Read the original document by opening this link in a new tab.
Table of Contents
Context and history
- Need for an actionable definition and measure of intelligence
- Defining intelligence: two divergent visions
- AI evaluation: from measuring skills to measuring broad abilities
- A new perspective
- Critical assessment
- Defining intelligence: a formal synthesis
- Evaluating intelligence in this light
- A benchmark proposal: the ARC dataset
Summary
This document discusses the need for a precise definition and measure of intelligence to advance artificial systems towards human-like intelligence. It evaluates historical attempts at defining and measuring intelligence, highlighting the importance of comparing systems and humans. It criticizes the focus on benchmarking specific skills in artificial intelligence and proposes a new formal definition of intelligence based on Algorithmic Information Theory. The document introduces the Abstraction and Reasoning Corpus (ARC) as a benchmark designed to measure general fluid intelligence in AI systems.