To the line
Capabilities on the main line · CONVERGENCE

AGI

In one viewA system that handles most jobs at least as well as a competent human — including jobs it was never specifically prepared for.

There is no agreed arrival date. New model records do not automatically resolve reliability, experience transfer and generality.

HYPOTHESISDate unknown
Thought experimentASI
AGICapabilities converge

Drag to rotate · or use the arrow keys

From individual skills to a general system

Select a capability and follow its connections to the core. AGI is shown here as memory, reasoning, action and verification working together.

Conceptual diagram. Its controls do not measure progress or predict dates.

Development chronology

In progress

2026

Some forecasts moved closer

About this eventIn the AI Futures sample, estimates updated in early 2026 shifted earlier, but this is not a scientific consensus.

2026

The spread stays enormous

About this eventFutureSearch's April 12 archive shows differences between AGI definitions and individual forecast updates. Its sample is neither a scientific consensus nor a fresh September forecast.

2026

The academic objection

About this eventIn his 2026 outlook, Stanford HAI's James Landay did not expect AGI that year. This is one researcher's forecast, published December 15, 2025, not a position shared by all academia.

Planned

planned

Agreeing on a definition

About this eventLabs and academia use different criteria. Without a shared yardstick, "achieved" will be declared before it happens.

Distant horizons

ahead

The economic test

About this eventThe most honest check is the share of jobs a system completes end to end and gets paid for by the market rather than a jury.

Sources and research

Primary material behind this dossier: papers, lab publications and official reports.

Capabilities on the main line