AI-Assisted                   

Agricultural Advisories

  • 21% yield increase for small scale farmers.

  • 1 billion dollars in additional economic output in 12 months

  • call for collaboration between UpliftAI, University of Agriculture Faisalabad, and Government of Pakistan.

Motivation

Low literacy and education rates limit agricultural yield. AI can mitigate this by providing expert quality, personalized advice to every farmer

Pakistan, with its abundant natural resources, strategic trade location, and a young, energetic population, holds the ingredients for great economic growth. Yet, this potential is severely constrained by a knowledge access crisis: 42% of adults are illiterate, and a concerning 77% haven't completed middle school education - a critical milestone for developing essential skills in information seeking and analysis. As a result, a large segment of the adult population struggles to acquire relevant knowledge or derive effective, personalized solutions that can improve their individual productivity and in return improve our economy.

For instance, in the agricultural sector, farmers find it particularly challenging to access the knowledge necessary to adapt to the rapidly changing environmental conditions affecting their crops – resulting in agriculture yields that are lowest amongst countries with similar climate and infrastructure. According to data from US Department of Agriculture, Foreign Agriculture Service, compared to Pakistan, the average yield per acre in India, Bangladesh and Uzbekistan is 35% higher for rice, 12% higher for cotton, 29% higher for wheat, 24% higher for corn, 70% higher for peanuts, 87% higher for soybeans and 215% higher for millets.

Government and private sector have tried to mitigate this issue through phone helplines, information campaigns and other outreach mechanisms. However, the effectiveness has been limited as such programs are hard to scale due to scarcity of experts who can provide high-quality personalized advice to each farmer. Historically this has been an unbreakable self-perpetuating cycle: to break this cycle, we urgently need experts to educate and upskill the farmers, but the system itself struggles to produce and retain enough farming experts.

Recent breakthroughs in artificial intelligence (AI) offer a scalable solution to break this cycle.

Solution

AI-assisted agriculture advisories: AI provides farmers with personalized agricultural advice in local language via voice notes, images, and videos

Recent advancements in artificial intelligence (AI) have made it possible for computers to understand languages spoken in Pakistan, including Urdu, Punjabi, Pashto, Sindhi, and Balochi, with remarkable accuracy across a broad spectrum of dialects and accents. This technical breakthrough opens the door to voice-based knowledge access systems for farmers.

Additionally, AI models can now be trained on vast amounts of data to replicate the knowledge and problem-solving strategies of human experts. This means that AI systems can provide personalized, high-quality step-by-step guidance and instruction tailored to the specific needs of individuals, addressing the shortage of experts.

By putting these technologies together we can create AI-assisted farming advisories: Farmers interact by sending a voice note in their native language to a WhatsApp number, and the AI replies in the same language with high quality advice. When suitable, it may also provide images or videos for enhanced clarity and effectiveness. If the AI doesn’t know the answer with high confidence, it shares the question with a trained human expert.

Such voice based AI-assisted systems bypass the hurdles caused by illiteracy, lack of advanced education and lack of sufficient access to expert advisors.

Proof of Effectiveness

AI assisted agricultural advisory increased yield by 21% in Telangana, India

A program called Saagu Baagu engaged 7,000 small-farm growers of chili peppers. Farmers were provided with AI-based tools including a Whatsapp chatbot that provided farming advice and information in the local language. The program also provided the farmers with soil testing facilities. The pilot program lasted 18 months, or three cycles of planting, growing, and harvesting peppers. Farmers in the program grew 21% more plants per acre while using 9 percent less pesticide and 5 percent less fertilizer, according to the World Economic Forum. It should be noted that improvements were primarily because of better advice available to farmers – there was no change made to farm mechanization or irrigation infrastructure. The Telangana government has expanded the program to 500,000 farmers who grow a wider range of crops including chickpeas, cotton, groundnuts, rice, and turmeric.

Similar agricultural programs in Africa, led by a company called Digital Green, are witnessing average yield boosts of 12%, with farmer incomes surging by up to 24%.

