Work and skills: learning AI and building with it

Two kinds of question bring people to AI as work. One is "how do I learn this?", from students, graduates and people changing careers. The other is "can it pay me now?", often from the same people. Both deserve straight answers, because both attract offers that cost money and deliver little.

The best introductory AI courses are free, the tools for practising run in a web browser, and some of the strongest AI communities in the world are African. Our guide to learning AI for free sets out a twelve-week plan: two free courses on what AI is and how AI projects work, the basics of Python and data, and a small project on data from where you live.

Machine learning, the part of AI where systems learn patterns from data instead of following rules written by hand, is where the building happens. A model is what that learning produces: the trained system that turns an input, such as a question or a photo, into an output, such as an answer or a label. Training an existing model further on your own data so it does one job, or speaks one language, better is called fine-tuning, and it is one of the ways small African teams build tools the large ones lack.

Meanwhile there is paid work. Companies building AI pay people to label pictures, transcribe speech, rate answers and write examples, work often called data labelling or data annotation, and much of it is done from Nairobi, Lagos, Kampala and Accra. It is real and it can pay, but it is gig work that a platform can end overnight, and the region has seen low pay, a platform switched off with little warning, and workers taking a global company to court. Our guide to AI training jobs in Kenya and Nigeria sets out what happened, what to ask for in writing, and how to spot the fake offers that borrow the real work's name.

The two questions meet in the same place. The tasks that pay best go to people with scarce skills, and the skills that last are the ones a learning plan builds: fluent writing in a language companies need, expertise in a subject, and the ability to work with real data. Use the paid tasks as a step, and keep climbing.

Whether you apply for a job, a programme or a client, the people deciding want to know one thing: can you do the work? A certificate says you finished a course. A small project says far more: a question you asked, the data you found, what you did with it, and what went wrong. Put it somewhere others can see, write a page about it, and mention it first.

Communities help here too. A challenge entered on a platform such as Zindi, a contribution to an open language project, or a talk at a local meet-up leaves a public record that anyone can check. Governments in the region are investing in the same skills, through programmes such as Kenya's Ajira Digital and Nigeria's 3MTT, which our guides describe.

AI training jobs in Kenya and Nigeria: the work, pay and risks
What the work really is, what has happened to workers here, and how to spot a scam.

How to learn AI for free in Africa: a 12-week plan
Free courses, African communities, and a plan that ends with your own project.