It tipped decades ago: in the United States, the labor force participation rate for men has been in secular decline since 1950 (and perhaps earlier—-the Fred data only goes back to 1948):

https://fred.stlouisfed.org/series/LNS11300001

For women, since about 2000: https://fred.stlouisfed.org/series/LNS11300002

Combined: https://fred.stlouisfed.org/series/CIVPART

What's the large gap between labor force participation rate and unemployment rate? Is it the rising number of elderly people?

https://www.bls.gov/news.release/pdf/empsit.pdf

Here’s the graph limited to ages 25-54:

https://fred.stlouisfed.org/series/LNS11300060

The shape is the same, with a peak at ~1999.

There is a large difference between the labor force participation rate and the unemployment rate because the LFPR can’t be gamed. A stark way to think about it is to consider that, if every person currently looking for work were to fail to such an extent that they just gave up, the unemployment rate would drop to zero.

OK, makes sense.

Using your preferred metric, we have about the same LFPR we had in 1980. It has gradually varied since then between about 81% and 85%.

> The shape is the same, with a peak at ~1999.

The trend is up since 2015, apart from the Covid dip.

Automation and AI have not had a dramatic effect yet on LFPR. Maybe there's a tipping point to come.

Yes, if you ascribe to the Singularity theory of technological development. The closer we get to the asymptote, the larger each automation jump will be, and we'll have less time to adjust as a society before another even larger jump in automation hits. Granted, at this point we're talking about basically being at AGI, but still, yes, the idea is that it will tip eventually