Just had to redo an entire experiment because the automated data processing software "optimized" the results in a way that was clearly not biologically plausible. Can we please slow down on the automation hype and make sure these tools are actually doing what we need them to do?