Why aren't SW developers working on more important problems?

I think you get me:)

Yes all of that is purely code and not a mix of multiple disciplines.

That is also done entirely by software developers and not electrical engineers and biologists...

Most actually useful software is a combination of multiple disciplines and the actual software part never needed someone with a computer science degree but a degree in a related field which also covered software development.

It's not a skill issue, it's a matter of software isn't the thing solving the problem in any way, it is only accelerating what was already possible without it.

Regarding your AI in cancer diagnosis a d autonomous tractors, these and not places where software pushed costs down but rather went the opposition direction and is still not available to the masses but instead are available to the 1% who were doing just fine without it.

All that stuff is being worked on by domain experts. I.e. you need to understand cancer or food production in order to think up the code that will help make progress on those subjects. Then, you either code it yourself if you have the coding skills and the time, or hire some coders to code it for you.

An example to help you understand this (as the term "domain expert" is foreign to you). You have a cancer researcher, who's spend years, or decades, learning whatever they can about cancer. They come up with a research problem which would be helped by some code - e.g. they want computer vision program to help them automatically classify some cells in pictures taken with a microscope. This image classification is in itself a small task in the grand scheme on studying cancer and ways to treat it, and it can't be even conceptualized without the deep domain knowledge the researcher has. This researcher is the "domain expert" I'm talking about.

because none of those industries pay what the top five industries you started this post with do. (healthcare is a major industry and pays tech talent somewhat competitively depending on the region)

Sort answer: have bills to pay. Degrees / training to keep current / technical certifications are expensive. More $$ for same amount of time helps pay down expenses quicker.

1) Software development is by definition research related.

2) IT services are non-research realted and have a defined/bounded focus/goal(s).

#2 / bounded development much easier to allocate resources over time (unknowns & knows are defined/generally known up front). Similar topic would be why companies like to outsource things (cloud, human resources, software,

#1 / unbounded development, much harder to allocate resources over time (many unknowns, and unknown "what don't know")

Simplifying things, management likes #2. #2 keeps things within budget/minimizes resources which leaves little room for funding/resources for research/learning.

#1, is similar to doing things for 1st time -- large amount of resources upfront and over time in order to possibily produce something. BUT, that something may not be useable/return resource investment until other known/unknown things happen. Examples: Original development of first IC. Laser. 8088 developed as IC for DataRace display (not as cpu computer). Xero Parc and the Alto. AT&T Unix (vs bsd/minux/linux). Dirty Operating System, Disk Operating Sytem, Gnu DOS. 1970's Polaroid employee developing the digital camera (pre-IC / cpu chips). Rise of Amazon, where amazon didn't return an investment for more than a decade. Nerf as a graduate project. Dell & Facebook as side projects. make/git as quick thing to help get research work done.

Related topics:

case studies/write-ups on why founders get tossed from founding company once company gets beyond start-up phase.

Diy medical device story: https://www.medtechdive.com/news/tidepool-loop-clearance-diy...

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6610599/