It gives us a higher chance of creating safe AI. AIs trained on open data are more likely to be neutral and trustworthy instead of biased by the interests of the corporation who created and trained them.