What is AGI (artificial general intelligence)?
AGI (Artificial General Intelligence) refers to AI that can learn and solve a wide range of problems on its own, like a human, rather than excelling only at specific tasks like today's AI.
It does not exist yet, and even its precise definition and the bar for reaching it are still debated.
How is AGI different from the AI we have today?
Today's AI is "narrow" AI, strong within defined areas such as translation, image recognition, or Go. AGI would not be confined to one domain and could teach itself to handle problems it has never seen.
In other words, the key difference is not whether it does one task well, but whether it can learn new tasks across the board.
How is AGI different from superintelligence (ASI)?
AGI generally means human-level general intelligence. Superintelligence (ASI, Artificial Superintelligence) means a stage that broadly exceeds human ability.
Both are still hypothetical, but discussion of what comes after AGI often brings up safety and control issues.
When will AGI arrive?
Even experts' predictions differ widely, and there is no consensus on what should count as "achieving AGI."
Rather than pinning down a specific year, it makes more sense to track progress across several dimensions at once: capability, safety, and social impact.