Скачать или смотреть A New Way to Understand AI: The 6 Core Functions

AI is moving faster than most people can keep up with. New models and buzzwords appear every week, but the capabilities behind them change much more slowly.

In this video, I introduce a function-based ontology of AI: a new way to structure the AI landscape around what AI does (capability), rather than around model names or training paradigms. The goal is to give you a stable mental map that still works as architectures and algorithms evolve.

The six core functions that cover most AI systems at the capability level are:
1. Represent — encode raw inputs into structured states
2. Reason & Infer — derive relations, constraints, or beliefs from representations
3. Optimize & Control — choose parameters or actions to optimize objectives under constraints
4. Simulate & Emulate — simulate dynamics or emulate expensive processes with surrogates
5. Generate & Create — produce samples, designs, or hypotheses under priors/constraints
6. Autonomize & Orchestrate — coordinate multi-step workflows and agents toward goals

A function-based view helps you:
• avoid getting trapped by rapidly shifting terminology
• understand what a model is doing in a workflow (its functional role)
• compare approaches across subfields using a shared capability vocabulary
• translate ideas across domains (science, art, engineering, etc.) without relying on fragile labels

In follow-up videos, I'll show how this stable skeleton can be applied across different domains.

This video is based on AI Core Function Ontology: A Two-Level Capability Framework (DOI: https://doi.org/10.5281/zenodo.17664037).