Public transcript

Untitled recording

Length: 4 min
Language: English
Text from: the video's captions
Transcribed: 14 October 2026

Read the full transcript Open on YouTube

Notes in English.

Summary

This short recording is an audience Q&A response, likely from Eric Schmidt, covering US-China AI competition, Europe's AI strengths including Mistral, and the long-term role of humans alongside AI. The speaker argues US-China collaboration is unlikely because China fears partnering more than the US does, and that most human functions will eventually be doable by computers. He uses a Waymo thought experiment to argue that AI systems must preserve human freedom and allow exceptions for human needs, or people will revolt.

Key points

The speaker was the first investor in Mistral and notes Mistral cannot raise money like its US competitors.
After five years studying Chinese peer competition, he concluded the Chinese were more afraid of partnering with the US than vice versa.
He believes the US and China must coexist but are highly unlikely to become close partners.
Humans are social animals and some services like medical care will still be delivered by humans.
Most other functions will be doable by computers, though whether we allow that is an open research question.
He uses a Waymo thought experiment: a fully optimized autonomous traffic system must have an exception button for emergencies like a pregnant woman needing a hospital.
If AI systems impinge on freedom of thought, motion, and assembly, humans will fight back, and he would lead that fight.

Questions it answers

00:00Europe and Mistral

What is Europe's position in AI, and what challenges does Mistral face?

The speaker, the first investor in Mistral, says Mistral cannot raise the money its US competitors can, though they are working on solutions. He notes Europe has enormous strength and talent despite being somewhat disorganized.

The speaker was the first investor in Mistral.
Mistral faces a fundraising gap compared to US competitors.

00:43US-China competition vs collaboration

Is US-China collaboration on AI realistic?

After five years studying Chinese peer competition, the speaker found the Chinese were more afraid of partnering with the US than the reverse. He believes the two must coexist but are highly unlikely to become close partners.

It takes two to tango — China was more reluctant to partner.
Coexistence is necessary, but close friendship is unlikely.

01:33Long-term role of humans

What will humans be useful for in the long term with AI?

Humans are social animals who want to be around other humans, and some services like medical care will still be delivered by humans. Most other functions will be doable by computers, though whether we allow that is an open research question.

Human-delivered medical care will persist.
Most other functions will be doable by computers.
Whether we allow computers to take them over is a research question.

02:20The Waymo exception-button thought experiment

Why must AI systems allow exceptions for human emergencies?

In a fully optimized autonomous-traffic New York, a pregnant woman needing the hospital requires an exception button. Without one, it is dominance of humans by a computer and humans will revolt; a system that reasons and adapts to human needs will be accepted.

No exception button means computer dominance and human revolt.
An adaptable system that responds to real emergencies gains acceptance.

03:42Freedom as the deciding line

What determines whether humans accept or fight AI systems?

It depends on whether AI integrates with human experience and needs, and whether it limits or increases freedom. If systems impinge on freedom of thought, motion, and assembly, they will be fought, and the speaker says he will lead that fight.

AI must increase rather than limit human freedom.
The speaker would lead the fight against systems that impinge on freedom.

Notes

Europe and Mistral

The speaker was the first investor in Mistral and describes himself as a big Mistral fan.
Mistral cannot raise the money its US competitors can, and they are working on solutions to that problem.
Europe has enormous strength and talent despite being somewhat disorganized.

US-China Relations

Prompted by Henry's work, the speaker spent about five years looking at Chinese peer competition.
He initially thought closer partnership with China was possible, but discovered the Chinese were more afraid of partnering with the US than the US was with them.
"It takes two to tango" — the two systems must coexist but are highly unlikely to become best buddies for the obvious reasons.

The Long-Term Role of Humans

Humans are social animals and want to be around other humans.
Some things, such as medical care, will still be delivered by humans.
Most other functions will be doable by computers; whether we allow that is a question for research.

The Waymo Thought Experiment

Imagine New York City with only Waymo and competitor autonomous cars, with traffic mathematically optimized for maximum street loading.
If a pregnant woman has an emergency and needs to get to the hospital, is there an exception button in the car?
If there is no exception button, that is dominance of humans by a computer, and humans will revolt and see it as oppression.
If the system instead reasons about the emergency and adapts to get the person to the hospital, it will be more accepted.

Freedom as the Deciding Factor

The key question is whether AI integrates with human experience and needs, and whether it limits or increases freedom.
Preserving freedom of thought, motion, assembly, and gathering is really important.
If AI systems impinge on freedom, they will be fought, and the speaker says he will be leading that fight.

Text from the video's own captions. Summary and notes written by AI from the transcript, so check anything important against the recording.