When will we have transformative AI?

Economic historians have often emphasized the role economic growth plays as perhaps the most important phenomenon in human history: economic growth is what separates 21st century humanity from our hunter-gatherer ancestors who had no technology and no civilization.

By estimating historical economic growth, economists have identified two crucial events in human history. These were the agricultural revolution around 10,000 years ago, and the industrial revolution around 200 years ago.

Many experts anticipate a third crucial event in which economic growth greatly accelerates, corresponding to the development of transformative artificial intelligence (TAI). There exist a few different definitions of TAI (see, e.g., Karnofsky, 2016), but TAI and related notions like explosive AI and AI takeoff are most commonly characterized in terms of economic growth (e.g., Barnett, 2020; Roodman, 2020), and 30% is the figure that is most commonly cited as the threshold for growth to count as transformative (e.g., Cotra, 2020; Davidson, 2021a; Davidson, 2021b; Karnofsky, 2021).

Economic growth is measured here as real world GDP growth over a year. However, since economic growth can be artificially high following a recession, this question will not ask about economic growth per se, but instead about GDP levels reaching a new height.

(Note: This question is intended to be our—Metaculus'—canonical TAI question. It draws from our other question "In which year will the world's real GDP first exceed 130% of its highest level from any previous year?", but explicity ties the resolution to TAI.)