Having said that, how many people blindly trust Google, the internet is a cess pit of data and people often play fast and loose with it. Years ago when I was teaching there were 2 sites mlk. Org and mlk. Com
One was an authoritative quality site, the other was borderline racist propaganda, Google indexed both of them…
Just saying, it’s like the old hifi adage, garbage in, garbage out.
We are just looking at the garbage from a different window!
And this is the problem I have with AI LLMs at the moment. The Internet is full of garbage and the search engines haven’t managed to filter it properly, but instead of taking AI LLM as a fresh starting point to actually put in only authoratative data, they just fed it “the whole internet”, garbage and all.
A lot of people claim that this is fine and “search results are never 100% accurate anyway, so why should AI be any more accurate?” 
One of the big differences for me, is that people have to go and read the source material to find the information they want, so anyone with an ounce of intelligence can usually work out fairly quickly, whether the site is reputable or not, or whether the answer on the linked site makes any sense.
With AI, it is giving you a summary of the information, with obfuscated links to the source material. Therefore people will assume it is correct, because it is presenting the information in a confident manner, if you don’t already have a vague idea of the answer, you won’t know whether the information is correct or not. You will still need to read those linked articles.
A couple of examples: my daughter and son-in-law are renovating the house they bought and regularly search for YouTube videos on how to accomplish certain tasks. It quickly became clear that a majority of the videos were useless, even misleading or dangerously wrong. In the end, we came down with a couple of independent creators that made good videos, plus the DIY stores and the product manfuacturers, who were making high quality videos with sensible, and safe, information.
Likewise, one of the first tests I did with Bing AI was a request on monetary limits for purchases in the EU. It was out by a factor of 10… On the side that would probably have me being probed by the local tax office, if I actually followed Bing’s advice!
This is similar to experiences I had with Google Translate a few years ago - it works now, because I put the correct translations in manually. But, at the time, Google had real problems with formal English, it could work with “don’t”, but give it “do not” and it would drop the “not” in the translation to German.
E.g.:
It was at a time when I was transitioning away from using Google services and I thought Google had gotten wind of it and was trying to bump me off! 
Google Translate looks amazing, unless you already know both languages - I had a tight deadline to translate an instruction manual from English to German & hoped Google Translate would help me speed up the process. It couldn’t, it was so inaccurate as to be dangerous and even what it got right was of such a low standard, you couldn’t use it in an user handbook, let alone an official document. (After I left the company where I was doing translations for the boss, I actually took an internship at a translation bureau, even with nearly 20 years of experience with German, my translation skills, which were way better than Google’s, were not good enough for working on official translations.
Interested to know what you think of the MS translation app, I have found it better than Google with French and Italian will have to find a German to try it on. I had someone reject some Welsh translation because they said it was coming up with stupid things - I had to explain that microwave oven on Welsh actually is pobolyping 
I don’t think it would be too hard to do an aggregated result and work with that
Paul Smith-Keitley
Adobe Creative Educator
Just tried it with your post and my post. It gets a lot more right than I expected, but it is inconsistent and gets some things wrong “bump me off” was translated as “annoy me”, for example!
In my text, it kept switching between KI and AI for AI, KI is the correct term. There were other inconsistencies and the wrong use of words based on context. Overall, I’d give it a 6/10.
Thanks for trying it, I wonder if it drops the not like Google
Paul Smith-Keitley
Adobe Creative Educator
No, it had no problems with the “not” - I fed it all of my post, including the examples, it translated them all correctly. Unfortunately, because I corrected Google all those years ago, I can’t use it to show the mistakes it made back then…