Once weakly general AI has been developed, as defined by this question, https://www.metaculus.com/questions/3479/date-weakly-general-ai-is-publicly-known/ (year x), will US total factor productivity (TFP) growth double within ten years?
I'll be using Fred's Total Factor Productivity at Constant National Prices for United States time series. Unfortunately, this data has very high variance, making a year x and year x + 10 comparison somewhat meaningless.
I will resolve this yes if the ten year average on year x + 10 is double the ten year average on year x. Unfortunately this is still highly susceptible to other things, like recessions, but I couldn't think of anything better. Suggestions welcome.
The twenty year average is much more stable, but the year x + 10 twenty year average would be pulled down by pre-AGI years, and so wouldn't work.
Related questions
@ChrisMillsc5f7 Annual TFP growth has been around .5% the past decade. There are only 2 years in the past 50 years where the ten year average has been 1% or higher. You're suggesting a return to 1% would not be suggestive of AI impacts, but instead of measurement error or something else? If so, any suggestions for alternative measures to look at?
@AmitAmin maybe your question needs to be more tightly worded then. Specify a date a which the 10byear avg tfp has to be above a certain level. I still dont think it is very meaningdul because thw impact of AI will only show up over time, bilut at least we coulsd be sure of what we are betting on...