Interesting that Torsten Bell, an economist , is looking to the potential of AI for UK growth
Interesting (as always). Argues that strengthening evidence of AI making individuals more productive at specific tasks is not (yet) driving the relatively strong aggregate productivity growth we’ve seen in the US (which is driven more by sweating the existing capital stock) https://t.co/jXRhLk8h0c
— Torsten Bell (@TorstenBell) August 5, 2026
Academic research is coalescing around the finding that AI drives meaningful productivity gains at the micro-level: tasks, workers, firms. These findings are particularly compelling because they largely use older generations of models. Likely that newer models even stronger. /2 pic.twitter.com/Po0l8JT9GA
— Ernie Tedeschi (@ernietedeschi) July 23, 2026
We run US productivity growth through a Markov-switching model, a statistical tool economists use to identify different distinct periods in the data. Our model suggests the US is currently in a high-productivity regime, though it’s uncertain how long this will persist. /4 pic.twitter.com/8jCz4J2laG
— Ernie Tedeschi (@ernietedeschi) July 23, 2026
1st, if AI were driving macroproductivity gains, we’d expect to see an acceleration in total factor productivity (TFP). TFP is what economists think of as “pure” technological & efficiency growth. But recent estimates from both @sffed & @BLS_gov have been weak. /6 pic.twitter.com/Phnvx1mj92
— Ernie Tedeschi (@ernietedeschi) July 23, 2026
How could it be that AI has microproductivity effects that aren’t showing up in aggregate yet? One possibility is that AI is speeding up certain tasks, but firms haven’t yet adjusted to ease downstream bottlenecks and unlock these gains in aggregate. /8 pic.twitter.com/mm8Nv2AJTn
— Ernie Tedeschi (@ernietedeschi) July 23, 2026
We don’t think this is a pessimistic read of AI, by the way. The microproductivity research is promising, especially as it shifts to newer models, & firms should gradually reorganize to unblock those gains, just as they have done in response to prior technological shocks. /10
— Ernie Tedeschi (@ernietedeschi) July 23, 2026
Full post here:https://t.co/sx84W2cDN0
— Ernie Tedeschi (@ernietedeschi) July 23, 2026
