People preferring uncensored bots is an unfounded, data-less claim.

There's probably a limit to how many times a bot can make up alternate facts before a user decides it's not worth asking anymore.

If OpenAI et al had positioned these generators as creative improv artists they might have found a niche market, but instead they presented it as an all-knowing oracle. Sure, they gave it a "may make false statements" disclaimer, but still the product they built is a helpful assistant that answers general knowledge questions with the accuracy of a coin toss.