Leveraging AI techniques and ideas in Malawi

🚀 The mission of the KAI Lab is to work on interesting and relevant projects and research in AI. And by doing that, it desires to facilitate debates and research, to create a channel for exchanging ideas, fostering innovation and bringing together those engaged in exploring or actively using AI in Malawi (and beyond).

🔗 Research at the KAI Lab covers several areas of interests: health informatics, data intelligence such as datasets creation and cataloguing, NLP, ML, HCI. If after reading through our projects on the Projects page you found some interesting, please let us know.

We need to explain especially to Chichewa speakers how the “K” in the name of the lab came about. “K” stands for “Kuyesa”, a verb that means “to test”, or “to experiment”. “Kuyesera” means experiment. So one can use “kuyesa” for something that may fail during experimentation and may never be used. Or one may mean something that is being tested with the expectation of success and of resulting in a successful outcome. Both meaning are understood. However, when one brings together a Chichewa word with a technological term in English, the meaning of the resulting phrase can sometimes be lost or misunderstood. So to avoid misunderstanding we refer to our lab as KAI lab.

Here is a comparison between testing versus experimenting taken from wikidiff: “A test is a trial a withness, while an experiment is a test under controlled conditions made to either demonstrate a known truth, examine the validity of a hypothesis, or determine the eficacy of something previously untried”.

💡 We aim to do both testing and experimenting in the KAI Lab. We want to know how AI techniques, tools, algorithms can be used to solve real problems in Malawi. It is not always easy for non-experts to differentiate between what constitutes an AI-based solution from a solution found with other means or methods. Similarly, it is not trivial to understand why, how and to what extent an AI solution is better than one say based on general classes of algorithms / methods.

💰 Funding Acknowledgment: The lab is sustaining itself through research funding and project work. We would like to acknowledge our grantors and partners for their support. We mention them on the project and events pages. We also seek new collaborations.

So now a few words about AI.

Artificial Intelligence (AI) is an area that is both old and new. It is old because it is one of the oldest area in computer science. The term AI was coined by one of the “founding fathers” of AI, John McCarthy, in 1956. AI aims to build intelligent machines that can do the same kind of intelligent tasks humans can do: problem solving, understanding speech etc.

It is new because, while significant progress has been made, the initial questions raised by AI are as relevant and open as ever. In the seminal book on AI, “Artificial Intelligence: A Modern Approach”, Stuart Russell and Peter Norvig state that the building of AI requires one to address some deep questions:

  1. Can formal rules be used to draw valid conclusions?
  2. How does mind arise from a physical brain?
  3. Where does knowledge come from?
  4. How does knowledge lead to action?

McCarthy wrote in 1955: “The study (of AI) is to proceed on the basis of the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.” Hence, in the early days as today, the emphasis was on developing programming languages / algorithms and formal systems for knowledge representation and reasoning that can equip machines with AI.

The Turing test for a intelligent machine required that AI possesses four capabilities: natural language processing (NLP); knowledge representation; automated reasoning; and machine learning (ML). The Total Turing test required two additional qualities: computer vision and robotics.

All these areas have over the years witnessed significant breakthrough. Natural Language Processing and expert systems were extremely vibrant areas of research in the early days and have again, in recent time, come at the forefront of AI research. ML and Deep Learning (DL), once modest areas of AI, have established themselves as leading techniques for building AI.

The AI community is once again anticipating the coming of Artificial General Intelligence (AGI). AGI (as opposed to Narrow AI that is good at one type of intelligent task) refers to an intelligent machine that performs multiple intelligent actions similar to what humans can do. And to achieve AGI, one needs to equip an AI (agent or program) with knowledge and reasoning. This is where Big Data and the internet have played a major role. The shift from logical reasoning to a DL-or ML-based reasoning seems to characterise the recent trends in AI research and discussions.

AI is old and new. It is also universal. While a few years ago, research and development in AI happened at a few top universities and companies, today, all organisations are expected to adopt some AI technology. Continents such as Africa have started to play an important role in the AI research and discussions. A lot of that has to do with “big data”.

AI aims to solve complex and hard problems. This requires sustained engagement in understanding and describing problems. And this entails developing research themes, constructing datasets and engaging AI in practical applications. It also means that AI is not something that can be easy to scale. A lot of experimentation and testing is needed. Hence, establishing an AI lab just like ours is essential to progress.

By experimenting and testing AI, one gets closer to understanding the crucial aspects and impacts that this area has on our life and society today and in the future. And we at KAI Lab look forward to hear your thoughts on AI.

The digital transformation of Malawi is here. Be part of the change! 🌐🔍✨

Posts

subscribe via RSS