I (23 terms)

Image Classification Assigning category labels to images. Intermediate Imitation Learning Learning policies from expert demonstrations. Advanced Importance Sampling Sampling from easier distribution with reweighting. Advanced Incident Response Process for managing AI failures. Intermediate Inductive Bias Built-in assumptions guiding learning efficiency and generalization. Intermediate Inference Cost Cost to run models in production. Intermediate Inference Pipeline Model execution path in production. Intermediate Information Asymmetry Some agents know more than others. Advanced Information Cascades Early signals disproportionately influence outcomes. Advanced Information Gain Reduction in uncertainty achieved by observing a variable; used in decision trees and active learning. Intermediate Informed Consent Patient agreement to AI-assisted care. Intermediate Inner Alignment Ensuring learned behavior matches intended objective. Advanced Inner Product Measures similarity and projection between vectors. Advanced Instance Segmentation Pixel-level separation of individual object instances. Intermediate Instrumental Convergence Tendency for agents to pursue resources regardless of final goal. Advanced Instrumental Goals Goals useful regardless of final objective. Advanced Instrumental Variable Variable enabling causal inference despite confounding. Advanced Intent Recognition Inferring human goals from behavior. Frontier Inter-Annotator Agreement Measure of consistency across labelers; low agreement indicates ambiguous tasks or poor guidelines. Intermediate Interpretability Studying internal mechanisms or input influence on outputs (e.g., saliency maps, SHAP, attention analysis). Intermediate Inverse Kinematics Computing joint angles for desired end-effector pose. Advanced Inverse Reinforcement Learning Inferring reward function from observed behavior. Advanced ISO/IEC 23894 International AI risk standard. Intermediate