The key difference between organizations with and without AI is that without AI you have no choice but to host expertise (as an active, dynamic, improving skill, as opposed to procedures or knowledge) in human brains. For HR, that means you need to hire expertise holders at the scale in which that expertise will have to be applied: as many marketing experts as marketing actions/interactions, as many design experts as things that will need to be (re-)designed, etc.
AI allows — or is rather defined by — hosting expertise in computers. To whatever degree an organization is AI-powered, this means that the role of experts is not to apply expertise but to teach expertise to the organization's AI, which then applies it. This radically changes what HR needs to look for. The scale of your operations is no longer the main factor, but rather its expertise footprint. If you have apply the same sort of expertise two times as often you might not need to hire anybody, while if you need a new form of expertise, you do need to hire somebody who you can use to teach and continuously improve that aspect of the AI.
This flips HR on its head:
- Roles are no longer defined in terms of activities but rather of specific expertise.
- The main map of the resources and needs of the organization isn't the org chart or the team assignments, but rather the cognitive map of what the organization as a whole, and specially its AI, or perhaps the organization as AI, knows and can think about.
- In a way, it becomes part of the AI development team. Scouting and hiring is the first step of development, and therefore it needs to plan and operate in terms of things like AI testing (you hire where your models fail or need to improve) and AI architecture: you hire very differently based on whether you'll add expertise to the AI based on domain knowledge, data, simulations, etc. None of those are cultural choices but rather technical choices driven by the mathematical analysis of your AI architecture goals.
All in all, whatever other impact AI will have on HR in terms of automating parts of it, it has already opened an entire new frontier of needs, challenges, and opportunities that has nothing to do with competing with AIs and everything to do with building them.