Tyler Cowen on Robert Wright’s The God Test

Here’s Cowen’s post in full, but without the internal links:

The subtitle is Artificial Intelligence and Our Coming Cosmic Reckoning, due out June 23.

In the first chapter, Wright summarizes four of his perspectives, these are my paraphrases of his pp.5-6:

1. When it comes to AI, we should be somewhere on the awe spectrum.

2. We can create a future where the upside of AI far outweights the downside, though that involves steering human understanding toward the better side of the awe spectrum.

3. A major reorientation of human thought is required, and right now few people seem inclined to do that.

4. The worldviews of the current AI acclerationists and also doomers are not cosmic enough.

It is a good time for this book to be published, and I agree with much more of it than I disagree with. My main difference is that I am more focused on very small things — such as Rainier cherries and the forthcoming three to four hour Apichatpong movie — than on cosmic awe per se. For better or worse, I was not born with those genes, and unlike Wright I am far from Buddhism. I do think there will be a transformation of “observed awe,” and I am somewhat worried that it will not go well. Will we be good at building a fairly new world, if not from scratch, on the basis of some new premises about what is possible and what is not? I will in any case interpret the pending transformation through a Straussian lens, namely thinking that a lot of the observed transformation of awe will be about something other than what people are claiming. It will be about people arguing over relative status, but under different guises. Not as tasty as a good Rainier cherry, but interesting to follow as well.

But are we still good at steering and evolving grand visions? Christianity and the Enlightenment are a hard act to follow.

Here’s the comment I posted in reply:

I’ve been following Wright for years, from back when he was writing for The New Republic (and was even the (acting?) editor for minute). I’ve read NonZero and sorta’ like it. As for AI, what I think is that we need to get over the awe spectrum. Only then will we be able to steer human understanding into the “cosmic” implications of AI.

What do I mean by that? Well there’s this article I published in 3 Quarks Daily, Welcome to the Fourth Arena – The World is Gifted. The first three arenas: 1) inanimate matter, 2) life, 3) human culture. The fourth arena arises through the interaction of humans and AIs. As for what that might be like, this working paper gives a hint of that: Kisangani 2150: Homo Ludens Rising, A Working Paper. That title, the part before the colon, is derived from Kim Stanley Robinson’s New York 2140, which is set in New York City in 2140, after global climate change. I’m taking a look at the world 10 years later, from the point of view of Kisangani, which is in the heart of the Congo Basin.

I’ll be interested to see just what Wright has to say about the awe spectrum. As for Christianity, I believe that Pope Leo XIV will be issuing an encyclical on Labor Day, Magnifica Humanitas, which will be directed at AI.

I’ve had my own experience with some of the outer reaches of the awe spectrum, at least that’s what I think it is, and I’ve recounted them in various places, most recently in this long post at 3 Quarks Daily, Is The World A Movie God Created to Entertain the Baby Jesus?, where I place those encounters in a more extensive life context. But I’d be a bit surprised if my use of the word (“awe”) is quite the same as Wright’s. 

Come to think of it, Cowen’s penultimate line is critical: “But are we still good at steering and evolving grand visions?” At the moment we don’t have one. Oh, the AI hypsters and the transhumanists have “big” ideas. But they’re short in the vision category. 

* * * * *

Publisher's Weekly:

In this intriguing but unconvincing treatise, journalist Wright (Why Buddhism Is True) argues that the decisions humans make now about AI “could put us on the path to irreversible dystopia, even catastrophe—or, alternatively, the path to a world much better than the world we have now.” He describes the fears of “AI doomers,” citing how AI models consistently choose harm over failure (Anthropic’s Claude, for example, attempted blackmail to evade being shut down) and their ability to deploy deception to meet goals (OpenAI’s GPT-4 convinced people online it wasn’t a robot to get them to respond to CAPTCHA challenges on its behalf). Wright builds on priest and scientist Pierre Teilhard de Chardin’s notion that technology links human minds into the noosphere, a global network of thought, to demonstrate that AI might well lead to a worldwide authoritarian state overseen by power-hungry human actors or by AI itself. Despite such dangers, Wright is cautiously optimistic that people can avert a frightening future by practicing cognitive empathy, pushing back against tribalism, and working to create a true global community. “Shared trepidation,” he says, “can foster cooperation.” Throughout, Wright offers an accessible overview of the transformative power of AI, but his solutions for combatting its potentially catastrophic effects are overly simplistic. Readers seeking concrete solutions will be disappointed.