A University of Cape Town (UCT) researcher has secured international funding to examine whether generative artificial intelligence (AI) can improve employment outcomes for South Africa’s unemployed youth.
Dr Tim Köhler, senior research officer at the Development Policy Research Unit in UCT’s School of Economics, is one of 19 early-career researchers selected from more than 300 applicants to receive support from Schmidt Sciences’ AI at Work programme.
In January, the US-based non-profit announced over $3 million (about R48 million) in funding for the global cohort.
Köhler’s project is the only South African-led study in the group and will generate what is expected to be the first rigorous developing-country evidence on how a generative AI-powered career guidance tool affects job-search outcomes for unemployed youth.
The AI at Work programme supports research into how AI adoption affects productivity, wages, employment and career pathways. The 19 awardees represent institutions across eight countries, including MIT, Harvard, Oxford and UCLA. Key partners include researchers from MIT’s Shaping the Future of Work Initiative and the Abdul Latif Jameel Poverty Action Lab.
Jobs scarcity
Youth unemployment remains one of SA’s most pressing socio-economic challenges. Statistics SA data shows around 60% of the youth labour force aged 15 to 24 and 40% of those aged 25 to 34 are unemployed, affecting roughly five million young people.
“A variety of factors explain these high rates, but for young jobseekers, severe information constraints can play a particularly important role,” says Köhler.
“They may not know which career paths are viable given their circumstances, what skills employers are actually looking for, or how to navigate both formal and informal labour markets.”
These constraints, he argues, make SA a relevant test case for examining how AI tools function in labour markets characterised by structural unemployment and limited opportunities.
“If AI can help young people navigate these complexities here, the lessons are likely to be relevant for many other countries across the continent and the Global South,” Köhler adds.
According to Schmidt Sciences, the objective of the programme is to generate credible, real-world evidence as AI adoption accelerates, outpacing policy and regulatory responses.
“This award places UCT and South Africa more broadly at the forefront of an emerging global conversation about AI and work,” explains Köhler. “One that has so far been dominated by evidence from high-income countries.”
Personalised guidance
Köhler’s study will evaluate an AI-based career guidance tool for unemployed youth. It is being co-designed with a Cape Town-based technology firm and informed by engagement with NGOs, researchers and government stakeholders involved in youth employment initiatives.
Powered by large language models and optimised for low-bandwidth environments, the tool will operate as a personalised, conversational assistant accessible via phones or computers.
Köhler explains: “Through a simple conversational interface, users can receive guidance tailored to their own circumstances. That might include identifying realistic job opportunities, improving a CV, preparing for interviews, or considering training or self-employment pathways when formal jobs are limited.”
The focus of the tool is on delivering context-specific advice rather than generic information.
“Many young people receive generic advice via the internet that doesn’t reflect local labour market conditions or their personal circumstances,” Köhler continues. “We want to see whether more relevant, personalised, dynamic guidance can help improve decision-making and, ultimately, well-being.”
The project will use a randomised controlled trial to measure impact. By randomly assigning access to the AI tool, researchers will assess its effect on employment outcomes, job search behaviour, confidence and expectations.
The study will also examine possible unintended consequences, including whether such tools could reinforce existing inequalities, or raise expectations without improving access to opportunities.
“Our expectation is not that AI is a silver bullet for solving unemployment,” he states. “But if well-designed, it may help ease specific constraints young people face as they try to navigate the labour market.”
If effective, the tool could serve as a scalable, low-cost complement to existing employment and training programmes.
Share