Caesar AI Atlas

Grounding (Verankerung)

Caesar AI Atlas Definition

Grounding ist der Prozess, die Ausgabe eines KI-Modells zum Inferenzzeitpunkt mit vertrauenswürdigen, überprüfbaren Informationsquellen zu verbinden. In generativer KI hilft es, faktische Zuverlässigkeit zu verbessern, indem relevanter Kontext bereitgestellt wird, ohne die zugrunde liegenden Modellgewichte zu verändern. Retrieval-Augmented Generation ist eine häufige Grounding-Technik.

Other Definitions

Grounding (Verankerung) Source

The process of basing all or part of an LLM's response on information retrieved from one or more trusted sources. For example, suppose a user prompts an LLM for today's weather forecast in Berlin. The LLM might ground the response on information it gathers from the European Centre for Medium-Range Weather Forecasts. Retrieval-augmented generation (RAG) is a common grounding technique.

Grounding (Verankerung) Source

The process of connecting a model's output to verifiable sources of information. Grounding helps to improve the accuracy, reliability, and usefulness of AI outputs by providing the model with access to specific data sources, and reduces the likelihood of hallucinations. A common type of grounding is retrieval-augmented generation (RAG).

Grounding (Verankerung) Source

Providing context or relevant knowledge to an AI model by connecting it to trusted data sources at inference time. This does not update the model itself.

Also Referenced In

Concept Comparisons

Related Terms