Not to be abrasive, but consider that past sentiment with major revolutionary technologies has been similar

The luddite movement in the early 1800s were certain new machines would replace all human labour. Whilst it did replace some human labor jobs, new jobs were created.

John Keynes expressed the same concerns in the 1930s about machines leading to mass unemployment Though he predicted it occurring within the next hundred years, technically he hasn't been proven wrong yet. However since the 1930s, unemployment didn't dramatically rise - the labor market adapted and evolved.

Then before the personal computer was popularized, people were also discussing its ability to displace workers at massive scale. Many, indeed most, wouldn't have predicted today's job market

Past technology did directly replace many of the existing jobs at the time and many people were unable to predict the jobs that would be created from the changes.

It's always occurred in the past with new technology, so there needs to be a more established reason as to why AI will buck this trend of job markets adapting and evolving alongside new technologies.

Don't use difficulty with predicting the future of jobs as a clear indicator that they won't be created, as that's also part of the trend