You're making a fallacy here, forcing the issue into black and white - if you mention first 'dataset bias', it must mean you 'offer some superficial causal mechanism that sounds like an explanation to that public'