Hunger

Combating World Hunger with
Machine Intelligence

World hunger is one of the major problems of our time. With the constant growth of the earth’s population, compounded by climate change and exceptional events like the COVID-19 pandemic and natural catastrophes, the situation is expected to become worse. Machine Intelligence (MI) can help tackle this challenge. 

Around 700 million people, or 9.2% of the world’s population, currently suffer from severe “food insecurity” (which is defined as the absence of reliable access to a sufficient quantity of affordable, nutritious food) [1]. At the same time, the world’s population continues to grow and, according to the UN, it will reach 9.7 billion by the year 2050 [2], which will result in a food demand 69% higher (measured in calories) than in 2006 [3]. Simply increasing the amount of arable farmland in the world will not solve the problem because the planet’s existing agricultural systems are already under extreme stress: agriculture makes up around 18% of global CO2 emissions [4] and uses up to half of the world’s habitable land [5], with devastating consequences for ecosystems, water supplies, and biodiversity.

Global land use for food production

Global land use for food production. Agriculture uses up to half of the world’s habitable land [5].

The solution to world hunger is not straightforward. It should have the following goals:

 

  • Reducing existing hunger,

  • Preventing future hunger outbreaks,

  • Finding a solution which is both sustainable and climate friendly.

 

In each of these goals MI can play a decisive role in the success of the solution. Let us consider some examples.

 

Reducing existing hunger

 

Providing those in need with food is an important short-term measure, but it does not tackle the problem at its source. A more sustainable solution would be to help agriculture in regions suffering from food shortage evolve into a self-supporting state. In fact, agricultural development can be one of the most powerful tools in combating hunger: according to the World Bank, growth in the agricultural sector can be two to four times more effective in increasing incomes among the poorest than growth in other sectors of economy [6]. There are several initiatives currently making a significant effort in developing local agriculture. Some of them already employ MI technologies.

 

For example, the German NPO Thriving Green combats hunger and malnutrition following an innovative concept [7]. They teach and help locals to cultivate a crop called microalga Spirulina, a nutritious superfood which can be harvested daily, delivers high yields, and requires only a small amount of water. Since it prefers high temperatures, spirulina is perfectly suited to nourishing people who live in deserts. In cultivating spirulina, Thriving Green provides locals with a simple and robust technology which includes an Artificial Intelligence (AI) component: The health condition of the spirulina population is observed by a camera connected to a small computer. This computer runs an AI model that was designed to identify sick microalga and alert the farmer when disease is present. In this way, the farmer can take action to stop the spread of the disease early enough to save most of the valuable harvest.

 

Indeed, one of the most promising uses of MI in agriculture is for the identification of plant diseases. Potential applications of MI in this area have become a topic of intensive research. For example, scientists in Switzerland and the US have been working on using AI to identify diseased plants from photos shot on smartphones [8]. This approach allows farmers to identify diseased plants early and to act in time to increase their yields. In fact, this research project was so successful that it has triggered the development of a comprehensive AI-based software called PlantVillage [9]. The software is not only able to identify plant diseases with a precision comparable to that of a human expert, but it also uses on-ground and satellite data to forecast events that can impact the yield of a farm. Furthermore, PlantVillage can automatically respond to questions posed by farmers using an AI-based human language comprehension system and a large open access library on crop health.

 

Other companies involved in the fight against global hunger have appreciated the power of their own data and the technologies which are capable to evaluate them: Companies like myAgro, Farmerline, and TechnoServe, which are dedicated to developing local agricultural systems in poor regions, have collaborated with the US-based non-profit startup Delta Analytics in order to better understand their data and to display them on dashboards that are easily understandable for humans [10]. Such advanced data analytics projects can provide important insights about the impact of various activities in the portfolio of an NPO and discover potentials for the optimization of internal processes and supply chains.

 

 

Preventing future hunger outbreaks

 

Preventing future hunger outbreaks is an even more challenging goal than that of reducing existing hunger. In fact, we will not be able to solve it in a sustainable fashion without some innovation and the use of new technologies. The good news is that there are already several promising MI approaches that can help tackle this issue.

 

For example, an interdisciplinary team of researchers at Carnegie Mellon University is running a program called FarmView [11]. This program involves exploring and deploying an innovative system of sensors, robotics and AI techniques to improve plant breeding and crop management. The focus of the program is the development of automated, data-driven decision tools to increase the yield of sorghum, a drought- and heat-tolerant grain that is often used in famine-prone regions. On test fields, drones, robots, and stationary sensors collect data about observable features of sorghum crops. This data is then analyzed by AI algorithms in such a way that breeders and geneticists can choose the plant’s traits that are most suitable for improving yield and make the plants most resistant to disease and drought. Using this approach, new, very fertile and resistant strains of sorghum are bred and provided to farmers in developing countries who can increase the yield of their crops while still using their traditional farming practices.

 

Another action necessary to provide enough food for all is to avoid wasting food that is already available. Instead of trashing food because it is abnormal in size, color or appearance, it can be donated to food banks or used in the production of processed food. For example, tomatoes which are not good enough for supermarket sale can be used to produce ketchup and unsellable potatoes can be used to make French fries or chips. Indeed, we are already on the way to minimizing waste in food production and companies like TOMRA already provide us with high-tech food sorting solutions, which use advanced analytical techniques for this purpose [12]. The next step will be to apply these technologies on a large scale and to enable developing countries to benefit from them too.

