Ablation: Removal of a portion of an AI capability
Augmented Reality: real-time integration of digital capabilities to the user's setting
Convolutional Neural Network: A neural network designed to work with structured data
Data fusion: Integrate multiple information sources to provide more accurate answers
Data poisoning: insertion of false data into a data set, to decrease quality
Emergent behavior: When an AI system behaves unpredictably
Jailbreak: manipulation of an AI model to bypass imposed restrictions
Natural Language Processing: A model's ability to understand conversationally phrased prompts
Reinforcement learning: AI models training based on a trial and error model
A framework for representing knowledge
AlphaEvolve: A coding agent for scientific and algorithmic discovery
Attention is all you need
BERT: Pre-training of deep bidirectional transformers for language understanding
Fairness and abstraction in sociotechnical systems
Gender shades: Intersectional accuracy disparities in commercial gender classification
Generative agents: Interactive simulacra of human behavior
Gradient-based hyperparameter optimization through reversible learning
Learning to reinforcement learn
Mamba: Linear-time sequence modeling with selective state spaces
On the measure of intelligence
Personality traits in large language models
Tackling climate change with machine learning
The mythos of model interpretability
The next decade in AI: Four steps towards robust artificial intelligence
Time-R1: Towards comprehensive temporal reasoning in LLMs
Using computational patients to evaluate illness mechanisms in schizophrenia
