This ^ The effort required to change a process shouldn't be underestimated. Especially considered that industrial environments are already (a) automated for l

This ^ The effort required to change a process shouldn't be underestimated. Especially considered that industrial environments are already (a) automated for lowest-hanging fruit things (e.g. moving stuff around at human height) & (b) optimized around human capabilities for the remaining things. Substituting a not-humanlike robot would require reconfiguring a lot of existing automation around it. If you have "like a human, but costs less" that can be plugged into any existing still-human process? You can literally swap them in. Eventually we'll get to hyperoptimized machines, but an easier sales story to say "We automate your existing human processes."