In the 90s there was the same hysteria around AI and a lot of promises that went unfulfilled before we went into another AI winter.

Generative AI is currently in the "frenzy" stage of Carlota Perez' Technology Cycle. We have a lot of experimentation with single task tools that have serious UX issues for consumers.

The technology, (generative deep learning), will work its way into existing products, services and processes and move away from the consumer for the most part in order to attain broader adoption and reduce consumer friction. Prompt engineering is temporal, it won't persist in the long term.

The differentiation between generative AI and other AI methods (discriminative deep learning, reinforcement learning, machine learning algorithms, etc...) will dissipate and all blend into the AI umbrella again.

Ultimately AI will once again become invisible to users and an expected aspect of the software they interact with.