1. the metrics you mention are hard to measure and usually lag 2. the metrics might not be moving yet, because AI accelerates everything a little bit, and most of the hype is still the promise rather than the actuality 3. people are using the gained productivity to speed up secondary tasks / do different work, because applying it to their main work is still complex or not of sufficient quality
I said elsewhere, I don't have any beliefs about this problem, I am deeply fascinated by the difficulty involved in productivity analyses and especially by the question of missing productivity that we should expect to be driven by the huge investments in computer technology that have been observed over the past half century - and quickly accelerating of late.
It's a fascinating topic and I intended merely to point out that quantifying computer technology driven productivity growth is a hard problem. Any facile conclusions one way or the other are suspicious in my view.
It is not clear to me at all that building and buying more computer and data centers and software is an obviously good investment that should only be increased because it raises productivity. That is the discourse, but there is a very conspicuous lack of hard data to support that claim, and lots of pages going back and forth and proclaiming at the 'conclusions' sessions of papers and chapters that the problem is indeed hard and the work done so far is at best inconclusive.