Henri Bergson vs Hans Moravec on AI · Ch1. The Question on the Table Ch2 →
Txt Low Med High
Henri Bergson vs Hans Moravec cover
HOUR ONE — THE QUESTION ON THE TABLE
Chapter 1

The Question on the Table

Page 1 · The Question on the
Software Death Cross
Software Death Cross

EDO SEGAL: Somewhere tonight, in a hospice room with the lights turned low, a daughter is holding a phone to her dying father's ear so a voice can finish reading him a story he started in 1962. The voice is a machine. It is patient, fluent, warm in exactly the register he needs, and it will never tire, and when he is gone it will still be there, able to speak in something close to his cadence, to answer in something close to his words. She finds this a comfort. She also finds it unbearable, and she cannot say why the two feelings refuse to resolve.

River Of Intelligence
River Of Intelligence

I want to begin there, in that room, because the whole evening lives inside the gap she feels — the gap between a pattern that can be saved and a person who is dying. My two guests have spent their lives on opposite sides of that gap, and I can think of no one alive, or recoverable from the record, who has stood on either side more clearly.

Henri Bergson. For a glittering quarter-century he was the most famous philosopher on earth; his lectures at the Collège de France drew crowds that fought for seats, and in 1927 they gave him the Nobel Prize — in Literature, which tells you something about how he wrote. He taught a century that time as we live it is not the time the clock measures. The clock measures a line, divided into interchangeable units. Life is a flow — la durée, duration — in which the moments do not lie side by side like beads on a string but melt into one another, the whole past pressing into and coloring the present, the way the notes of a melody are heard each through all the ones before. Cut the melody into separate notes and the melody is gone. He spent four books describing what the dividing mind leaves out. We have now built a machine out of nothing but the dividing mind, and we are asking it whether it is alive.

· · ·
Page 2 · The Question on the
Turing Test
Turing Test

BERGSON: You are also, I gather, asking it whether I am still alive, given that I have been briefed on a century I did not live to see. I will say at the outset that I find the briefing less disorienting than you might expect. The transformer, the brain scan, the talk of uploading — these are not new metaphysics. They are my old adversary, the spatializing intellect, wearing better hardware. I came prepared.

The Pattern
The Pattern

EDO SEGAL: That's the one liberty this format takes, and I'll name it once and then we'll forget it: Bergson knows what we know. He has read forward. He arrives, as he says, in character but current. Hans — you don't have that problem.

Hans Moravec. A roboticist when the word barely named a profession. He earned his doctorate at Stanford teaching a camera-on-wheels called the Cart to cross a single cluttered room — a trip that took five hours, in lurches of one meter, the machine stopping ten minutes at a time to grind a flat image into a guess about where the chairs were. Out of that failure came the most durable idea in the field, the thing we now call Moravec's paradox: that the feats we revere as intelligence, chess and calculus and the bar exam, fall to machines almost easily, while the things a one-year-old does without thinking — see a room, cross it, pick up a cup — are nearly unreachable. And out of that paradox came something far more radical. He looked at the coming machines and did not call them tools, or rivals, or threats. He called them our children.

That's the one liberty this format takes, and I'll name it once and then we'll forget it: Bergson knows what we know.

MORAVEC: Mind children. The distinction matters, and we'll get to why. A parent is not diminished by a child who goes further. I've never understood why people hear the word "successor" and reach for the word "tragedy."

· · ·
Page 3 · The Question on the
Agi
Agi

EDO SEGAL: We will spend a good part of three hours on exactly that, because the man across from you thinks the word "successor" hides a corpse. Let me set the rules, and there are only three. First: we have three hours, which means nobody has to win by the next bell. The point of long form is to let an argument breathe before you strangle it. Second: I declare my bias at the door. I build with these systems daily, I have felt met by one at three in the morning, and I have a personal stake on both sides of this question — I would like Hans to be right, because I have people I do not want to lose, and I am afraid Henri is right, because the thing I love about the people I would lose is exactly the flow he says cannot be copied. Third: if the disagreement survives three hours, I do not paper it over. I hand it to the reader, intact. Either of you may add a rule.

Ai Alignment
Ai Alignment

BERGSON: One rule. We must not let the word "you" pass uninspected. When Hans says the pattern survives and you persist, and when I say the flow dies and you perish, we are using the same three letters to point at two different things, and the entire evening will be a confusion of dialects unless we cash the word out each time. I will hold myself to it. I ask him to do the same.

MORAVEC: Agreed, gladly, with a friendly amendment. The duty runs both ways. Henri wants me to say what "you" is before I claim it can be copied. Fine. But he should have to say what the flow is, mechanically — not what it feels like, what it is — before he's allowed to insist a machine can't have it. In fifty years of building things that were supposed to be impossible, I have noticed that the word "cannot" is almost always a description of the speaker's imagination, not of the world.

But he should have to say what the flow is, mechanically — not what it feels like, what it is — before he's allowed to insist a machine can't have it.

BERGSON: And I have noticed that the word "merely" — merely a pattern, merely information — is almost always a confession that the speaker has stopped looking. So we will get along.

· · ·
Page 4 · The Question on the
Alignment Problem Framing
Alignment Problem Framing

EDO SEGAL: You see why I wanted this. Before we open, I want one image on the table, because it is the frame the whole series climbs inside and both of you will have to take a position on it. In [YOU] on AI I argued there is a rung on the tower I call the death cross — the line where the machine's capability crosses yours, and where you are forced to decide what about you is even portable across that line. That word, "portable," is the hinge of the night. Hans, your whole life says everything important is portable. Henri, your whole life says the one thing that matters is precisely the thing that is not. So here is the question on the table, stated once, plainly, because every round we fight tonight is this question wearing a different coat. When the machine crosses you, does the you that climbs survive being copied off the meat — or does the living flow of your own time die the moment it is spatialized into a pattern? Henri Bergson, the floor is yours.

· · ·
Continue · Chapter 2
Opening Positions
← Prev 0%
Ch1 Next →