The problem and solution are one and the same. Google is trying to automate tasks that are nowhere near ready for automation.
This doesn't even really make any sense either. You might argue its because they're huge, but the numbers don't add up. Google netted about $31 billion in 2018. It's hard to tell exactly how many apps are published on Google Play per year, but it seems to be in the ballpark of 200,000 [1]. Let's hire some low wage employees to spend just 1 hour per app which is generally enough time to determine if an app is at all legitimate. That's obviously 200,000 man hours. Creating these jobs in the USA could be done for about $10/hour, or $2 million, or a minuscule fraction of 1% of their their annual net income. Problem solved, happy customers, happy developers, jobs created.
But Google gonna Google. Though even here if they really wanted to Google, they could even hire these guys in e.g. India or wherever and pay a fraction of even that fraction of what they'd have to pay if they created these jobs in the US. But I'm certain they've Googled that the $ cost of pissed off customers and developers is less than the $ cost of providing a decent solution.
The one confounding issue here might be that the 200k is lowballing the number of apps. After all that's only the apps that are published, not the ones that are refused publication. But all of the above holds true even if you push it up to 2,000,000 apps trying to publish per year. Perhaps the more salient point is that Google also charges $25 before you can even try to publish an app. Why not direct that money to a human being and actually have a half decent review process? That same human being might even be able to answer complex questions like, "Hey, I don't even have any ads in my app. Why was my app removed for deceptive ads?" in a timely fashion.
[1] - https://www.statista.com/statistics/266210/number-of-availab...