ORIAS
An ORIAS research initiative
Worldfold
Discover new sources of capability. Investigate reality, mathematics, computation and intelligence deeply enough to find the powerful principles that have been overlooked, and build them into machines, engines, models or entirely new technologies.
Plate VIThe edge of the map is where the work starts.
The possibility.
Significant forms of capability may remain out of reach not because reality forbids them, but because we haven't found the principles, representations, interactions or methods yet.
We want to understand whether the limits of present technology are fundamental or come from the languages, organisations and assumptions we currently depend on. We are especially interested in capabilities that change what can be done repeatedly, not merely what can be demonstrated once.
The Dream Machine is a hypothesis, not a plan.
One possible destination is an intelligence that is not bound to the mathematical language and procedures it started with, and can grow capabilities beyond what its designers programmed.
Another is a powerful mathematical or physical mechanism that needs no such intelligence at all. Another may be something we don't yet have the vocabulary to imagine. An extraordinary mathematical discovery followed by a fixed but revolutionary engine would count just as much.
The open questions.
Is there an unfamiliar computational primitive with enormous practical leverage?
Is there a form of intelligence whose principle differs from inference, search, learning and planning?
Can a machine originate new representations when the language of its problem is inadequate?
Can mathematical discovery become a scalable source of capability?
Which forms of technological power does our vocabulary of intelligence and algorithms miss?
Where is the boundary between what reality permits and what we can reliably make possible?
What it is not.
Not a finished machine, an announced product, or a fixed design for a future AI. Not an AI scientist, a theorem prover or a collection of agent frameworks.
We begin with software, mathematics, existing computers and rigorous experiments. We do not begin by assuming we know what the final invention is.