The app's design is to get you to mindlessly doomscroll and not really think too hard about what it's showing you. If it shows you something insane occasionally it's no big deal, it's just the algorithm trying something new right?
It's very difficult to show any concrete evidence for how a secretive algorithm controlled by an adversary behaves. In an ideal world the burden should be on the platforms to prove that their algorithms are fair and not biasing towards certain viewpoints, but that might never happen.