Thread by @peterhartree

Peter Thiel explains how @ESYudkowsky & @MIRIBerkeley went from transhumanism to luddite despair, in the course of 15 years.

From alignment research to death with dignity...

1/15

Thiel: "Nick Bostrom, he's an Oxford academic... most of these people are interesting because they have nothing to say. They're just mouthpieces, it's like the mouth of Sauron... they're useful because they tell us where the zeitgeist is."

2/15

"The Vulnerable World Hypothesis (2019) goes through a litany of these different ways that science & technology are creating all these dangers for the world. What do we do about them? It's the precautionary principle, whatever that means..."

radiobostrom.com/1/the-vulnerab…

3/15

"Bostrom has a 4 part programme for achieving stabilisation.

We 'only' need to do these four things:

1. Restrict technological development..."

4/15

"2. Ensure that there does not exist a large population of actors representing a wide and recognisably human distribution of motives..."

[wry smile from Thiel; audience laughter]

5/15

"3. Establish extremely effective preventative policing."

"4. Establish effective global governance."

6/15

"I would return to Classical Liberalism: however dangerous science & technology are, it seems to me that totalitarianism is far more dangerous."

"1 Thessalonians 5:3—the political slogan of the Antichrist is: peace & safety."

7/15

Again, Thiel is one of the better critics of Oxford EA.

That said: Bostrom does not underestimate the risk of totalitarianism.

The VWH paper does not express an all-things-considered view on the topic. In that sense, Thiel is right that Bostrom has "nothing to say".

8/15

So Bostrom offers definite pessimism all the way down? No.

Bostrom suggests a strategy: differential progress.

We can't stop progress in science and technology across the board. But we can speed up safety-promoting ideas & technology, relative to the rest.

9/15

We can never be sure which technologies are safety-promoting, and which are not.

But: we can probably do better than chance.

Gain of function research? Democratise CRISPR? Maybe not.

cold-takes.com/the-track-reco…

10/15

And wait: might there be ways to reliably prevent highly illegal actions that would *not* come bundled with unacceptable risk of totalitarianism?

This question seems understudied.

I suspect the topic is neglected for understandable, but bad reasons (taboo => stigma).

11/15

Robin Hanson thinks we already have something close to a world government.

We have the IAEA, IPCC & Montreal Protocol—all are sort of working, maybe, without obviously unacceptable downsides.

Maybe we can muddle through on AI & biosecurity too.

https://twitter.com/robinhanson/status/1308511754968690688

12/15

Thiel was speaking at the Stanford School of Business Academic Freedom Conference, November 4, 2022.

Full video here:
drive.google.com/file/d/1b2PpPR…

Podcast version @TheValmy:
thevalmy.com/85

First 20m = familiar stagnation story
Next 7m = excerpts above
Last 30m = Q&A

13/15

Want more Thiel?

https://twitter.com/peterhartree/status/1583509982099607552

14/15

Want more Bostrom?

I started @RadioBostrom so you can listen to his most important papers.

We've released 27 episodes so far:
radiobostrom.com/episodes

If you're new, start here:
radiobostrom.com/introduction

15/15