It's an interesting experiment, but I've noticed something about many things we call "fallacious". They're not lies, they're just heuristics, which in fact, in aggregate, help reach more correct conclusions sometimes.

Say an argument from authority. Sure, some guy who digs dirt may know more about building skyscrapers than an architect who spend 40 years doing it. But, still?

So in that context, I hope we start recognizing what fallacies are. They may not match standards of formal logic, but also if formal logic worked then we wouldn't need to grow artificial intelligence from noise, we'd instead feed it rational thought models and then it it comes up with flawless conclusions. Which is how we imagined AI for decades prior and all out attempts failed. They failed because formal models are highly restrictive and throw the baby out with the bathwater.