An important milestone in my opinion because this is where AI can start to auto develop itself with real world feedback, leading to auto developing (Robot-ai) Systems.
This is bad news unless we develop strategies to make it right. Even if such robots actions where limited to manufacturing, auto developed robots can leed to a lot of industrial accidents, so, many workers could be injured.
To prevent injuries i would see, in near future, factories where humans are excluded. Which would lead to an even greater risk : auto developing hostile systems.
A strategy to control this risk will be, in my opinion, to have at least two competing (Robot-ai) industrial complex that would supervise each other’s work so to allow humans ultimate supervision.
Whatever will happen, beyond my forecasting, will be, i believe, much more complicated than what we have today.