Your point is still wrong, because "productivity investments" are more likely to destroy jobs - certainly stable, well-paid jobs - than to create them.
In a very obvious and mundane way, how often do people find that when someone leaves a job they're suddenly doing their own work and the work of the person who left? But without a pay rise?
That's a productivity increase. It is very much not a "generated job."
Add automation to this and it's even more obvious.
The only time jobs are generated is when new markets are opened and this allows new kinds of activity which weren't possible before.
The textbook example is government seed funding of computing in the 50s, which created completely new kinds of business activity. But only after the initial research was paid for by the public. And at the cost of many traditional jobs. (Secretaries, clerks of all kinds, etc.)
Traditional economic theory has almost nothing of value to say about processes like these. There's plenty of mythology, but nothing that will accurately quantify wider social costs/benefits in a strategically useful way.
In reality technologies have downsides as well as upsides, not just in terms of direct externalities but in terms of job losses - and sometimes in terms of productive value losses.
It's exactly this feature of "economic progress" which conventional economics fails to analyse, quantify, or - often - even notice.