Why We Need Better Evidence to Inform the Development and Use of AI in Career Guidance for Young People

AI is beginning to shape young people’s transitions from education to work. 

Increasingly, career development practitioners are using large language models to support guidance sessions and for resource creation and administration. Careers-focused chatbots, powered by AI foundation models and drawing on labour market data, are being developed to provide access to ‘personalised’ support through hybrid – AI and human guidance – models that then signpost the young person to a professional adviser. 

Formal career guidance aside, young people are also turning directly to easily accessible generic AI tools such as ChatGPT for career exploration, job applications and interview preparation. Meanwhile, AI is transforming the labour market and creating uncertainty for young people’s career planning when the jobs of today that they aspire to may not exist or may look different in the future. 

The Nuffield Foundation, with its deep expertise in education and youth transitions, and the Ada Lovelace Institute, with its focus on the relationship between AI and data-driven technologies and the people who are affected by them, collaborated to explore this emerging field. In ‘Navigating the future: a landscape review of AI in career guidance for young people’, published today, Renate Samson and I describe the ways in which AI is currently being used to support young people’s transitions, assess the opportunities and risks, and identify future ways that AI developers, policymakers, education leaders and career development practitioners can engage with AI to support positive and equitable outcomes for young people. 

Through a literature review and exploratory conversations with career development practitioners, educators, employers, technologists and policy officials from across the UK, we identified areas where AI could potentially improve access to careers information, enable a hybrid – human and AI – approach to advice and guidance, with the potential for wider access to employment for marginalised young people and an increase in efficiency for practitioners. 

Given the challenges facing young people today, reflected in the high rates of unemployment and economic activity, these opportunities could have a positive impact. However, AI adoption also carries risks.

Evidence that young people are using generic chatbots for career guidance is particularly concerning. The human-like qualities of chatbots can give young people the impression that the model has empathy and understanding and is equivalent to human guidance. In reality, the outputs are unreliable and are not based on the ethical and professional standards that guidance requires. 

More broadly, issues with accuracy and bias affect how decisions are made. The challenge of hallucination, a feature of AI, can lead to people relying on inaccurate, misleading or simply made-up information. Further challenges that are coming to light are a growing over-reliance on the technology, and risks of cognitive offloading leading to potential critical skills erosion. There are also risks of growing inequalities in access to information via digital tools, and as AI is adopted more broadly, to human support. Many young people are concerned about the environmental impact of AI. 

To mitigate these risks and promote an appropriate, ethical and effective approach to the use of AI in career guidance, the report emphasises the need for: 

• A purpose-led (rather than tech-led) approach guided by a deeper understanding of young people’s career support needs, use of AI as a support not a substitute for professional careers advisers, and collaboration between AI developers and careers experts. 
• Comprehensive upskilling of the careers workforce on ethical use of AI, tech transformation of the labour market, and on how to support young people to develop the skills needed to use AI effectively in job applications and in the workplace. Employer engagement and work experience are important for young people and advisers to stay abreast of the latest trends. 
• Regulation and evaluation, including independent, impartial evidence about the effectiveness of AI tools for education and employment outcomes. At present, there is very little evidence that goes beyond implementation feasibility and acceptability. Building such evidence will be challenging given the pace of technological change and new tools appearing on the market, the time needed to evaluate medium- to longer-term outcomes, and the multifaceted nature of career guidance. 

To support progress in this area, the Nuffield Foundation has awarded funding to the Institute for Employment Research at the University of Warwick to develop and trial an assessment protocol to be used for evaluating AI careers advice in contrast to human advisers. The study will take a collaborative approach to build sector consensus about appropriate evaluation approaches and thresholds for evidence to inform practical guidelines and specific policies on tool usage. 

As with many other sectors, the career guidance sector is keen to identify where adoption of AI can benefit young people and practitioners alike. Our report shows that a great deal of considered research, exploration and trialling has been taking place. There is more to do to ensure AI adoption is appropriate. We hope our work supports a positive direction of travel for the sector that brings lasting benefits for all. 

By Emily Tanner, Programme Head of Post-14 Education & Skills at the Nuffield Foundation