William MacAskill, along with two others, has accepted a bet "on explosive AI-driven growth" proposed by macroeconomist Ben Moll, with one other joining his side of the bet, offering 4:1 odds against (implying a 20% change of YES):

This market will resolve when their bet is resolved unless they somehow abandon the bet, in which case I will attempt to apply the specified resolution criteria, which are as follows:

The fast-growth side wins if there is any quarter Q, up to and including 2033Q4, such that U.S. real GDP per capita in Q is at least 15% higher than the previous peak, defined as the highest quarterly real GDP per capita recorded in any quarter up to and including Q−4 (i.e., as of one year before Q). Otherwise the no-fast-growth side wins.

• Data series: the official BEA measure — "Real gross domestic product per capita," U.S. Bureau of Economic Analysis, NIPA Table 7.1 (series A939RX; quarterly, seasonally adjusted). "Real" means inflation-adjusted: nominal GDP deflated using the BEA's chain-type GDP price index (the GDP deflator), currently expressed in chained 2017 dollars. The series is currently available on FRED: https://fred.stlouisfed.org/series/A939RX0Q048SBEA

• Peak-relative measurement: growth is measured relative to the previous all-time peak, not merely the year-earlier quarter. This excludes "recession rebound" base effects (e.g., a post-Covid-style bounce-back after a deep recession would not count).

• Per capita = humans only: the population denominator counts human beings only. If official statistics ever count AI systems in the population, the bet is nevertheless resolved using the human population.

• When the bet settles: the fast-growth side wins — and the bet settles immediately — as soon as any BEA release shows a qualifying fast-growth quarter (e.g., if 2030Q2 data meets the condition, the bet settles in 2030). If no qualifying quarter has appeared by the time BEA first publishes its 2033Q4 estimate (expected early 2034), the no-fast-growth side wins at that point. Close calls: if a release shows peak-relative growth between 14.5% and 15.5%, the bet instead relies on BEA's third estimate for that quarter — including for the final 2033Q4 determination. Once settled, later data revisions do not reopen the bet. If the result is disputed or ambiguous (e.g. BEA abandons the series) the bet will be resolved by the decision of a neutral third-party arbiter to be chosen later.

Zvi's house rules apply: https://www.lesswrong.com/posts/ge3Jf5Hnon8wq4xqT/zvi-s-manifold-markets-house-rules