SEARCH · ENGLISH EDITION
Find an AI concept.
Search titles, summaries, article text, source-language titles, and known aliases.
AI Agent
An AI system that selects and executes actions over multiple steps to pursue a goal.
AI Agent Oversight
Human control over agent goals, tool permissions, execution and outcomes, including meaningful reviews and recovery from failures.
AI Alignment
Research and design for relating AI behavior to human intentions, values and constraints, including reward proxies, feedback and operational evaluation.
AI Memory
Stored conversation state, preferences and task records, distinguished from context windows and model training, with examples of freshness and access boundaries.
AI for Science
AI-assisted scientific prediction and candidate selection, with attention to experimental validation, data leakage and the scope of automated research.
Anthropic
An AI company that develops the Claude model and product family and conducts AI safety research.
Attention
A mechanism that lets a neural network weigh the relevance of different positions when computing a representation.
BERT
Bidirectional Encoder Representations from Transformers, an encoder-based language representation model.
Benchmark Data Contamination
The presence of evaluation examples or closely related material in training data, potentially inflating benchmark results.
ChatGPT
A conversational AI product from OpenAI that provides natural-language interaction with AI models.
Claude
A family of general-purpose AI models and conversational products developed by Anthropic.
Context Window
The amount of tokenized input and generated text a model can consider within one inference context.
Data Provenance
Information about where data came from, how it was collected, and how it changed before use.
Datasheets for Datasets
A documentation framework for describing how datasets were created, composed, maintained, and intended to be used.
Embedding
A numerical vector representation used to encode semantic or structural properties of data.
Fine-tuning
Additional training that adapts a pre-trained model to a narrower task, domain, or behavior.
Foundation Model
A broadly trained model that can be adapted to many downstream tasks and applications.
GPT
Generative Pre-trained Transformer, a family of autoregressive language models associated with OpenAI.
GPT-6 Astra
An OpenAI GPT-family model described for complex reasoning, coding, research and tool-based work, with a practical framework for task-level evaluation.
Google AI Ecosystem
A guide to separating Google's research organizations, models, products and Cloud infrastructure when reading AI announcements.
Google DeepMind
Google's AI research organization, formed from the combination of DeepMind and Google Brain teams.
HELM
Holistic Evaluation of Language Models, a framework for evaluating models across scenarios and metrics.
Hallucination in AI
A generated statement that is unsupported, fabricated, or inconsistent with the available evidence or context.
KV Cache
A cache of attention keys and values used to avoid recomputing earlier context during autoregressive generation.
Knowledge Distillation
A training method in which a smaller student model learns from the outputs or representations of a larger teacher model.
Large Language Model (LLM)
A general-purpose language model trained on large datasets and usually built at substantial scale.
Latency and Throughput
Two core serving metrics describing response delay and the amount of work a system completes over time.
MMLU
Massive Multitask Language Understanding, a benchmark covering many academic and professional subject areas.
Mistral AI
An AI company providing models and inference services, with a guide to model-specific licenses, hosted versus self-operated deployment and task-level comparison.
Mixture of Experts (MoE)
A model architecture that routes each input to a subset of specialized parameter blocks called experts.
Model Card
A structured document that describes a model's intended use, evaluation, limitations, and other important context.
Multimodal AI
AI systems that process or generate more than one type of data, such as text, images, audio, or video.
OpenAI
An artificial intelligence research and deployment organization known for GPT models and ChatGPT.
Physical AI
AI that connects perception and decision-making to physical actions, including robotics toolchains, sim-to-real gaps and evaluation of complete tasks.
Project Suncatcher
Google Research's exploration of AI computing in space, with solar power, accelerators and optical links examined as a research system.
Quantization
A technique that represents model weights or activations with lower numerical precision to reduce resource use.
RLHF
Reinforcement Learning from Human Feedback, a method for aligning model behavior with human preferences.
Red Teaming
Adversarial testing intended to discover failure modes, unsafe behavior, or exploitable weaknesses in an AI system.
Retrieval-Augmented Generation (RAG)
A method that retrieves external information and supplies it to a generative model as context.
Safety Evaluation
Testing designed to measure risks, harmful behaviors, and the effectiveness of safeguards in an AI system.
Sovereign AI
Control over AI data, infrastructure, models and operational capabilities, examined separately from data location and supplier branding.
System Card
Documentation describing a deployed AI system, including capabilities, evaluations, mitigations, and limitations.
Tensor Processing Unit (TPU)
Google's specialized machine-learning accelerator, explained through matrix computation, software compatibility and workload-level evaluation.
Transformer
A neural network architecture that processes sequences primarily through attention mechanisms.
Vector Search
Search that retrieves items by similarity between numerical vector representations.
Voice AI
Speech recognition, synthesis and conversation systems, including full-duplex interaction, voice replication and evaluation of actual task outcomes.