> If AI really helped like it expected to, you would grab any dev you could so that you could have an army of 100x devs.
This seems maybe a bit reductionist.
AI will have diminishing returns because at a certain point, coding is not the bottleneck and coordination is or some other thing that hasn’t been optimized yet. The exact bottleneck seems like it depends on the organization.
My theory is that in general, augmented devs are much more productive, but 100% of that gain doesn’t translate into 100% more software delivered to customers, and there is a point where coding isn’t the longest pole.
But I don’t think most orgs are at that break even yet, and I think we can still get more out of engineering before we plateau.
This seems maybe a bit reductionist.
AI will have diminishing returns because at a certain point, coding is not the bottleneck and coordination is or some other thing that hasn’t been optimized yet. The exact bottleneck seems like it depends on the organization.
My theory is that in general, augmented devs are much more productive, but 100% of that gain doesn’t translate into 100% more software delivered to customers, and there is a point where coding isn’t the longest pole.
But I don’t think most orgs are at that break even yet, and I think we can still get more out of engineering before we plateau.