FWIW, people had the same thoughts when Macromedia Dreamweaver came out in 1997 and suddenly graphic designers could directly generate HTML code.
Technology always has a convenience vs flexibility trade-off, and a new tool that offers more convenience will be useful for fewer use cases, while a tool that is designed around flexibility will spawn a new ecosystem of supporting elements to make it convenient enough to use.
There is always a mass market that doesn't really need the flexibility, and these are well served by "no coding" platforms -- but they have been served by those for decades now, and this market has never had enough of a profit margin for humans in the loop anyway.
As a web developer, you do things that are not handled by off-the-shelf software.
Web sites that are the product, like service platforms, will always require more and more features to distinguish themselves from the competition, and there are a lot of other companies that need special tweaks for their websites as well.
So it is rather unlikely that the amount of work will significantly change.
What does change are frameworks, because every time a framework got far enough into "convenient" territory, someone else will find it too inflexible, and build a new one that handles something that wasn't easily doable before.
ML/AI is a growing field, but most of the demand they have is in designing training sets and verifying models, which are theory heavy and do not have a tight feedback loop, so the work style in that field is completely different. It may still be worth looking at it if it's interesting to you, though.
