The calculator’s advent for mass adoption in the 1970s and 80s brought fears of eroding learners’ math abilities. Artificial Intelligence (AI) chatbots today are widely feared because they’re thought to hamper critical thinking.
The fear of calculators proved exaggerated. A meta-analysis of 79 studies found calculators improved problem-solving at nearly every grade level, although a Vanderbilt study showed they helped most once learners already had basic number sense.
AI has no such wall. Ask ChatGPT to solve the whole equation or write the whole essay, and it will. MIT Media Lab’s “Your Brain on ChatGPT” study found essay writers using ChatGPT showed the weakest brain connectivity and struggled most to recall their own work, a pattern researchers called “cognitive debt”. In the US, RAND found students using AI for homework rose from 48 to 62 percent in a single year.
For Uganda, the risk is double-sided. Students with AI risk that cognitive debt while those without access risk being left behind. A survey of 564 medical students at four public universities found 76 percent use AI tools, mostly ChatGPT. The same survey found Makerere students were 66 percent more likely to use AI than their peers at Gulu University. This was because Makerere sits in the capital, where connectivity and exposure to technology are greater. Another study of Ugandan university learners found students eager for AI but ranked access to technology the weakest factor.
The divide stretches beyond classrooms. About 72 percent of Ugandans – 36.5 million people – were still offline at the start of 2025, and only 28 percent are connected to the national grid.
A commercial farmer with AI advisory tools gets ahead, while a smallholder relying on guesswork does not even get the option. There are steps being taken to boost access and improve protection as more Ugandans learn about AI.
The Ministry of ICT has promised a national AI and emerging technologies strategy, targeting agriculture, health and public administration, with an ethics chapter anchored in the Data Protection and Privacy Act. Uganda validated its AI readiness assessment with UNESCO in June.
On infrastructure, a $1.2 billion Aeonian Project is building an AI data centre to be powered by Karuma power dam. The first phase is due this year, with full capacity by 2028. With an estimated 98 percent of African data processed abroad, local computing power is a matter of sovereignty.
There are more projects. Last October, the ministry launched Sunflower, a Sunbird AI model that works in Luganda, Acholi, Ateso and other local languages. The New York Times reported in August that a coffee farmer in Mityana now uses it to manage his farm. Home-grown tools such as IntelliFarmAI for pest identification and Makerere’s Mak Ocular for malaria and TB screening show Ugandans can build for Ugandan problems.
But a supercomputer at Karuma does little for a pupil in Kotido without power, a device or data. And none of these efforts addresses the first track: what AI does to learning for those already using it.
What should be done?
First, finalise and publish the AI strategy with a dedicated education chapter. The Ministry of Education and the National Curriculum Development Centre should issue clear guidance on when learners may use AI, insisting on foundations first, just as calculators came only after children mastered basic arithmetic.
Second, redesign assessment. Take-home essays that AI writes in seconds should give way to in-class writing, oral presentations and project work, which the competency-based curriculum already favours.
Third, treat access as infrastructure. Prioritise power and internet for rural schools and negotiate with telecom companies to zero-rate local tools like Sunflower so that cost is not the barrier. Fourth, train teachers first. Teachers who understand AI can set the guardrails the technology lacks.
