The best solution, in my humble opinion, is to focus on making AI stay as close to the unfiltered objective truth as possible

There's a ton of problems with that that make it unlikely to be possible, starting with the fact that genAI does not have judgement. Even if it did, it has no way of determining what the "unfiltered objective truth" of anything is.

The real solution is to recognize the limits of what the tool can do, and don't ask it to do what it's not capable of doing, such as making judgements or determining truth.