Citation needed on there being less ridiculous patents being filed, because I doubt it.
As far as Google Search goes, standard advice for any tech company to their employees is don't go searching for patents, just submit potentially patent-able ideas to legal. This is because if someone has a patent on something and you violate it and they can prove that you knew about the patent via subpoenaed records then you owe them three times as much as you would otherwise, so the more ignorant you are of existing patents the better.
And on the legal side of things, IP lawyers just search the USPTO records directly, not through Google Search.
I've read hundreds of patents over the past 7 years or so, so the first citation is simply my experience. You won't find, for example, a broad patent on something like bytecode like this one https://www.google.com/patents/US4667290 these days
However, I've been researching the patent system quite a bit, and have come across a number of studies about patent quality [1-12], including "software" patents [1, 2, 3, finding that software patents are no worse than other patents]. Unfortunately, there are many problems with these studies:
1) Judging patent quality is hard and time-consuming, and doing it for a significant number of patents to get data is much more so. Hence many studies use proxies to judge patent quality, such as back-citations [4, 5], or more complex multi-variate models [10] . And then other articles point out flaws in using proxies like this, or actually showing that the metrics actually show the opposite of what they conclude [6]. And then other studies that propose new metrics [7, 8, 9]
2) Related to 1, the other problem is looking at not enough data, looking at different subsets of data, or using different methodologies. Studies are usually careful to highlight these shortcomings and try to logically make an argument for generalizing these results, but it's hard to reach a general conclusion, especially since there is no real consensus.
Now, back to your [Citation Needed]: Few studies convincingly evaluate USPTO (or other) patent quality over time. One study that does find an improvement over time [10] attributes it to the KSR decision rather than Google. Another says that EU patents are getting better over time, though not the US [11]. Another one does briefly argue that US patent quality has been increasing over time [12] based on a bunch of other studies, but some of these use metrics that are not fool-proof, as I mentioned in point 1).
And no, patent lawyers and examiners do absolutely use Google for search.