Potential Impact in Pakistan

Potential to increase agricultural output by up-to 10 billion dollars

We project that AI-assisted agricultural advisories can quickly unlock over $1 billion in additional economic output for Pakistan, even with low adoption rates. If adoption is widespread, $10 billion is plausible.

The graph on the right shows the potential increase in agricultural output as a factor of adoption (%farmers reached) and effectiveness (%yield increased).

It should again be noted that our average yield per acre is the lowest amongst countries with similar climate and infrastructure. According to data from US Department of Agriculture, foreign agriculture service, the average yield per acre of India, Bangladesh and Uzbekistan is 35% higher for rice, 12% higher for cotton, 29% higher for wheat, 24% higher for corn, 70% higher for peanuts, 87% higher for soybeans and 215% higher for millets.

Call for Action

AI research teams, agricultural experts and government departments should work together

UpliftAI is developing the foundational technology required to build such AI-assisted advisories.

However, for such systems to work well we need a close collaboration with agricultural experts who understand the science and practicalities of farming in different regions of Pakistan. Such experts will author and share the data, instructions and advice that the AI will learn to replicate. As such, we would like to enter into a formal collaboration with University of Agriculture Faisalabad to provide said expertise.

The potential rewards are high: a leap forward in Pakistan's journey towards sustained growth and development. By embracing this opportunity, we can not only address critical knowledge access challenges in Pakistan but also set a precedent for how AI can be leveraged for social good on a global scale.

About UpliftAI

An AI research and development group focusing on improving Pakistan’s economy

UpliftAI is an artificial intelligence (AI) research and development focused on uplifting Pakistan through technical innovations. This group is led by a team of experts in artificial intelligence, software development, and user experience design with successful track records at leading technology companies like Apple, Meta, Microsoft and Amazon.

Team Members:

Hammad Malik has eight years of experience in developing some of the world's leading voice assistants. His expertise includes four years with Alexa at Amazon and another four with Siri at Apple. At Apple, he was the AI/ML search lead for Finance, Stocks, Flights and #Images. Hammad earned a Master’s in Computer Science from Cornell University, USA and a Bachelor's in Computer Science from LUMS, Pakistan

Faizan Ahmad is an AI Research Engineer at Meta where he plays a pivotal role in the development and evaluation of cutting-edge Large Language Models, with a focus on security aspects. He co-authored the renowned Purple LLAMA. He holds a Master’s in Computer Science from the University of Virginia, USA,  concentrating on Deep Learning; and a Bachelor's in Computer Science from FAST, Pakistan.

Shehryar Ali is the CEO and founder of Conglo, a leading media production and marketing company in Pakistan. He has led some of the Pakistan’s most successful marketing campaigns. Notably, his work for the Pakistan Super League achieved over one billion+ views. Shehryar holds an MBA and a Bachelor's from LUMS, Pakistan.

Haseeb Niaz is a Senior AI/ML Engineer at Apple, focusing on AI Information Intelligence. He has over eight years of experience at leading tech companies including Apple, Microsoft Research, and Meta. Haseeb was the lead developer of CrystalNet, Azure’s high-fidelity cloud emulator, showcased at Microsoft Build. Haseeb has a Master’s in Computer Science from Stony Brook University, USA and a Bachelor's from LUMS, Pakistan.

Advisors:

Dr. Hafiz Asif is a Professor of AI at Hofstra University, New York, USA. His research focuses on analyzing AI systems and deploying them safely. He is the author of "Intelligent Pandemic Surveillance via Privacy-Preserving Crowdsensing," which won the 2022 Best Paper Award by the IEEE Computer Society, highlighting his forefront position in AI research for societal benefits. He holds a PhD from Rutgers University, USA and a Bachelor's from LUMS, Pakistan.

We are currently engaging with highly accomplished academics and executives in the field of AI, Agriculture and Education to join our advisory team.

Explore Prototypes

Please visit the Early Prototype section of the website to explore prototypes. More advanced versions will be demonstrated in person.