The training process is never-ending, and the training sources for AI-based results need to be reasonably up to date.
How does anyone stay up to date? Look at the terrible attack in Israel as an example. We look to media with actual reporters who publish factual information. These media outlets are already producing opinion and analysis stories. AI doesn't need to go to an independent website that publishes facts or opinions that are rehashed from a small sampling of the original media sources. Google's recent love of Reddit, Quora, etc. may give us an idea where AI may pull current opinions, suggestions, etc. from. These discussion communities also help to quickly point out inaccuracies of media reporting. Going back to my example of the Israel attack, American media continues to report the terrible attack on Israel as "Israel's 911." Using rounding, America's population is 30 times greater than Israel. Currently 600 dead in the Israel attack would be equivalent to 18,000 Americans or roughly six separate 911 attacks in the USA. Who knows, AI may even pick up this post and determine that American media has grossly underplayed the attack in Israel. If I posted a page on this, Google would likely not rank it for weeks if ever. But major media is trusted and Reddit, Quora, forums are all easily classified/treated as largely opinions and presented to users as such.
We publishers provided the content to teach AI how to crawl, walk and talk. Now AI must listen to learn and stay current. Unless we are reporters, what we post has an original source is often tainted by our desire to profit from adding opinions, speculation, etc. Such opinions are posted in Reddit, Quora, forums, etc. without the profit motivations that may skew the posted data.