in GMO, you're making a very deliberate change to some 200 (or less) nucleotides, and you know exactly what that change does, because you've already observed its results before putting it on the market.
Oh, really? Please tell us then why the Golden Rice (the GM rice with a transgene for pro-vitamin A)
(a) produces large amounts of zeaxanthine
(b) the maturation period is lengthened;
(c) grain yield is reduced.
none of these traits were known until this rice was produced.
Need more examples? Take the Bt gene in GM crops. The Bt toxin is known to exude from the roots of Bt rice and Bt corn, but NOT from the roots of Bt cotton and Bt canola.
Yet another example of your EXACT predictability: the transgene for pectin synthesis does no harm in GM tobacco containing it, but in the GM apple with the same transgene, premature leaf shedding is reported.
And then there are a plethora of examples of gene silencing, gene overexpression, and variable expression of the transgene in different tissues of the same plant. Enough with the technophilic triumphalism and the hubris of "Exact knowledge", "precision" and "predictability", already.
You may be interested to read all references to high-profile scientific literature, in my article "Genetic engineering in agriculture: Uncertainties and risks" [in GMO Food: A Reference Handbook (ed. D. Newton), ABC-Clio: Santa Barbara/Denver/Oxford (2014)].