When reading statistics like this, though, it's important to ask two things:
1. If the report lists how many jobs were lost due to automation, does it also talk about how many were created due to automation? If 20k are lost one year, and 19k are created, the article may not mention the latter, though it is relevant. In one sense, only the delta is important.
2. But, in another sense, the delta isn't what we care about. New jobs created by automation is misleading too. Because, are the jobs created by automation better than the ones that were lost? Is it the same people who lost their manufacturing jobs who got the new jobs? Are those people happy about the trade?
My point is that it's a complicated situation that has been (and likely will be) commonly oversimplified.