Results for "system dynamics"
System Dynamics
AdvancedEquations governing how system states change over time.
Think of system dynamics as a way to understand how things change over time, like tracking the growth of a tree or the flow of water in a river. It uses equations to describe how different parts of a system interact and influence each other. For example, if you were studying a city, you could use...
Some agents know more than others.
Supplying buy/sell orders.
Effect of trades on prices.
Sudden extreme market drop.
Groups adopting extreme positions.
Emergence of conventions among agents.
Rate at which AI capabilities improve.
How many requests or tokens can be processed per unit time; affects scalability and cost.
System design where humans validate or guide model outputs, especially for high-stakes decisions.
Extracting system prompts or hidden instructions.
Monte Carlo method for state estimation.
Optimal estimator for linear dynamic systems.
Formal model linking causal mechanisms and variables.
Correctly specifying goals.
AI used in sensitive domains requiring compliance.
Mechanism to disable AI system.
Guaranteed response times.
Maximum system processing rate.
Startup latency for services.
Hard constraints preventing unsafe actions.
Deep learning system for protein structure prediction.
A system that perceives state, selects actions, and pursues goals—often combining LLM reasoning with tools and memory.
Inferring the agent’s internal state from noisy sensor data.
A datastore optimized for similarity search over embeddings, enabling semantic retrieval at scale.
Model trained to predict human preferences (or utility) for candidate outputs; used in RLHF-style pipelines.
Plots true positive rate vs false positive rate across thresholds; summarizes separability.
Central system to store model versions, metadata, approvals, and deployment state.
Observing model inputs/outputs, latency, cost, and quality over time to catch regressions and drift.
System for running consistent evaluations across tasks, versions, prompts, and model settings.
Attacks that manipulate model instructions (especially via retrieved content) to override system goals or exfiltrate data.