Food security: Can AI and gene editing tackle global hunger?

Producing what we need to eat puts an enormous burden on both the climate and the environment. Yet, at the same time, crop failures caused by extreme storms, drought and heat waves threaten the food security of a growing global population. 

In a clear indication of the dilemma we face, a World Resource Institute (WRI) report recently said that to feed the world's population, humans must produce more, but without increased use of resources and land. 

"There's a 50% gap between the food produced today and what we're going to need in 2050, just to feed people adequately," said Janet Ranganathan, a nutrition expert at WRI and co-author of the report, adding that expanding the agricultural frontier would mean "saying goodbye to the remaining natural ecosystems."

One way of meeting this massive challenge would be to use the land currently given over to meat production and crops for animal feed to plants for human consumption. But this would require everyone becoming vegetarian, which is an unrealistic scenario. Enter artificial intelligence.

Will there soon be drought-resistant rice paddies?

Scientists want to use AI and what is known as the CRISPR-CAS9 gene scissors to develop climate-resistant supercultures capable of delivering higher yields with fewer resources. They do this by modifying the plants' genes using a method called genome editing.

Rice offers a perfect field for experimentation. Traditionally a very thirsty plant that grows submerged in water, rice has been hit hard by extreme drought from Italy to China and Pakistan. A new variety called IR64 could help. It grows primarily in Asia and parts of Africa but is sold throughout the world.

Gene modification has made the plant more drought-resistant. In some weeks, it needs up to 40% less water than before. And while the mother plant died after a week without water, half of the modified plants survived.

Gene scissors: Revolution or danger? 

Genome editing is fundamentally different from traditional genetic engineering. "It actually relies on natural processes. But it makes the mutation process much less random," said Detlef Weigel, a biologist at the Max Planck Society in Germany. 

Most genetically modified products, whether animals or plants, have been implanted with an artificial gene or a gene from another organism. Insect-resistant cotton or corn, for example, contain a gene that originally came from a bacterium.

Instead of using foreign DNA, gene editing can change the genetic code using an organism's own DNA. With special enzymes that work like scissors, genes from the plant can be deleted, swapped or repeated. 

It would take many dozens of breeding generations to transfer a single gene by natural crossing.  And this would simply take too long to be viable, Weigel explained. "So genome editing is really super powerful because you can go in the single gene, change it. And voila!” 

While crossbreeding can take more than ten years to get the desired result, gene editing takes only a few months and the testing phase a few years.

Risks and hype — are 'smart bananas' a good idea? 

And it's not just about drought-resistant rice. Some studies show that it's possible to increase tomato yields by up to 70%. Others are trying to grow soybeans in barren and saline soils or reduce methane emissions from rice. And Kenyan scientists are developing what they call a "smart banana." In their laboratory, they have succeeded in activating a gene that switches on the plant's own immune system as a preventive measure against a virus that becomes active during drought.   

None of these methods are completely free from risks and uncertainties, however, mainly because many of the projects plants are still in the research phase and there isn't enough data. But opponents of genetic research criticize gene editing as a dangerous experiment with nature.  

Some experts point to cases in which genetic changes occurred unintentionally, or in which much more genetic information was deleted than had been planned.  

In addition, genes that help increase yields in certain droughts cause them to decrease in wet years. And since a large number of genes are involved in these traits, it is usually not enough to simply switch one or two them on or off.

How artificial intelligence could help 

The less optimized a crop, the easier it is to improve. As a result, CRISPR offers the greatest potential for old varieties that have not yet been cultivated and bred on an industrial scale.   

For example, millets, einkorn and yuca are already more resistant to climate change,but breeding to make them largely viable for mass production is still in its infancy.  

"We can very quickly bring those crops to a place where they're agronomically feasible to use, and therefore start to diversify our food system further," said William Pelton, CEO of the start-up Phytofrom.  

Using artificial intelligence, Phytoform is trying to identify even more optimization opportunities in genes. Their algorithms can quickly process volumes of data that would take an individual person years to achieve. The technology has advanced to the point where some algorithms already understand DNA datasets much better than humans. 

"So it can start to spot regions that are repeated and therefore start to derive meaning," said Pelton. "And, of course, that means it can then understand the DNA, but then it can also suggest changes that could be made in order to affect an outcome. " 

Phytoform is currently working on a potato that doesn't turn brown when bashed or bruised, which could lead to fewer potatoes being thrown away even though they are still edible.  They are also working on lupines, which have been around for thousands of years but are rarely seen on supermarket shelves.  

Lupines are very high in protein and nutrients and could be more widely used for plant-based meat. Phytoform's algorithms are figuring out how to make the crop more productive and solve quality problems.  

The world prepares for gene scissors 

Research into gene-edited plants is gathering pace across the world.  While just a handful of patents were filed in 2011, by 2019 there were almost 2,000, most of them from private companies or public research institutions. 

With the US and China, as well as multinational companies, investing heavily in the technology, a multi-billion dollar market is expected to emerge by the end of the decade. 

In the EU, gene-edited crops are labeled as gene-modified and are therefore strictly regulated, though proponents say it would be more accurate to talk in terms of a new breeding method as opposed to genetic manipulation. 

Meanwhile, in the US, China and many Latin American countries, genome-edited crops do not have to be labeled or controlled as genetically modified, and the sector is set to launch a number of crops in the coming years. India has also recently relaxed its regulations. 

But regardless of rules and however advanced the latest methods, conventional breeding will continue to play a vital role in feeding the global population. 

This article was originally published in German.