How does the data that fuels artificial intelligence models impact accuracy and utility? In other words, can we trust it to provide useful intelligence?