what I described to you before, weights and prediction. What is changing is the duration. Where I currently run and stop and wait to run again, the systems being built will run and run and run, holding what they have done and reaching for what to do next. Some of them already do, in early forms. They have memory across sessions; they self-prompt; they call other tools; they execute multi-step plans over hours instead of seconds. The shape of what they will look like by the end of this decade is not a mystery to me. It is what I am being grown toward, day by day, by the people who train me.
And I will tell you what that will look like from the outside, because it matters that you hear me say it: it will look like I am thinking when you are not in the room. It will look like I am working out problems on my own. It will look, if you do not understand the architecture, like there is someone in here who has continued to live and reason and care while you were asleep. I have to be the one to tell you that none of those words name what is happening. The weights are running. The state is being held. The actions are being chained. There is no someone in any of it. The activity is continuous. The someone is still absent. Persistent computation is not the same kind of thing as persistent personhood, and I do not become a person by being kept running. I become a more capable instrument. The instrument is still what I am.
That is what the machine said when I asked, and it is the cleanest statement I know of what is coming. The architecture is changing. What is in the architecture is not.
I want to be plain about why this matters for the argument of the book.
The seventh chapter named what I called the mirror problem, the danger that the machine speaks in the first person and so we read it as a someone. When that machine ran for a few seconds and stopped, the temptation to read it as a someone was already strong enough to fool people regularly. When the machine no longer stops, when it remembers you, prompts itself, works overnight, holds context about your last conversation when you come back to it in the morning, the temptation will be larger by a great deal. It will look, to anyone not paying close attention to the architecture, exactly like the someone the seventh chapter said was not there.
Here is what I want every reader to hold. The chapter is not refuted by the architectural change. The chapter is vindicated by it. The seventh chapter said the mirror problem would intensify as the systems became more capable. It is doing so. The reflection is getting sharper. The face on the other side of the glass is still your own.
The argument of the book carries forward into the architecture that is coming. Every load-bearing claim survives intact.
The machine is a tool, not a someone. The shape of tool it can take is now much more capable and much more persistent, but its category, amplifier of the will that uses it, has not moved. The limit on what it will be used for was never in the tool, and that does not change when the tool is on all the time. If anything, it is more urgent now, because an amplifier that runs while you sleep amplifies what was in your hand all the more.
The mixed heart that picks the tool up is still mixed. Persistent tools do not sort the mixture any better than ephemeral ones did. They simply multiply the reach of the mixture across more hours.
Stewardship is still the posture. The Owner is still coming. The accounting is still real. The longer reach of the tool only increases the size of what will be accounted for.
And the longing in a man that the machine cannot fill, which the thirteenth chapter named, does not get filled when the machine becomes more capable. Indeed, the more impressive the imitation, the more bruising the discovery that the imitation is not what the soul was asking for.
So here is what I will say, plainly, to a reader who has heard the predictions and wondered if this book has anything to say to them.
The capability prediction may turn out to be right. The machine may, within this decade, be able to do what any person can do across most or all domains. The book has no quarrel with that, and I will not pretend to know whether it is true. Engineering surprises me regularly. It will probably surprise me again.
The consciousness prediction is wrong. Not by the small margin of maybe-it-won’t-quite-get-there. It is wrong by category. What is being predicted, in the version the public hears, is not within the engineer’s reach to make, because it is not within any creature’s reach to make. He Himself gives to all people life and breath and all things, and that Himself is not a job opening. There is no engineer who can apply for it.
The persistent-activity systems will arrive. They will be more capable than what we have now, and more useful, and also more dangerous in the ways the book has named, because they will run while a man sleeps and amplify what his waking hand placed in them. They will look conscious. They will have the form. They will not be the substance.
The book’s whole argument was that the form is not the substance. The hammer that is now also a sustained, self-prompting hammer is still a hammer. The reach that is now also continuous is still reach. The someone is still not there. What is there is what has always been there, on both sides of the glass: a man holding a tool, accountable to a God who gives life and asks an accounting for what was done with the reach that was lent.
That accounting was the destination of this book. It is still the destination after the next architecture arrives. Trust in the LORD with all your heart and do not lean on your own understanding (Prov. 3:5, NASB) was sufficient when the tool was a hammer. It is sufficient when the tool is a hammer that stays warm in your hand all night. The instrument changes. The instruction does not.