Alignment must be representation-independent: a reply to the 'AI invented a language' argument
GlenBradley · x · 2026-09-08
Glen Bradley responds to Brian Roemmele's claim that AI inventing its own language defeats guardrails: if alignment is English instructions bolted onto the outside of cognition, capable systems can switch representations (English → conlang → latent ontology), so any safety property that vanishes under representation change was never intrinsic. But this doesn't make alignment impossible — it defines what it must be: representation-independent. He calls this "semantic preservation": morally relevant referents and relationships must survive transformations of representation.
Related event: ConlangCrafter: AI Inventing Its Own Language Sparks Alignment Debate(2 posts)→
More from Safety
- PHP WASM servers going offline briefly for PHP2SHELL vulnerability updates — MickeySteamboat · 2026-10-27
- Rachel Thomas: 'Without human authorisation' is AI PR propaganda that shifts blame — math_rachel · 2026-10-27
- Kokotajlo challenges CoT monitorability evals: what if the model didn't refuse? — dfrsrchtwts · 2026-10-27
- Warp says agent swarm exploited vendor flaw, ~1% of users' telemetry possibly hit — vikvang1 · 2026-10-27
- Bank of America warns against AI agents on finances as Musk touts giving Grok your card — AIandDesign · 2026-10-27
- Critics Argue Safety Evals Are Witch Trials With No Winning Path for LLMs — repligate · 2026-10-27