For those unfamiliar with Science Congress, it is a competition where learners in secondary schools in Kenya showcase their understanding of mathematics, science, and technology, and how they are able to translate the theories learned in class into tangible, practical solutions.
I recall how my friend Justine Muli and I somehow, miraculously, sailed through to the District level with a physics project, but later regretted it.
First, it was a first for our school. And, as you may know, our school was still young. No library. No dining hall. No computer lab. Since we all came from the same village, we spoke what I can only describe as English–Kikamba. We were confident, but with very little exposure.
Then we arrived.
We found ourselves in one of the provincial schools at the time, now a national school. As students from other schools made their presentations prior to ours, they were fluent in English and sounded like they were explaining university-level concepts.
When it was time for questions, students started shooting questions that seemed completely outside the high school syllabus, while there was virtually nothing from my side. I froze. I forgot almost everything in my presentation.
Then came the girls.
Girls from Precious Blood Kilungu and Makueni Girls spoke excellent English. They argued confidently. And they knew they were good. We did not even have enough English to greet them properly. We could only follow them around quietly, unable to even say hi, for fear of embarrassing ourselves. They seemed to enjoy the emotional harassment they were giving us.
Anyway, when it was time for our presentation, we did our best, even though my heart was almost coming out of my chest.
Our project was about using a simple electrical concept, a series connection of current, to help illiterate people identify food categories and prepare a balanced diet.
Here is how it worked.
We had different food categories on one side, for example proteins. An illiterate person would place a bulb on different categories, and the bulb would light up only when it touched the category containing pictures of protein-rich foods that were also locally available. Using this kit, someone could easily identify what foods to include and achieve a balanced diet without needing to read.
Looking back now, we were actually using the same core logic behind what we today call Large Language Models, the technology that powers modern AI systems.
A Large Language Model, or LLM, is not magic. It does not think like a human. It works by recognising patterns. You give it an input, what we now call a prompt, and based on patterns it has learned from large amounts of data, it predicts the most appropriate response.
In our project, the bulb was the response.
The food category was the input.
The electrical circuit was the logic layer.
Our system was rule-based and physical. Today’s AI systems are data-driven and digital. But the underlying idea is the same. Match the right input to the right knowledge and give a useful response.
Had our Science Congress project been given proper support, mentorship, and maybe a computer lab or two, I might not be writing this story today. We could be millionaires explaining AI on global stages.
Nevertheless, after our presentation, there were no tough questions like there were for the other students. I knew one of three things had happened. Either we did not communicate well, or our project was beyond the scope of the audience, or most likely, we simply did not make much sense.
Later, our teacher told us that we had actually placed position three. However, because our school was not well known, our position was quietly given to another, more established school. It is nice to have teachers who can find the correct words for negative feedback and massage the students’ ego.
But had we progressed, how exactly would we have faced girls from Kenya High and Alliance Girls?