Agriculture is hard. Actually getting things to grow in the ground which produce enough enough excess energy to eat is a difficult problem! Our current industrial age agricultural approach is "bang the ground flat and work it with chemicals, machines, and row-upon-row of identical plants." It's inspired by the industrial revolution, by standardization, by quality control and by the fundamental machine-age paradigm of "identical products from identical processes." Every ear of corn alike!
However, there are radically more location-aware and intelligence-added approaches to agriculture, both ancient and modern. "Adaptive Agriculture" is agriculture that pays attention to every square meter of land, producing higher yields with less inputs.
Let's look at a historical example of adaptive agriculture first. The Inca were the first to cultivate the potato. Their model of cultivation is radically different from the factory farming model - the culture knew of around 3,000 varieties of potato, of which over 1200 are still in use, with 250-300 varieties being grown in a single plot [great link]. Because the Inca had to get every last calorie out of the ground, lived in the same places for many generations and had a stable culture to pass on wisdom about species and place, they bred crops to work well in every available niche and for as many purposes as possible. The practiced complex crop rotation and were, in general, great farmers. This is using human intelligence to assess local conditions to maximize agricultural productivity.
The high-tech big-yield model of adaptive agriculture is Precision Agriculture, also called Precision Farming. As farm plots have increased in size, the human knowledge about each hectare of land has fallen and fallen. Precision Agriculture seeks to remedy that by replacing human knowledge with machine knowledge. By using satellite imagine and GPS locators on tractors, yield monitors which measure meter-by-meter of land and dose-precise dispensers for fertilizer and pesticide, PA seeks to put knowledge back in the farming process. Each meter of the field gets individual treatment within an industrial farming context, reducing use of pesticides and fertilizers to the minimum-required levels for a given yield. Precision Agriculture leaves on thing invariant: what is in the field. It still works largely with the concept of monoculture as an invariant.
At the other end of the spectrum, we find the delightfully holistic practice of Seedballing. A seed ball is an earth-and-clay ball with seeds and fertilizer added. Each seed ball contains a mixture of different varieties and different species. even a child can prepare them. When planting time comes, the seed balls are scattered on the land, each falling in place at random. In each location whichever seeds are best suited thrive in their protected mud starter-home, so the field becomes a non-uniform melange of intermixed species.
In seedballing, the intelligence is not human, but evolutionary. By seeding many different possibilities, and letting nature winnow them down to what works and what does not, seedballing puts the right plant in the right place without heavy technical infrastructure. It puts the intelligence back into agriculture in a profoundly different way.
Global scale food production is likely to require the great-grand-children of current Precision Farming approaches. But it's a loveless post-industrial practice. Seedballing, at least to me, has a beautiful and delightful human-scale rightness. Even if it won't feed the planet, it's still beautiful and inspiring. Perhaps future Precision Farmers will use the tools of their ancestors, but have the aesthetics of seedballers, letting each plant grow in it's right place.
Variations on the latter approach are quite common in Permaculture and organic agriculture, for example the Fukuoka's method (One Straw Revolution) is based on broadcasting a mix of seeds, and I've heard of seed-saving networks (e.g. in India) distributing mixed seeds so that each farmer will - after a season or two - end up with a different crop optimised to their location. From what I've heard, yields of this mix quickly surpassed the one-genetic-clone type of seed bought from the big companies, because that could only be optimised for a single situation.
As tiny unmanned aircraft are developed, it would be nice if robots could be used to cultivate and harvest like bees. This could lessen the need for monoculture.