Artificial general intelligence

What makes this fascinating

• No agreed finish line — There's no measurable definition of “human-level, general” intelligence — which is exactly why “are we there yet?” is so contested.

• Broad but brittle — Today's models are remarkably general yet still fail in surprising ways outside their training distribution. Closing that gap is the crux.

• Capability isn't alignment — Building a system that's generally capable and ensuring it does what we intend are separate problems — possibly equally hard.

Frequently asked questions

What is artificial general intelligence (AGI)?

AGI is a hypothetical AI with the broad, flexible competence of a human mind — able to learn and reason across any domain rather than being narrow to one task. Today's systems are powerful but specialized.

How close are we to AGI?

Unknown and hotly debated. Estimates among researchers span from a decade to many decades or never, partly because there is no agreed definition or test for when a machine matches human general intelligence.

Is AGI dangerous?

Potentially. A system with general competence and goals misaligned with human values could be hard to control, which is why AI safety and alignment are active research fields. The level and timing of any risk are debated.

More summits in Computer Science

• The foundations of cryptography

  All modern security rests on a bet: that some problems are truly hard. Can we ever prove it?

• Quantum computing's true power

  Which problems can quantum machines crack that ordinary computers never will?

• AI interpretability

  We build neural networks we can't read. Can we ever truly understand what they learn inside?

• Provably correct software

  Can we ever write software proven, mathematically, to be free of bugs and exploits?