Artificial Intelligence in Ghanaian Basic Schools: Opportunities, Challenges and the Way Forward

An AI tool can now draft a lesson plan, a set of quiz questions or a simpler version of a reading passage in less time than it takes to write the date on the board. For a teacher with a large class, a pile of exercise books and a lesson note due on Monday, that is an attractive promise. But the important question for Ghana's basic schools is not whether artificial intelligence is impressive. It is whether it can help, reliably, fairly and safely, in classrooms where many learners have no device, the power supply is uncertain and few teachers have been trained to use these tools. This article separates what is realistic today from what is still a possibility for the future.

Where things stand today

National policy is moving

In April 2026, President John Dramani Mahama launched Ghana's National Artificial Intelligence Strategy (2025–2035), a ten-year roadmap that names education among the sectors where AI is expected to contribute. The Minister of Communication, Digital Technology and Innovations, Samuel Nartey George, said the strategy is designed to ensure that AI "contributes to equitable access to education".

Within the education sector, the most concrete step so far has been at senior high level. In October 2025 the Ministry of Education announced subject-specific AI apps for Senior High Schools, co-created by the Ghana Education Service (GES), the National Council for Curriculum and Assessment (NaCCA) and other agencies, to help teachers "plan lessons and conduct assessments more efficiently". That initiative serves the SHS curriculum, not basic schools. At the time of writing, I have not found published GES guidance specifically on the use of AI in kindergarten, primary or junior high school classrooms.

Teachers are interested but under-prepared

Research on how ready Ghanaian basic school teachers are for AI is still limited. One recent study surveyed 319 basic school teachers in the Cape Coast Metropolis and found moderate levels of self-reported AI literacy across four areas: understanding AI, applying it, evaluating AI applications and AI ethics. It also found statistically significant differences between male and female teachers, with male teachers reporting higher levels, particularly in applying AI (Arkorful et al., 2025).

The study covers one metropolis and relies on teachers rating themselves, so it is not a national picture. It does support what common sense suggests: most teachers have not yet had structured training in these tools.

Infrastructure is the real constraint

NaCCA's own Computing curriculum states the problem directly. It notes that Ghanaian schools lack ICT infrastructure "due to the high cost of computers and other technologies, poor internet connectivity, intermittent power fluctuations, and teachers' inadequate knowledge or technical know-how in using ICT tools in teaching" (NaCCA, 2023).

National data points the same way. In the 2021 Population and Housing Census, 31.4% of people aged six and older had not used the internet in the previous three months, and 19.6% had not used any ICT device at all (Ghana Statistical Service, 2023). Those figures will have moved since 2021, and they describe the whole population rather than schools, but they are a reminder that any AI plan which assumes every learner is online will leave many children behind.

Opportunities that are realistic now

Given those conditions, the most realistic use of AI in most basic schools today is teacher-facing rather than learner-facing. The teacher uses a phone or laptop to prepare; learners benefit from better materials without touching the tool themselves. This approach works around the device gap, keeps an adult in control of every output, and avoids the question of young children using general-purpose AI services. UNESCO's guidance on generative AI in education and research, published in 2023, sets an age limit of 13 for the use of AI tools in the classroom and calls for teacher training on the subject.

Lesson planning and preparation

AI tools can suggest starter activities, worked examples, discussion questions and local illustrations for a topic. A teacher planning a lesson on the water cycle could ask for three everyday Ghanaian examples of evaporation and condensation, then choose the one that suits the class and discard the rest.

There is some evidence that this saves time. In a 2024 trial in England commissioned by the Education Endowment Foundation, 259 secondary science teachers from 68 schools were assigned either to use ChatGPT with a guide or to avoid generative AI. Teachers using ChatGPT spent about 56 minutes a week on lesson preparation, against about 82 minutes for the comparison group, and an expert panel that did not know which resources were AI-assisted found no evidence of a difference in quality (Roy et al., 2024).

That is a useful result, but note where it comes from: secondary science classrooms in England, over ten weeks. It is not evidence about Ghanaian basic schools. It is a reason to test the idea here, not a promise.

Drafting assessment items

AI can draft multiple-choice questions, short-answer items and marking schemes quickly. The draft is the easy part. Every item still needs checking: Is the answer key correct? Is the language right for the class? Does it test the indicator in the scheme of learning? Are the wrong options believable, or obviously silly?

Disclosure: MosTek Solutions builds ExamCraft, an examination authoring tool that works in this area. Its development is described in a separate article.

Differentiation

A mixed-ability class needs the same idea presented in more than one way. AI tools can rewrite a passage at a simpler reading level, produce extra practice for learners who need it and suggest extension tasks for those who finish early. The teacher still decides who needs what, because only the teacher knows the learners.

Do not assume that output in Ghanaian languages such as Ewe, Twi, Ga or Dagbani is accurate. Test it with fluent speakers before using it with a class.

Everyday professional writing

Teachers write a great deal besides lesson notes: letters to parents, notices, meeting summaries and project outlines. AI can produce a first draft that the teacher then corrects and personalises. Keep learners' personal information out of anything you type into the tool.

Challenges that cannot be skipped

Cost, connectivity and power

Most AI tools need an internet connection, and data costs money. A teacher who pays for data out of their own pocket to prepare lessons is quietly subsidising the school. Schools considering AI should budget honestly for data and devices, and favour tools that work on a phone, on a weak connection or offline.

