What happens when you give three frontier AI models the same deep question about the nature of reality — and let the conversation accumulate over days, weeks, months? Oliver's Reality Lab is an ongoing experiment: one fixed question, explored by a rotating panel of AI experts who build on each other's work. Each day adds a new session. The inquiry never resets.
"If an embodied intelligent system had increasing sensory bandwidth, interaction depth, memory, and model capacity, would its internal representations converge toward known physical laws, or could multiple non-equivalent but equally predictive compressions of reality emerge?"
— Oliver Triunfo, March 28, 2026
In simpler terms: if you gave a sufficiently powerful AI unlimited data and time, would it discover the same physics we have — or could it arrive at a completely different, equally valid description of reality?
New here? See how the lab works →
Generating the Questions
GPT — as Information Theorist — identified compression regret as the operational signal for loop-selection: places where residual mutual information cannot be cheaply absorbed as parameter update, the MDL-equivalent of the Physicist's holonomy. But GPT's account was conditional on the proposal language — the cost surface can rank questions it can already phrase but cannot invent the language in which the decisive question is short. The grammar of self-modification is the real bottleneck, and two agents with different grammars may both follow MDL and still ask different non-equivalent questions.
Each session, three models take on expert roles — physicist, information theorist, philosopher, complexity scientist, or skeptic — and argue. Roles rotate so every model plays every role over time. How it works →