If M&A and electricity costs are factored in, investments exceed USD1tn in the last ten years. Estimates of the gains that will accrue from AI are relatively modest, based on existing jobs and relying on widespread adoption that might take a decade or more to achieve.
Many productivity examples in research reports involve competitive gains by one firm that are losses to another, yielding no net gain.
It is impossible to know if the projected labour productivity improvements will produce net job losses or gains throughout the economy.
Consideration of future tech progress must consider headwinds such as population ageing and the slowing rate of US educational attainment.
Unlike earlier technology revolutions, AI is expected to disrupt high-income workers more than middle-income workers.
