Skip to main content

Glossary of Terms

Jump to: A · B · C · D · G · H · L · O · U

A​

  • ABM (Agent-Based Model): A modeling product that uses individual entities (like households or biological organisms) as "agents" to simulate how they interact within an environment to identify specific outcomes.
    • Example: The Harvest model simulates how individual households navigate a city to obtain food, helping planners identify neighborhoods at risk of food insecurity.
  • Agentic Artificial Intelligence: AI systems designed to autonomously pursue complex goals and take actions with limited human intervention, adapting based on environmental feedback.
    • Example: An autonomous drone system used in animal ecology that can independently adjust its flight path to track a moving wildlife population without a remote operator's manual steering.
  • Artificial Intelligence (AI): A machine-based system capable of making predictions, recommendations, or decisions that influence real or virtual environments for a given set of human-defined objectives.
    • Example: Using machine learning for Land Use and Land Cover (LULC) classification, where an algorithm automatically distinguishes between forest, water, and urban areas in satellite imagery.

B​

  • Backend: The "data stuff" and underlying coding of a model, as distinguished from the user-facing interface.
    • Example: The hidden software architecture of the Patra system that manages the flow of agricultural data and tracks how different models are performing.

C​

  • Cyberinfrastructure (ICICLE): Digital infrastructure (mostly software) designed to make using and accessing future AIs easier, more equitable, and beneficial to the public.
    • Example: The ICICLE platform itself, which provides researchers with a "plug-and-play" environment to run complex AI workflows like Smart Foodshed analytics without needing to write low-level code.

D​

  • Development: The work of coding an AI model, encompassing both backend data infrastructure and the user-interface design.
    • Example: The engineering process of building the Harvest web tool so that city officials can test different scenarios, such as how adding a new bike path affects food accessibility.

G​

  • Generative Artificial Intelligence (GenAI): A software system that generates new content or data autonomously by learning patterns from existing datasets.
    • Example: An ICICLE science agent that uses a natural language interface to automatically generate the Python code needed for a researcher to analyze a new field dataset.
  • Grounding: The process of connecting an AI model to verifiable, authoritative sources to reduce hallucinations and improve accuracy.
    • Example: Connecting a chatbot to the ICICLE Knowledge Graph so it only provides agricultural advice based on verified, peer-reviewed data.

H​

  • Human in the Loop (HITL): A process where a human actively participates in the operation or decision-making of an automated system to ensure safety and ethical standards.
    • Example: Using the Interactive Knowledge and Learning Environment (IKLE) system, where human domain experts review and approve AI-generated maps before they are used for regional planning.

L​

  • LLM Model Card (or System Card): A structured document for a trained AI model that provides information on its intended use, performance, and limitations.
    • Example: Patra Model Cards are used within the ICICLE project to provide a "living record" of a model's metrics and training data, ensuring accountability and trust throughout its lifecycle.

O​

  • Ontology: A hierarchical description of how different kinds of data interact and network together in a machine-readable format.
    • Example: Within Patra or IKLE, an ontology helps the computer understand that a specific sensor reading for "soil moisture" is a relevant sub-category of "environmental data" for crop management.

U​

  • User Interface (UI): The "Frontend" or user-facing side of software, designed to make complex models usable by non-experts.
    • Example: The interactive dashboard of the Harvest tool that allows community advocates to visualize food access patterns on a map without having to understand the underlying code.

Sources:

  • Gabriel Wilkins, "Gabe's Guide to Artificial Intelligence" (May 23, 2024).
  • State of Oregon DAS EIS, "Responsible Usage of Artificial Intelligence Policy/Packet" (2026).
  • ICICLE AI Training Catalog & News (icicle.osu.edu / arawireless.org).
  • Patra Model Cards & ICICLE-ai SDK Documentation.
Last updated on