Results for "aligned agents"

Agent calls external tools dynamically.

Interleaving reasoning and tool use.

Agent reasoning about future outcomes.

AI systems that perceive and act in the physical world through sensors and actuators.

Agents optimize collective outcomes.

Agents have opposing objectives.

Market reacting strategically to AI.

Collective behavior without central control.

Agents copy others’ actions.

The text (and possibly other modalities) given to an LLM to condition its output behavior.

A high-priority instruction layer setting overarching behavior constraints for a chat model.

Expected cumulative reward from a state or state-action pair.

Continuous cycle of observation, reasoning, action, and feedback.

Learning from data generated by a different policy.

Separates planning from execution in agent architectures.

Models evaluating and improving their own outputs.

Internal sensing of joint positions, velocities, and forces.

Using production outcomes to improve models.

External sensing of surroundings (vision, audio, lidar).

Randomizing simulation parameters to improve real-world transfer.

Predicts next state given current state and action.

Directly optimizing control policies.

Modifying reward to accelerate learning.

Reward only given upon task completion.

Modeling environment evolution in latent space.

Acting to minimize surprise or free energy.

Imagined future trajectories.

Perceived actions an environment allows.

Human-like understanding of physical behavior.

Humans assist or override autonomous behavior.