Results for "theory discovery"

Supplying buy/sell orders.

System-level behavior arising from interactions.

Early signals disproportionately influence outcomes.

Emergence of conventions among agents.

AI capable of performing most intellectual tasks humans can.

Awareness and regulation of internal processes.

Existential risk from AI systems.

Stored compute or algorithms enabling rapid jumps.

Tendency to gain control/resources.

Isolating AI systems.

Intelligence and goals are independent.

Goals useful regardless of final objective.

Regulating access to large-scale compute.

Restricting distribution of powerful models.

Designing AI to cooperate with humans and each other.

Inferring and aligning with human preferences.

A measure of a model class’s expressive capacity based on its ability to shatter datasets.

The field of building systems that perform tasks associated with human intelligence—perception, reasoning, language, planning, and decision-making—via algori...

Decisions dependent on others’ actions.

Agents copy others’ actions.

Learning from data by constructing “pseudo-labels” (e.g., next-token prediction, masked modeling) without manual annotation.