> injecting the genome of deep sea fish species
You inject genes, not genomes. Injecting genomes is how breeding works, not how genetic engineering works.
> to produce new chemical elements
New elements are exclusively the domain of stars, nuclear physics, and particle accelerators. They have nothing to do with biology. I'll continue on the assumption that you meant "molecules."
> no fucking idea of what they do
It's the other way around: in selective breeding you're trying random genomes until you get literally anything that works. It would take tons of additional study to figure out why it worked, how it worked, and what the side effects are. Contrast to genetic engineering where you start with a very specific biological goal in mind, often the production of a single molecule or small set of molecules whose structure and characteristics have been extensively studied.
Suppose you set out to produce more Bt protein in the plant's leaves so that it's toxic to insects while the competition breeds a plant that seems to be similarly resistant to insects. Which plant is safer? In both cases you have to do a study to find out, but in the case of the GMO you can use ultra high sensitivity studies because you can purify Bt and test its effects on mice and humans. In contrast, the plant produced by breeding would have to have an absurdly huge sample size to achieve the same sensitivity. Historically, plant breeders don't bother.
Both approaches are almost always good enough, but of the two genetic engineering allows for much more precise risk prediction and characterization. That's one of genetic engineering's strong points, not weak points.