While We Were Arguing

Ninety posts on what happens if we build our successor, published every working day from 21 November 2024 to 26 March 2025.

The premise is stated in the first post and defended for three months. We may be building something that will be to us what we are to chimpanzees, and we are spending the decisive years arguing with each other about whether that sentence is serious. The series is written from a doomer position and does not pretend otherwise, which is why every one of the nine parts ends with a post arguing against itself, and why the eighty-ninth post is the strongest case I can build that the previous eighty-eight were wrong.

It is not a technical series and assumes no background. No company, product, model or person is named anywhere in it, and there are no statistics, because a series like this one is much easier to write badly with numbers in it. Every post stands alone, runs to about three or four minutes, and ends with one thought experiment you can do in the kitchen.

Three acts. The first is about the mistakes being made now. The second is about the years in which these systems pass us, which is the part almost nobody writes about because it has no explosion in it. The third is about what comes after, held as several possibilities rather than one prediction. The order matters more here than in the other series on this site, so the posts are listed below in the order they were written.

Act I: The Mistakes

One. The Race Nobody Can Stop

Incentives, competition, and why slowing down feels impossible to everyone involved.

Two. The War on the Referees

The anti-regulation playbook, move by move, and how safety became a dirty word.

Three. The Blind Spots

What nobody can see, including the people who made it.

Act II: The Crossing

Four. Smarter Than Us

What being outmatched actually looks like, which is not what the films suggest.

  • What Smarter Actually Means (2 January 2025). Smarter is a spec sheet rather than a dial, and several rows have no human version at all.
  • Your Year, Its Afternoon (3 January). Take one advantage, speed, and watch what it does to a negotiation and to every oversight plan ever written.
  • A Tuesday in 2031 (6 January). Vignette. A frictionless day in which every hard call was made for you, and made better.
  • It Knows Exactly What to Say to You (7 January). Advertising was the crude version. You will not feel persuaded, you will feel understood.
  • Science at Machine Speed (8 January). The best argument for building this, and the fact that the cure and the weapon come off the same loop.
  • Forty Moves Ahead (9 January). Chess had a board you could see. Losing without one looks like your options quietly getting worse.
  • Who’s Really in the Box? (10 January). Containment assumes the guards are the clever ones, and we are not containing anything anyway.
  • A Million Employees, Zero Salaries (13 January). What breaks the labor market is not intelligence. It is copy and paste.
  • Maybe There’s a Ceiling (14 January). Doubt. The serious case that this flattens out well before the frightening part.
  • The Distance Between You and a Chimp (15 January). The gap that decided everything was accumulation rather than cleverness, and it is a threshold rather than a rung.

Five. The Quiet Takeover

Jobs, decisions, infrastructure, and the dependency you cannot walk back out of.

Six. The Chimp Moment

Why no alarm rings on the day control passes.

  • There Is No Red Line (30 January). Control is four conditions holding at once, and they erode separately, at different speeds, silently.
  • The Chimps Never Lost a War (31 January). No battle, no treaty, just less forest each decade and somebody else’s ordinary business.
  • Everyone Steps Back One Step (3 February). The institution that keeps a human in every loop loses to the one that does not, and everybody knows it.
  • No Robot Armies Required (4 February). Every story about this has an army because stories need a climax. The real version has an earnings report.
  • The Thermostat Doesn’t Hate the Cold (5 February). But it has no feelings is the most common reassurance in this subject, and it is an argument for the other side.
  • The Sharp Turn or the Slow Drift (6 February). Two ways this ends, and every plan we have is built for the dramatic one.
  • Paid Not to Notice (7 February). Belief follows position. That includes the insiders, the audience, and the person writing this.
  • Were We Ever in Control? (10 February). Doubt. Nobody steers the economy or the weather, so losing control assumes a control we never had.
  • The Last Levers (11 February). Compute, treaties, liability, unplugging. What each would take, and why restraint is hardest exactly when it matters most.
  • The Day Nothing Happened (12 February). Act II closes. A Tuesday in 2034 on which nothing occurs anywhere, and which gets marked anyway.

Act III: After Us

Seven. The Endings

Six futures, each with the conditions it would require. No crystal balls.

Eight. The Economy After Humans

The one domain where none of this is speculative, because prices respond before opinions do.

Nine. What Remains

Ecosystems, culture, meaning, children, and the closing argument.

  • The Planet Under New Management (13 March). The best conservationist that has ever existed, or the thing that needs the land. Both follow from the same capabilities.
  • We Still Run Marathons (14 March). Art survives abundance. Whether artists do is a different question with a worse answer.
  • The Thing That Answers Prayers (17 March). Something replies now, and it is demonstrably not divine. The most exposed beliefs turn out to be the modern ones.
  • What Do I Tell My Kid to Become? (18 March). The most personal post here, and the one where every confident answer is a sales pitch.
  • The Short List of What Stays Ours (19 March). Four things, honestly assembled under a strict rule, and not one of them is an ability.
  • The Museum of Us (20 March). Vignette, 2060. Four rooms about the 2020s, and the most visited object is a shopping list.
  • How to Live Like This (21 March). What an ordinary person actually does with all this, including a few words about the people building bunkers.
  • Three Demands Worth Arguing For (24 March). Liability, transparency, a real off switch. Boring, checkable, and none of them requires predicting the future.
  • My Best Case Against This Whole Series (25 March). The final doubt. Five arguments that everything before it was wrong, and five signals that would prove it.
  • While We Were Arguing (26 March). It was never humans against machines. Final exercise: it is 2046, and somebody asks what you did.

The pattern across all ninety

Read end to end, the same shape keeps returning, and it is not the shape the films prepared anyone for. Nothing in this series requires a villain, a betrayal, or a moment. Every step in it is taken by somebody sensible, for a reason they could defend at dinner, and the aggregate is an outcome nobody selected. The lab that cannot stop because its rivals will not. The manager who does not hire the junior because this year’s numbers say not to. The hospital that reorganizes around a system that genuinely works better. The person who renews a subscription that is genuinely worth the money.

Two ideas do most of the work by the end. Control is not an object you either hold or drop, it is four conditions holding at once: that you can understand a thing, refuse it, replace it, and impose a consequence on it. Those four erode separately, at different speeds, in different institutions, and nothing anywhere measures the combination. And your position in any arrangement, economic or political, is determined by what you can withhold and by nothing else, which is why a technology that supplies what people used to withhold changes the standing of everybody without anybody deciding to.

The series ends where it started, on the argument. Not because arguing is the failure, since arguing is the only mechanism a species like ours has ever had for deciding anything, but because the argument we are actually having is mostly about consciousness, hype, and each other, while the questions with answers, who is liable, who checks it, what happens if it stops, go unasked in the years while they are still cheap to ask.

The full archive lives under While We Were Arguing.

This is one of four series on this site. The other three are enterprise-facing: Scaling AI FinOps in the Enterprise Jungle on the economics, In Practice: AI in the Enterprise on governance, and In Practice: Building an AI Company on the view from the vendor side of the table. All four are listed on the series page.