> Early tractors didn't do much for underdeveloped nations. Mainframe computers didn't either. The default state of capital-intensive technology is to improve p

> Early tractors didn't do much for underdeveloped nations. Mainframe computers didn't either. The default state of capital-intensive technology is to improve productivity, but not to bring capital and work into underdeveloped nations. Whatever benefits accrue there are the diffuse rewards of expanding global production. When early tractors were being deployed, all regions and countries were quite undeveloped in agricultural mechanization, compared to today. Likewise, when early computers were being created, all countries had primitive infomation technology. The default state of capital intensive technology is to be deployed first where capital is relatively cheap and labour is relatively expensive. The "diffuse rewards" of expanding global production are pretty much the whole point, since the magnitude of those rewards are huge. The recent economic boom of much of the less-industrialized world was characterized by (compared to the industrialized world) labour-intensive production, true - however, it was much less labor intensive than had been the case before. It was absolutely characterized by people adopting more mechanization, more automation, more technology. So this is why it doesn't make sense to me to think that a high-productivity technology will stall growth. If a t-shirt robot is more expensive than a Bangladeshi textile worker's wage, then we'll see continued high use of human labour in that industry. If the robot becomes cheaper, then the Bangladeshi textile workers will shift to other industries (as they mostly did when they went from being farmers and handcraft workers to being textile workers, over the last few decades.)