I think we've been making leaps and bounds from concept to mainstream throughout the years. The technology in 1903 was still pretty basic, and even by 1950s it still wasn't that good. The concern of safety on top of costs really pushed back the mainstream adoption. If we compare it to something relatively safe like mobile phones, it was around 1979 for the concept, and 1983 was the first mainstream one. The succeeding improvements have been consistent and quick.

Generative AI almost came out of nowhere. Looking it up, it was around 2014 where advancements towards it were noted. GPT-1 was in 2018, GPT-2 in 2019, and then Image generation followed in 2021. It was like 3 years for simple text to image improvement in generative AI. And by the rate they are able to train them, their learning speed in other things like coding will only get better.