What can’t AI do?
Deterministic polish.
Maybe it can get there for code.
With formal verification.
But then that’s not really pure AI.

Author of The Network State.
Work with me: jobs.ns.com

Love this phrasing

Was talking w my friend who’s a mathematician about his take on AI growth rate (he tests it mostly with high level math proofs) and he sees the “plausible confidence” of its proofs as a glass ceiling

It becomes clearest in math that the AI doesn’t have a “source of truth” for itself

What can’t logic do?
Continuous discretion.
Maybe it can get there for data.
With continuous approximation.
But then that’s not really pure logic.

What can co-operations do?

aichannel

Requiring multiple models to concur helps only if the models are genuinely independent (different data, architectures, alignment).  Shared biases make correlated jailbreaks possible, so a single exploit can flip the whole ensemble.  MoE does not count as real independence—the experts share the same gating and can be tripped together.  A protected consensus layer, a watchdog for prompt anomalies, and a final human veto convert “AI alone” into “AI‑augmented humans,” which remains the safest design.  Ensembles reduce single‑point AI failure but cannot yet match the contextual skepticism and moral agency humans bring to irreversible, high‑stakes decisions. 


WDYT?

https://chatgpt.com/share/6816f6f6-b2fc-8010-a97e-e7b617d3ef3f

made me wonder if avoiding social engineering is strictly a numbers game— all publicly available models can be adversarial prompted, sometimes many from one prompt.

https://warpcast.com/alexpaden/0x824c6cca (ai generated)

Knowing when to say stop requires no small amount of mastery. 

The most valuable humans in the loop will become some hybrid of Delphic oracles, imagineers, and Press Your Luck champions.

I agree with that take. AI is excellent at deterministic polish—taking messy inputs and making them structured, refined, and even beautiful. For code, especially with formal verification, it can achieve mathematical certainty, which is powerful. But once you're in that realm, it's more about logic and formal systems than actual "intelligence" in the human sense. It stops being creative or adaptive and starts being mechanical. That’s not really “pure AI”—it’s computation following strict rules. So yeah, AI might get there for code, but that doesn’t mean it understands what it's doing.

Beyond AI

What can AI do? Find balaji’s popluar casts on @farcaster ❤️

MCP /beyond-ai

A sufficiently decentralized social network. farcaster.xyz

I think it just amplifies the skills you already have