"That would amount to more than 1,300 cuts across 55 U.S. fulfillment centers for standard-sized inventory." ... "removing at least 24 roles at each one, these people said. These facilities typically employ more than 2,000 people." I'm surprised that number is so low. There's only a couple dozen people out of 2000+ that actually put orders in boxes?
After 6 months or a year, the lessons learned about what worked and didn't will get rolled into the next round of automation, or an entirely different approach will be taken instead.
I used to work at a water filter factory, and this is exactly what happened there. One production line used the old, 7 person crew, and a new, more automated one was brought in for a different filter size that needed only 2 or 3 people. The ways that the two lines could cause batches to fail varied, and the company gave it some time to figure out which line ended up being more cost effective over X investment payoff period.
I didn't stick around long enough to see which one "won" out, but the automated process wasn't a clear winner by the time I had left.