As Artificial Intelligence reshapes the workplace, universities face growing pressure to equip graduates with the practical, critical and ethical capabilities needed to thrive in an AI-driven economy. Dr Mahmoud Mousa, Associate Professor at the School of Mathematical and Computer Sciences, Heriot-Watt University Dubai, tells us why higher education must look beyond prompt engineering and focus on critical thinking, data literacy, ethical reasoning and domain expertise to develop genuinely AI-ready graduates.
According to a survey of 1,000 global employers, collectively employing more than 14.1 million workers, 86% of employers expect AI to drive transformation within their organisations over the next five years, while 40% of core skills are projected to change by 2030.
The message to higher education is unmistakable: adapt, or become irrelevant. Universities around the world have heard the call. The response, however, has often been reactive rather than strategic. Institutions are racing to add prompt engineering workshops and tool tutorials, treating Generative AI proficiency as the ultimate marker of graduate employability.
In doing so, they risk prioritising a transient skill set over the deeper capabilities that will truly define the AI-ready graduate.
Prompt engineering is a transient skill. The interfaces through which we interact with AI are evolving rapidly, and the specific techniques that work today may be obsolete tomorrow. What employers are desperately seeking is a deeper set of capabilities such as critical thinking, data literacy, ethical reasoning and domain expertise. These are the foundations of what it means to be truly AI-ready.
There is a distinction between AI literacy and AI readiness. AI literacy means knowing what AI, Machine Learning and large language models are. This is achievable by understanding core concepts and being able to discuss them. AI readiness means being able to design, build, deploy and maintain AI-enabled systems in real environments. This would require working with cloud infrastructure, data pipelines and production tooling; and collaborate effectively with product and platform teams. As a result, an AI-ready graduate can be onboarded into a team and contribute in months, not years.
Most universities are still asking the wrong question. They ask: “How do we teach our students about AI?” The question employers are answering is: “Which graduates can build, deploy, and maintain AI-enabled systems in real environments?” This gap between literacy and readiness is where too many graduates are falling through.
A survey by FDM Group found that over half of graduate jobs now require AI skills, with 54% of organisations saying all early-career roles will require AI ability. Yet only 6% of teams currently possess strong AI skills. When asked about specific skills required in graduate roles, 25% of employers identified prompt engineering as a top requirement. But critically, 22% highlighted critical thinking and applied problem solving, while 21% cited data engineering and pipeline development. In terms of soft skills, 18% considered critical thinking important, 15% valued creativity and innovation and 14% emphasised ethical judgement.
The Graduate Management Admission Council’s 2025 Corporate Recruiters Survey reinforces this picture. Problem-solving and strategic thinking remain the top skills employers desire today and tomorrow, while knowledge of using AI tools tops the list of skills employers value most five years from now. Globally, 31% of employers indicated that knowledge of using AI tools is an important skill when hiring, up from 26% in 2024.
Employers want graduates who can do more than generate output from a chatbot. They want people who can think critically about that output, analyse the data that feeds it, make ethical judgements about its use and apply genuine domain expertise to interpret its meaning.
The following are the four pillars of genuine AI literacy that should exist in our graduates.
Critical thinking
The most dangerous thing about Generative AI is not related to what it cannot do, but it is more focused on what it can do plausibly, confidently and often incorrectly. Hallucinations, biased outputs and subtle errors are baked into the technology. A graduate who cannot critically evaluate AI-generated content is not a knowledgeable user; they might be a liability to the organisations they are joining.
Students need to be taught to question AI outputs, verify claims against primary sources, recognise when an AI is confidently wrong and understand the limitations of the models they are using.
Data literacy
An AI model is only as good as the data it consumes. Yet the data literacy gap in the workforce is staggering. Deploying AI without data literacy is like giving someone a powerful tool without teaching them how to handle it safely. Universities must ensure that graduates understand data provenance, quality, bias and interpretation, not just how to prompt a model to analyse a dataset.
Ethics and Responsible AI
Ethics and Responsible AI are essential components of AI education and should be integrated into the curriculum to ensure that future AI professionals develop technologies that are fair, transparent, accountable and beneficial to society. As AI systems increasingly influence critical areas such as healthcare, education, finance and public services, students must understand not only the technical aspects of AI but also the ethical implications of their design and deployment.
Including ethics and Responsible AI in AI education helps learners recognise and mitigate issues such as bias, privacy violations, discrimination, misinformation and unintended societal impacts. It also fosters a culture of responsible innovation, encouraging students to consider the human, legal and social consequences of AI-powered solutions. By embedding these principles into the curriculum, educational institutions can prepare graduates to build trustworthy AI systems that align with ethical standards, regulatory requirements and societal values.
Domain expertise
In the AI era, domain expertise is more valuable than ever because only human experts can truly judge AI-generated output for quality and accuracy. Universities should not simply turn all degrees into computer science courses; instead, they must train graduates to be deep experts in their own fields along with being able to critically and effectively use AI tools.
To conclude, an AI-ready curriculum should be designed to combine intensive hands-on learning with the development of uniquely human capabilities. It needs to include real-world projects, practical application and portfolio-based assessment, enabling students to graduate with tangible evidence of their skills while also fostering critical thinking, rigorous data analysis, ethical reasoning and deep domain expertise. These competencies prepare graduates not only to work effectively with AI technologies but also to contribute value that extends beyond what AI alone can achieve.
