SIGNAL//DESK
AI safetysrc: curated AI glossary

AGI

AGI is a hypothetical type of computer program that could learn and perform any intellectual task a human can do, rather than being limited to just one specific job like playing chess or translating languages.

AGI refers to a theoretical autonomous system capable of generalized learning and cross-domain reasoning, possessing the flexibility to achieve goals across a broad spectrum of cognitive tasks at or above human parity.

Artificial General Intelligence: a hypothesised AI that matches or exceeds human ability across most cognitive tasks, in contrast to today's narrow, task-specific systems.

evolution

  1. 1956 · history
    Dartmouth Workshop

    The term 'Artificial Intelligence' was coined, establishing the foundational goal of creating machines capable of human-level reasoning.

  2. 1997 · history
    Deep Blue defeats Kasparov

    The first major milestone where a machine outperformed a world champion in a complex cognitive task, sparking debates on the limits of narrow vs. general AI.

  3. 2012 · history
    AlexNet and Deep Learning

    The breakthrough in deep neural networks demonstrated that scaling compute and data could achieve unprecedented performance, shifting AGI research toward connectionist approaches.

  4. 2020 · history
    GPT-3 Release

    The emergence of large language models showed emergent capabilities across diverse tasks, leading to widespread speculation that AGI might be achievable via scaling laws.

  5. 2023 · history
    GPT-4 Technical Report

    OpenAI published evaluations suggesting the model exhibits human-level performance on various professional and academic benchmarks, intensifying the discourse on AGI safety and alignment.


← all terms