We should assume people will use tools in the manner that they have been sold those tools yes.
But these tools include research like this. This research is sold as proof that AI models have problems with bias. So by your reasoning I'd expect doctors to be wary of AI models.