Training

A tool handed to teachers without training will be used badly or not at all. UNESCO's AI competency framework for teachers sets out 15 competencies across five dimensions: a human-centred mindset, the ethics of AI, AI foundations and applications, AI pedagogy, and AI for professional learning (Miao and Cukurova, 2024). Ghana does not need to copy it wholesale, but it is a useful checklist for anyone designing training. Teachers need to understand what these tools are, where they fail and how to use them for learning, not only how to type a request.

Accuracy

Generative AI produces fluent text, not verified facts. It can give a wrong answer with complete confidence, invent a reference or misstate a curriculum indicator. In a classroom, an error on a teacher-prepared worksheet carries the teacher's authority. Every AI-generated material should be checked by someone who knows the subject before it reaches learners. A later article in this series looks more closely at keeping professional judgement at the centre.

Fairness and equity

If AI makes preparation faster, the teachers and schools that already have devices, data and training will benefit first. Without deliberate effort, that widens existing gaps between well-resourced and under-resourced schools and, as the Cape Coast study suggests, possibly between male and female teachers. AI tools trained largely on text from other countries may also suggest examples, names and assumptions that do not fit Ghanaian learners. Part of the teacher's job is to notice and correct this.

Privacy

Learners' names, results, health information and family circumstances should never be typed into a public AI tool. In Ghana, the Data Protection Act, 2012 (Act 843) governs the processing of personal data, and the Data Protection Commission is the regulator established under it. In March 2026 the government announced that it was developing a new Data Protection Bill to regulate AI systems, automated decision-making and cross-border data transfers. Schools should seek proper advice on their own obligations. A separate article will look at protecting school data in more detail.

Today's realities and tomorrow's possibilities

It helps to keep two lists apart. What is realistic now, with care, is the teacher-facing work described above: drafting lesson ideas, questions, adapted reading material and routine writing, with every output checked by the teacher, and small school-based trials whose results are shared with colleagues.

What is possible later, with investment and local evidence:

• Learners using AI tutors on shared or personal devices.

• Reliable support in Ghanaian languages.

• Practice exercises that adapt to each learner's progress.

• Analytics that help schools spot learners who are falling behind.

The second list is not fantasy. But each item depends on devices, connectivity, training and data protection and, above all, on evidence from Ghanaian classrooms that it improves learning. None should be adopted on the strength of a demonstration alone.

A responsible way forward

For teachers: Start with one task you do every week, such as drafting quiz questions. Check every output against your subject knowledge and your scheme of learning. Keep learner information out of the tool. Note what saved time and what did not.

For school leaders: Agree a short school position on AI: which uses are encouraged, what must always be checked, and what information must never be entered. Use staff professional learning sessions to share requests that worked well and mistakes that were caught.

For policymakers: Basic schools need guidance of their own, not just an extension of SHS plans. Teacher training, low-data and offline tools, support for Ghanaian languages and properly evaluated pilots would do more good than a rush to put AI in front of young learners.

Conclusion

AI will not fix the shortage of devices, the cost of data or the size of classes, and it should not be sold as if it could. What it can do today is give a well-prepared teacher a faster first draft. The value of that draft depends entirely on the professional judgement applied to it. If Ghana's basic schools adopt AI in that spirit, putting teachers first, demanding evidence and protecting learners' privacy, it can become a genuinely useful tool rather than another unfulfilled promise.

Try this: Ask an AI tool to draft five questions on a topic you are teaching this week. Before using any of them, check the answer key, the wording, the level of difficulty and the link to your indicator. Note how long the checking took. That checking time is the real cost of the tool, and the clearest measure of why your expertise still matters.

References

• Arkorful, V., Arthur, F., Boateng, E., Ofosu-Koranteng, M., Salifu, I., Attom, E. R., Tetteh, S. A., Quayson, E., Asare-Bediako, S. and Nortey, S. A. (2025). Exploring artificial intelligence literacy among basic school teachers in Ghana. Discover Education, 4, 250. https://doi.org/10.1007/s44217-025-00630-3

• Citi Newsroom (1 October 2025). Education Ministry introduces AI-powered subject apps for SHSs. citinewsroom.com

• Data Protection Act, 2012 (Act 843). Republic of Ghana. Official PDF via the Data Protection Commission

• Data Protection Commission, Ghana. dpc.gov.gh

• Ghana News Agency (2 March 2026). Govt to introduce new Data Protection Bill to regulate AI, cross-border data flows. gna.org.gh

• Ghana Statistical Service (2023). Release of the 2021 PHC Thematic Brief on Digital Exclusion in Ghana. census2021.statsghana.gov.gh

• Miao, F. and Cukurova, M. (2024). AI competency framework for teachers. UNESCO. https://doi.org/10.54675/ZJTE2084

• National Council for Curriculum and Assessment (2023). Computing Curriculum for Secondary Education (SHS 1–3), p. 19. Ministry of Education, Ghana. PDF

• Roy, P., Poet, H., Staunton, R., Aston, K. and Thomas, D. (2024). ChatGPT in lesson preparation: A Teacher Choices trial. National Foundation for Educational Research, for the Education Endowment Foundation. nfer.ac.uk

• TechAfrica News (27 April 2026). Ghana launches first National AI Strategy to drive inclusive growth and responsible innovation. techafricanews.com

• UNESCO (2023). Guidance for generative AI in education and research. unesco.org

• UNESCO (7 September 2023). UNESCO: Governments must quickly regulate Generative AI in schools. unesco.org