What potential risks can you think of for this approach?
A poorly designed dsRNA that has too much homology to a gene of a related beneficial species could impact it if it took up the dsRNA. dsRNA doesn't hang around in the environment, so the beneficial organism would have to be there at the time of spraying. For these reason, frameworks are being put in place to ensure due diligence is done in the design phase.
Like any pesticide, what the active ingredient is applied with may not be terribly nice (surfactants and the like). That's a case-by-case thing, which is hard to generalize.
Roundup is a synthetic herbicide - that they both could be used in agriculture is when the similarity ends.
At some point, you take calculated risks, because there is no action in life that comes without risk. Even your preferred solution of eliminating all chemicals and switching to alternative farming practices comes with substantial risk: it’s very likely it won’t scale. Millions would starve.
Organisms may be able to handle everyday quantities of dsRNA, but beyond a certain concentration, you may suddenly see unexpected problems, which then cause domino effects throughout the ecosystem. These kinds of relationships are unknown unknowns, and they rarely behave linearly.
I really like the RNA-based approach, don't get me wrong, but the other commenters here are right to point out that hubris is dangerous.