Across the projects I’ve built in these past couple of months, productivity gains from AI are wildly uneven. Some work sees orders of magnitude. Some sees nothing. That gap is creating four problems
-
Market valuation: Markets eventually readjust when productivity and cost shifts. But “eventually” is doing a lot of work. Real-world equilibrium takes years that supply-demand macroeconomics simplify away, and the interim turbulence benefits some and disadvantages others.
-
Unrealistic expectations: The aforementioned 2 ends can coexist within the same organisation. Commodity, AI-automatable work becomes the standard everyone else is judged against. The scarcest skill sets suddenly get confused with being luddite. Speed is confused with impact. This naivety too will adjust, but only “in due time”.
-
Who captures the upside of productivity gain: Employers are trying to capture the value, employees have to push for their fair shares. The salary and employment terms negotiated before the AI revolution - does it still reflect the fair value creation by the employee?
-
The “zombie” business: A poorly-managed business, about to go bankrupt due to bad decisions and inefficiencies from years prior - then comes the AI saviour. Suddenly employees are squeezed for productivity gains to compensate for years of mismanagement. In a free market, businesses can fail, and only people who know what they’re doing survive. Then suddenly AI makes marginal business arrangements viable. But marginal viability is fragile. When AI providers stop investing and start extracting, suddenly the bad businesses take the hit. The current state in the meantime, is a zombie state, neither alive nor dead.
I don’t have answers to these questions from the economics or policy maker angle. As one of the technical practitioners, I can only navigate with clear-eyed awareness of the revolutionary time.
I just want to leave this out here as something I’m seeing. What do you see from your perspective?