AI systems will increasingly develop epistemic tools to help determine truth, making trust and verification the critical challenges ahead.

AI-powered epistemic tools that help people determine what's true could have significant impact across multiple fields and cause areas.

As R&D becomes automated, AI systems will play a larger role in developing these truth-tracking tools, with compute availability becoming the main limiting factor.

The hardest problem will shift from building useful epistemic systems to ensuring people trust the right ones and verifying accuracy in difficult-to-verify domains.

Early versions of these tools will have lasting effects due to incumbency advantages, making current decisions about their development particularly important.