One of the advantages of artificial intelligence (AI) is that it can help us carry out tasks faster and with fewer errors than humans. I wanted to test this on analyzing labor market demand and skills gaps. So, earlier this year, I partnered with Headai, a Finnish company, to apply an AI-enabled labor market assessment tool in Kenya. We used the tool to analyze: (a) online job advertisements from select online job portals in Kenya and (b) computer science curricula from the University of Nairobi and Moi University to identify the gaps between what the labor market is looking for and what the university curriculum is providing.
There are at least 20 online job portals in Kenya including some mobile apps. For our pilot, we selected three local online job portals – Jobweb Kenya, Kenyan career, and Jobs Kenya One – since they had a large amount of machine-readable online job advertisement data with sufficient web traffic.
In total, the analysis was completed in two months, which is much faster than what humans can do – and with greater accuracy. After several weeks of setting up a web-crawling function, Headai’s tool in a few days read and stored more than 60,000 online job advertisements from 2015 to June 2019. The AI tool, which had been trained to recognize skills-related words, then in a couple of days analyzed stored data and produced a skills map and a skills-demand trend. The AI tool also read and analyzed curricula from the universities’ websites and identified the gap against the skills demand using the same methodology. For this part of the study, I selected a computer science program, on account of its standardized occupational categories as well as the growing interest in digital skills across the continent. We analyzed BSc and MSc degrees at the School of Computing and Informatics at University of Nairobi, along with BSc and MSc degrees at the Department of Information and Communications and Technology at Moi University. Although the tool could read only the curriculum available online, more than 100 relevant courses were available for the analysis, which was a good representation for the gap analysis. The tool analyzed the top 20 most demanded industries, job titles, and skills from the labor market information set, and analyzed a skills-vs-curriculum gap and presented this as a visual map. After the tool produced these analyses, the HeadAi team analyzed the results to figure out the meaning of the co-relations. However, the limitation of this approach is the detail and coverage of analysis. As the AI tool analyzes texts at the “word” level, human efforts are required to validate and analyze the relationship between the words. For instance, the job title “provider” can be “solution provider”, “service provider”, and “health care provider.”
Among the top 20 skills in demand, accounting ranked first, while computer science and engineering area saw an increase in skills demand for data, networks, developers and maintenance between 2015 and 2019. Within the software and ICT industry, 30 percent of the skills demanded are business and soft skills, such as experience, service, and communication, while a further 47 percent of the skills demanded are technical skills such as data, excel, and computers. The analysis suggests that, whereas both universities teach business and soft skills (Among available courses at select departments, 15 percent of courses offered in University of Nairobi and 25 percent of courses offered in Moi University focus on business and soft skills), they need to offer more courses to meet employers’ demand.
The AI tool also mapped the gap between the skills demanded and the curriculum offered, using the technology called “digital twin.” This map shows the skills demanded and taught at university in yellow, skills demanded but not taught at university in pink, and skills not demanded but taught at university in green.
To analyze these gaps further, we have divided skills demand into four categories.
Category 1: Professional Software Developer
A gap between skills demanded and curriculum offered exists even in the latest digital skills, such as big data analytics, machine learning, use of cloud computing and a computer programming language called python.
Category 2: Industry Platforms and Tools
The gap is found in industry standard such as ISO 9001, enterprise software including SAP, IBM, Oracle as well as a mobile phone platform like Android.
Category 3: Financing and Business-Related ICT skills
Universities lack providing business relevant ICT skills, such as data analytics as well as subject matter and business knowledge including financing, accounting, banking, auditing, logistics, and contracts.
Category 4: Basic ICT skills
The gap exists in Microsoft office, editing as well as creative digital skills.
Overall, the research found that the curricula offered were not too far from the skills demanded by industry. Some of the gaps may not be gaps at all, but rather a reflection of differing nomenclature. For instance, whereas academia tends to speak of “artificial intelligence”, the private sector uses “data science” and “big data.”
Based on the results, Moi University has proposed modifications to their curriculum development practices to include online learning, further teacher training, multi-disciplinary teaching and industry partnership.
As the skills demand changes rapidly and new jobs continue to emerge, many training institutions and policy makers would benefit from being able to analyze the demand and supply gap quickly when planning their curriculum and workforce development policy.
Moreover, as more employers post their jobs online, industry coverage will increase. Already in Kenya, some mobile applications such as Lynk, connect informal workers and individuals to jobs. If we get access to their data, informal sector labor market analysis can open a wealth of new possibilities.
In short, as the number of online job portals across Africa is growing, leveraging online job advertisement data especially in the area of digital skills will be increasingly useful. AI tools can make the analysis much faster, easier, and cheaper for policy makers, universities and training providers. The analysis will help them identify their “gap” and take appropriate measures to narrow it through using online learning, linking with industries, and revising the curriculum to bring them closer to the needs of the everchanging labor market.
Download Final report: Labor market analysis and curriculum gap assessment using big data in Kenya
The author would like to thank Moi University and University of Nairobi for their participation and Ruth Charo, Senior Education Specialist, World Bank for the implementation support.