Banque de France finds rapid AI uptake but weak productivity gains

Banque de France’s bulletin is a useful counterweight to the hype cycle. It confirms that generative AI is spreading fast inside firms, especially in support and cross-cutting roles like admin and sales, but it also notes that the impact on measured productivity is still modest. That suggests we are in a familiar phase of general-purpose technology diffusion where tools get sprinkled on top of old processes before organizations rewire themselves to take full advantage.

For the AGI race, this matters in two ways. First, it hints that macro-level productivity explosions may not show up until long after the underlying capabilities are in place, which could lull policymakers into underestimating risk and opportunity. Second, the fact that most use is still in general-purpose tools means specialized, deeply integrated AI systems have plenty of room to grow. As frontier models like Astra arrive, they will plug into an enterprise landscape where basic AI literacy is there but workflows are still immature. That is fertile ground for fast follow-on gains once companies figure out how to hand real authority to AI systems.