With our current models they definitely are at odds. There is a strong correlation between higher average performance and lower explainability when comparing machine learning approaches like neural nets, linear models, decision trees, etc for nearly all tasks.