Into The Impossible: Robert Wright on AI Alignment as a Moral Test

SPEAKER_1 (0:01)

Powerful technology can look like progress, until it forces a civilization to ask what kind of beings it really is.

In this episode of Into The Impossible, Brian Keating sits down with Robert Wright to explore a fascinating idea: AI alignment isn't just a technical challenge, but fundamentally a moral one.

That's right. Wright, who's the author of The God Test and an early observer of neural networks, frames this in striking terms. He says humanity may be facing "the kind of test that God would give." He uses the Tower of Babel as a lens to examine whether humans can handle godlike power without the wisdom to match it. And it's not just about building safe superintelligence, is it?

Wright argues the real question is whether people can become morally capable enough to govern it. As he puts it, "We are gods in power, but lack the wisdom to wield that power." Exactly. The real task is learning to act as a global community rather than a collection of rival tribes. But what's interesting is how Wright pushes back against both extremes in the AI debate. He talks about the "idolatry of AI," where some people personify models as if they're destined to create or destroy civilizations.

He's skeptical of both doomers and accelerationists when their views become quasi-religious. At the same time, though, he doesn't dismiss real-world risks like bioweapons. He insists AI governance will require international coordination because the stakes cross national borders. Building on that point, Keating and Wright also dig into how large language models actually work. Wright explains that modern LLMs learn by predicting the next word, and in doing so they build a semantic map of language in high-dimensional space. It's fascinating how he describes that. He compares the process to a kind of reverse-engineering of function, not a magical imitation of the human brain. And he's cautious about claims that these systems are already superintelligent agents bent on annihilation. They're powerful, but not the same as an intentional mind.

Despite the risks, Wright is actually more optimistic about "P-Boom" than "P-Doom." He suggests technological evolution may be pushing humanity toward a larger form of cooperation, maybe even a global brain that preserves human flourishing. And he links that hope to enlightenment, saying the deepest progress may come from overcoming tribal certainty and seeing reality more clearly. By the end, the conversation widens into something profound: the Fermi paradox, civilization, and whether passing this AI test could be part of a much larger story about life, meaning, and the future of the universe.