The Frontier

Artificial General Intelligence

Artificial General Intelligence (AGI) refers to systems capable of human-level reasoning, learning, and adaptation across essentially any cognitive task. Whether — and when — AGI will be built is the most consequential open question in technology.

Key takeaways

There is no consensus definition of AGI, but most center on broad task generality and self-directed learning.
Forecasts from leading labs and researchers cluster between 2030 and 2050 — with wide uncertainty.
Alignment — ensuring AGI pursues intended goals — is widely viewed as the core technical risk.
AGI would reshape labor, science, governance, and possibly the trajectory of civilization.

What you'll learn

A balanced examination of AGI: definitions, timelines, technical paths, alignment, governance, and the civilization-scale stakes.

Explore the topics

Deep explainers across the field, from foundational concepts to frontier research.

Defining AGI

Why the term resists a single definition.

Timeline Predictions

What top researchers and labs forecast — and why estimates differ.

Superintelligence

What comes after human-level — and why it matters.

Consciousness Debates

Could AGI be conscious? Would it matter?

AI Alignment

The technical challenge of building AGI that pursues human-compatible goals.

Existential Risks

Scenarios where advanced AI causes catastrophic harm.

Economic Impact

Labor displacement, productivity surge, and the future of work.

Societal Transformation

How AGI could reshape education, science, and governance.

AGI Governance

International coordination, treaties, and oversight regimes.

Frequently asked questions

What is AGI?

Artificial General Intelligence is a hypothetical AI system that matches or exceeds human cognitive ability across virtually any domain — reasoning, learning, planning, and creative problem-solving.

When will AGI be built?

Estimates vary widely. Surveys of AI researchers cluster around 2040–2060 for a 50% probability, but leading lab CEOs have suggested timelines as short as 3–10 years.

Are current LLMs AGI?

Most researchers say no. Frontier LLMs show striking general capability but still struggle with consistent reasoning, novel problem-solving, and long-horizon planning.

What is the alignment problem?

The challenge of ensuring an AGI system reliably pursues the goals its designers intend, even as it becomes more capable than its overseers.

Is AGI dangerous?

Many serious researchers consider misaligned AGI a major risk. Risk levels depend on the path to AGI, governance frameworks, and the success of alignment research.

Glossary

Full glossary

AGI: Artificial General Intelligence — broad, human-level machine cognition.
ASI: Artificial Superintelligence — intelligence substantially exceeding human ability.
Alignment: Research aimed at making AI systems pursue intended goals safely.
Mesa-optimization: When a learned model internally develops its own optimization process.
Capabilities Overhang: Latent capabilities that emerge unexpectedly as models scale.
Foom: Hypothetical rapid, recursive self-improvement leading to superintelligence.