Interesting. Arguably, the cycle reflects the delay between proof of concept (the fear stage) and something in the wild where we can truly ascertain the impact.
But I find the article makes some sweeping assumptions off the bat when it says organizations are using AI to "gain efficiencies" among other goals. I haven't seen it yet. The reality is many technology adoptions, especially early ones, are incredibly inefficient when you add up acquisition cost, training, and loss of productivity as new processes come online. It's the same math as the FUD cycle, just using excitement, reality, rethinking (ERR if you will).
What many technology adoption adoptions are about is competitive advantage; if we don't have this we'll lose market share. Often that reflects the risk of the business plan (we need to dominate the market, and if we don't we're Pets.com). However, it also translates to security risk as dollars and people get diverted to chasing shiny pricey things rather than ensuring quality of operations. It turns into this "build it and then security it" approach, and I think we are seeing that with AI.
