AI skills should begin before university, but the starting point is not an advanced application or a complicated model. Learners first need to understand what AI can do, question its output and use it responsibly. Technical building becomes more useful when those habits already exist.
An age-based pathway can move from literacy to prompting, fact checking, research, responsible use, automation, APIs, agents and applications. It is a progression of judgement as well as technical skill. Students in Kenya can follow it through classroom activities, independent practice where permitted and practical projects matched to their readiness and available resources.
Begin with AI literacy
Recognise that AI tools have purposes and limits. They can generate useful content while making mistakes, and they do not take responsibility for how the content is used. A learner should be able to name a helpful use, a risk and a situation that needs a person’s decision.
For primary-school learners, start with simple questions, safe sharing and adult-led examples. AI Smart & Safe develops those foundations. The ability to protect information and seek help matters before independent access to a public tool.
Learn to prompt with a clear purpose
A useful prompt describes the task, relevant context and constraints without adding unnecessary personal information. Ask for the kind of support you need: an explanation, practice questions, possible approaches or feedback permitted by the teacher.
Compare results rather than collecting “perfect prompts”. If an instruction gives an unsuitable answer, identify what is missing and revise it. Good prompting includes knowing when the task should not be delegated. A clear request cannot make an inappropriate use acceptable or turn an uncertain answer into evidence.
Make fact checking routine
Students should practise identifying a claim and choosing a dependable source to verify it. For a calculation, work through the method. For a quotation, locate the original text. For an image or news claim, investigate source and context rather than trusting appearance.
Keep a short record of what was checked and what remains uncertain. This habit is useful in schoolwork and later technical projects. A developer who cannot recognise plausible but incorrect output may build a system that repeats the same mistake for many users.
Develop research skills alongside tool skills
AI can suggest questions and search terms, but students need to read original material and connect evidence to an argument. They should distinguish a research lead from a source they have actually examined. An invented reference must not enter a bibliography.
Practise asking who produced the source, what it supports and what its limits are. Compare perspectives where relevant and explain why particular evidence was selected. Subject knowledge remains important: useful research requires enough understanding to notice when an answer is incomplete or the question itself needs improvement.
Keep responsibility and integrity visible
Know the school’s rules and the assignment objective. Explain permitted assistance and retain your own reasoning. Protect information about yourself and others. Do not use generated content to mislead, impersonate or harm somebody.
UNESCO’s student competency framework includes human-centred and ethical dimensions alongside technical learning. That is a useful reminder that responsible behaviour belongs throughout the pathway, rather than being postponed until a learner begins coding.
Understand automation as a sequence of decisions
Begin with a repeatable task and draw its steps. What comes in, what decision is made, what comes out and where should a person review it? A simple diagram can reveal whether AI is needed at all or whether ordinary software rules would be clearer.
For example, a fictional enquiry workflow might sort sample requests, draft a response and wait for human approval. Consider errors: a request in the wrong category, missing information or an inappropriate draft. Good automation includes a safe stopping point and a way to recover, not just the happy path.
Explore APIs when the foundations are ready
An API lets software request a service and receive a response. Older learners can study that exchange using an appropriate learning environment. They need to understand inputs, outputs, failures and the difference between a demo and a dependable application.
Protect credentials, use synthetic or authorised data and consider cost and access limits before connecting a service. Do not paste keys into public code or shared prompts. Practical learning can begin with a small request and a clear explanation of what happens when the service is unavailable or returns something unexpected.
Learn what agents and tool calling change
An AI agent may be designed to carry out a bounded task using tools. That expands the consequences beyond producing text. A tool might read information, update a record or trigger an external action, so permissions and approval points matter.
Start with low-risk sample tasks. Give the system only the access it needs and decide when it should ask a person. Test misleading input, unclear requests and failed tools. A learner should explain why an action is allowed and how the system stops, rather than celebrating that it can do many things automatically.
Build applications around real needs
Choose a small problem with a clear user and a way to evaluate success. A question-answering helper for public student-club information or an organiser for fictional requests can be more useful than a large demo with no defined purpose.
Combine ordinary software skills with AI: readable code, version control, testing, error handling and user communication. Review generated code rather than assuming it works. Show cases where the system should decline, ask for clarification or hand control to a person. Responsible boundaries are part of the product design.
Prepare for deployment without rushing it
A working classroom example is not automatically ready for real users. Consider information handling, permissions, costs, monitoring and what happens when the model or service changes. Test representative cases and document limitations in language users can understand.
Kenya’s National AI Strategy 2025–2030 provides national context for AI development. It does not guarantee an individual learner a job or endorse a particular course. Build evidence of skill through useful, reviewable projects rather than promises about career outcomes.
Choose the next stage thoughtfully
Learners can revisit earlier skills as projects grow more complex. Better research improves a prototype; better privacy judgement improves a workflow; clearer communication improves both prompts and user instructions. Progress is not a race to the most advanced tool.
Junior and senior students can strengthen the foundations through AI Ready. Universities and colleges can enquire about AI Builder, where practical learning extends towards applications and workflows. Cohort arrangements and prerequisites are confirmed directly with BritaCore.
Common questions
Must every learner become a developer?
No. Literacy, checking, research and responsible use are valuable across subjects. Technical building is a later option for learners whose goals and readiness fit it.
What should a project portfolio show?
Show the problem, choices, evidence, tests and limits. Include what you changed after discovering a failure and where human judgement remains necessary.