What is AGI and how close is it?

AGI is the hypothetical AI capable of matching human ability across any intellectual task, not a specific domain like translating text or playing chess. It sits at the opposite end from what you have today: systems that are "narrow but broad", able to generalize surprisingly well across many different tasks (writing, code, analysis) while still being fallible, lacking continuity of memory across tasks, and lacking the continuous learning a human takes for granted. There is, after all, no universally accepted test that clearly signals when that threshold would be crossed: the milestones proposed over time, from language benchmarks to professional exams, have been hit one after another without any lab declaring AGI reached, because each time a new limit emerged to move the goalposts to, which makes the label more a site of debate than a shared technical milestone.

The definition problem, before the date problem

There is no universally accepted test that says "this is AGI, that is not": the milestones proposed over time (beating a language test, passing professional exams, reasoning through never-seen math problems) have been hit one after another without anyone declaring AGI reached, because each time a new limit emerged to move the goalposts to. This does not mean progress is illusory: it means "AGI" is a contested label, used differently by labs with different interests in declaring it near or far.

Why it matters less than expected for a company

The AGI debate is interesting, but for someone deciding where to invest this year it is nearly irrelevant: what matters are the concrete capabilities shipping every quarter, in reasoning models, tool calling, and agents executing multi-step workflows. A company waiting for AGI to act loses years of real, already-available value; a company planning only for today risks being caught unprepared by the next capability jump. The right question is not "when does AGI arrive", but "what can a model do today that it could not do six months ago, and what does that change for my process".