I think so but the problem is with the raw processing power tor train these data sets, the computing/storage costs alone disincentivize open source development.
Even so I'm still cautiously optimistic.
Follow
@basedbagel @Chimi_Chuang_Tzu you can always rent a server with a shit ton of cores and ram and train it yourself and then ask it questions, there is a vid on youtube about a guy who did exactly that and trained it on a 4 yr 4chan database the crasy shit that ensued.
actually p interesting, even if just from a "how easily can pol get psyopped by 1 (one) botposter" angle