Brief research says it does vary wildly based on crop,country, and phase of agriculture. The FAO says:

China has been mechanizing in the last two decades and is up to ~73% overall but 80-90% in wheat, rice and maize.

Sub-Saharan Africa sits at ~10% tractor use across crop farmers, with Ethiopia at ~1% utilization in land prep, and Nigeria at 7% tractors/25% animals for "traction".

India & South Asia ~40-50% across all farming activities

India tilling/plowing "over 70-80%"

India sowing & Harvesting ~30-40%

Major contributors to these figures seem to be relative low cost of labor, lack of capital, farm fragmentation/small plots, and terrain.

I used to work with a guy that started a nonprofit providing tractors to sub-saharan africa. He routinely met farmers/land-owners who could technically purchase the machines, but would choose not to because the profit margin was actually higher if they used human labor. They would accept the machines as donations because at that point the cost of the machine actually made the machine viable economically against the human labor.

I imagine this also extends to politically unstable regions. Buying the machine "pencils" to an American who can be relatively reasonbly sure that they will be able to farm the land for the number of years required to recoup the expense of the machine, but less so for an individual not entirely certain that they will be allowed to own the land for that long.

Yes; In short, if you can depress human costs enough, automation may not be competitive. This is not really an optimistic result for people.