 

Looking at the above examples of how MI technologies can transform agriculture, one could imagine a holistic MI system which continuously observes farmland and crops, collects and evaluates data and simultaneously optimizes yield, resource consumption, and environmental impact. And indeed, there are already attempts to develop such systems, e.g., that of the Israel-based startup Prospera [13]. Even though such solutions are not currently implemented on a large scale, it is only a matter of time and investment until our agricultural systems are completely revolutionized. This may sound like good news, but we should keep in mind that such technologies should not only be accessible to large corporations from the developed world but must be made available also to developing countries. Only in this way we can prevent future outbreaks of hunger.

 

 

Agriculture and environment-friendly solutions

 

While it is only a matter of time before a revolution in technology-driven agriculture happens, we should ensure that when it does, it does not put more of a strain on the environment than the agricultural systems it replaces already do. The environment can also benefit from the power of MI technologies in agriculture, as it should.

 

As an example, MI technologies can help to considerably reduce the use of herbicides. Currently, herbicide use is akin to “carpet bombing”: the whole field is treated with herbicides. This approach leads to the emergence of herbicide-resistant weeds and creates an unnecessary strain on the environment. A company called Blue River Technology wants to change this situation [14]. It provides farmers with smart agricultural equipment which uses AI-powered computer vision technology to identify weeds. This equipment can spray herbicides only where they are really needed, avoiding “carpet bombing”. Thereby, the amount of herbicides used can be reduced by about 90%. This approach is good for both farmers’ wallets and the environment.

Machine Intelligence identifying weeds

Computer vision system developed by Blue River Technology in action. On this photo, the system distinguishes between cotton plants and weeds [14].

A new but promising alternative to conventional farming is so-called vertical farming, in which plants are grown indoors in vertically stacked layers [15]. Although vertical farming is very expensive and consumes a lot of energy, it does reduce the amount of farmland, thus saving a lot of natural resources, and it is less disruptive to sensitive ecosystems. Therefore, it should be considered as a promising sustainable alternative to conventional farming and it could be a part of a future solution to hunger and environmental problems. Because vertical farming plants grow in a controlled environment that requires continuously measuring humidity, light, temperature and so on, thus generating a large amount of data, there is a lot of potential to apply MI technologies in this field. Not surprisingly, the market leader in vertical farming, AeroFarms, already utilizes AI, predictive analytics, machine vision, and data science to optimize resource consumption, ensure quality and improve yield [16].

 

Another opportunity to reduce the impact on the environment through agriculture and food supply is by reducing the consumption of meat and animal products, since the livestock sector is responsible for 14.5% of human-induced greenhouse gas emissions [17]. However, for most people it is very difficult to go without animal products. MI can also be helpful in this field: a Chilean startup NotCo specializes in using AI to produce plant-based foods that look, smell, and taste like their animal-based counterparts [18]. The central role in this startup is played by an AI algorithm that its founders call Giuseppe. It has already been trained on an enormous number of plant combinations used to replicate animal products in taste and other characteristics. Currently NotCo provides its customers with vegan milk, ice cream, mayonnaise, and burger patties that were developed with the help of MI.

 

 

Conclusion

 

Innovative approaches that utilize the power of Machine Intelligence can be employed in the battle against hunger. In fact, we are already well on our way to revolutionizing agriculture using MI techniques. However, we need to ensure that this revolution is a sustainable one which positively impacts agricultural systems in developing countries. Thus, think creatively and use the power of modern technologies in your organization to fight hunger [19].

References

 

[1] https://undocs.org/en/E/2020/57.

[2] https://population.un.org/wpp/Download/Standard/Population/.

[3] https://www.wri.org/blog/2013/12/global-food-challenge-explained-18-graphics.

[4] https://ourworldindata.org/emissions-by-sector.

[5] https://ourworldindata.org/land-use#breakdown-of-global-land-use-today.

[6] http://documents1.worldbank.org/curated/en/700061468334490682/pdf/95768-REVISED-WP-PUBLIC-Box391467B-Ending-Poverty-and-Hunger-by-2030-FINAL.pdf.

[7] https://www.thriving-green.com/en/.

[8] https://arxiv.org/abs/1604.03169.

[9] https://plantvillage.psu.edu/.

[10] http://www.deltanalytics.org/past-grant-recipients.html.

[11] https://www.cmu.edu/work-that-matters/farmview.

[12] https://www.tomra.com/en/sorting/food.

[13] https://home.prospera.ag/.

[14] http://www.bluerivertechnology.com/.

[15] https://en.wikipedia.org/wiki/Vertical_farming#:~:text=Vertical%20farming%20is%20the%20practice,hydroponics%2C%20aquaponics%2C%20and%20aeroponics..

[16] https://aerofarms.com/.

[17] http://www.fao.org/news/story/en/item/197623/icode/#:~:text=Total%20emissions%20from%20global%20livestock,of%20all%20anthropogenic%20GHG%20emissions.&text=On%20a%20commodity%2Dbasis%2C%20beef,the%20sector's%20overall%20GHG%20outputs..

[18] https://notco.com/us/.

[19] We, at MI4People, are eager to collaborate with you to help you understand the potential of MI and identify impactful research projects that we can undertake to find MI based solutions that help deliver Public Good better as well as assist you in meeting your organization's goals more effectively.

https://www.mi4people.org/contributetomi4people.