You are over-simplifying green chemistry. While you do not define "using/learning/controlling Natural Interfering Systems", I interpret that to mean that the chemistry should integrate with (as opposed to interfere with) natural systems. Green chemistry is more than just minimizing toxicity in the product. Green chemistry is also about recognizing the hazards associated with starting materials, solvents, and other inputs, both in terms of the characteristics of those inputs (e.g. a toxic solvent) and the up-stream footprint to make them (e.g. resource extraction).
Green chemistry includes consideration of the reaction conditions and minimizing the energy use and safety risk (e.g. avoid potentially explosive conditions).
You are right that it is very difficult to predict how a particular substance will intereact with "expanded biological space", but chemists can consider some well-understood properties in their molecular design. What they design may not be perfectly green, but it can be a big improvement over what it's replacing.
There are numerous groups working to improve in silico toxicity measurements so that it becomes easier and cheaper to predict potential toxic effects based on structure rather than relying solely on in vivo testing after the molecule is designed and synthesized.
