1. Herbicide resistance. We now have weeds that are simultaneously resistant to many different modes of action. Even if one were able to eliminate the impact of pesticides on the environment, it is impossible to develop new chemicals and stacked herbicides faster than evolution can find ways to work around them.
2. There are other drawbacks to monoculture ag, namely soil loss, lower quality of soil structure/microbiota, loss of ecosystem services such as habitat, etc. Increasing our yields per acre is great but we need to factor in all of the externalities. I was glad to see mention of finding new ways to associate rhizobia with cereals, but that's a long way off and for the time being no one has figured out how to have agriculture avoid massive imports (even in organic ag) of nutrients into the system.
3. Exclusive focus on the biggest staple crops, with far more emphasis on quantity than quality and producing food that is necessary for a well-rounded diet.
4. Nearly all of these solutions ignore the possibility of using ecological interactions, unique spatial designs, intercropping, perennials and other agroecological strategies to reduce weed pressure, increase nitrogen fixation etc. These other techniques are far harder and are more difficult to commercialize, but if one wants to tackle "hard" problems in agriculture, those would be the ones to address.
Just to be clear, much good will come out of the technological developments, but for the moment they're primarily tinkering around the edges when really we need a whole-systems redesign for agriculture. I recently finished a PhD focused on optimizing decisions given all of the uncertainty/variability in precision ag systems, and what I ultimately concluded (at least for row crops) is we still have a long way to go.