Redundant by design: AI and the need to rethink universities

In February 2026, Pakistan's Higher Education Commission (HEC) directed universities to introduce a mandatory three-credit-hour course on Artificial Intelligence across all degree programmes. Universities are responding in predictable ways: some are rushing to add AI literacy modules, others are announcing new centres and committees. Most simply want to tick the box and signal compliance.

This urgency is understandable, driven by real anxiety about graduate employability in an automated world. But it is also a trap, which lies in the illusion that adding a technical module will protect graduates, even as the broader system trains them in routine cognitive tasks. We are preparing students as if they will compete against other humans, when they are competing against the machine itself. On the terrain of compiling information, summarising data, and generating standard answers, the machine has already won. And most institutions have yet to notice.

The real question is not whether universities should adopt AI, but whether they understand what they are adopting AI into. A university is not a content warehouse. Its purpose is to develop judgment, intellectual maturity and technical competence. Learning at universities is not just the acquisition of information but the development of capabilities, the ability to apply knowledge and skills in real-world situations.

Yet our institutions are built to distribute knowledge, not to develop capabilities. This made sense in an age of scarcity, when access and learning appeared indistinguishable. That world no longer exists. Information is now abundant. The challenge is no longer access, but discernment. Access is no longer the problem. Judgment is. And it demands that we reimagine the architecture of our universities.

This is where Russell Ackoff, a pioneer of systems thinking, becomes useful. He warned that the more efficiently we do the wrong thing, the wronger we become. The HEC's directive is a necessary first step, but not sufficient. Bolting an AI module onto a scarcity-era distribution system will not protect individuals or institutions from redundancy. Ackoff argued for dissolution rather than solutions, not fixing the existing system but redesigning it so the problem itself disappears. That is the challenge now before us.

What would such a redesign look like? We can begin by looking at the evolution of the very technology we are trying to adopt. Over the last two decades, software architecture moved from rigid, linear "waterfall" models to flexible microservices and is now transitioning into dynamic, agentic networks that adapt in real time. Our educational architecture, however, remains stubbornly monolithic. Universities are still trapped in siloed departments, rigid faculties and slow-moving statutory bodies. It takes a Board of Studies and an Academic Council six months to approve a syllabus change; in that same window, AI models evolve through multiple generations. We are attempting to house a fluid, agentic technology inside a rigid bureaucracy.

Because AI integration requires architectural rethinking, a standardised manual cannot solve it. We must return to first principles and re-examine the fundamental relationship between the university and the act of learning. A genuine redesign begins by confronting three questions.

First, if universities aim to develop capabilities, why does infrastructure maximize administrative friction while classrooms eliminate academic struggle? Administrative processes, such as timetabling, hostel applications, fee issues and transcript delays, drain the mental energy learning demands. These should be frictionless. In the academic domain, the opposite is true. Struggle is not a flaw; it is how understanding is built. AI that prioritises efficiency risks making learning shallower. It should act as a Socratic interrogator, not an answer machine.

Second, why does academic assessment still rely on lagging indicators? When technology allows real-time visibility into learning, why not move toward formative assessment? Instead of judging final outputs, systems can observe how students navigate confusion, adapting difficulty and guiding learning in real time. While there has long been a movement toward "Continuous Quality Improvement" in higher education, institutions have historically lacked the round-the-clock visibility required to implement it properly.

Third, how can institutions claim to build adaptable minds when they punish failure? True development requires the safety to test ideas, fail and iterate. An AI-integrated architecture fundamentally lowers the cost of failure, transitioning the university from a courtroom that judges final outputs only into a laboratory that encourages rapid iteration.

Crucially, this laboratory mindset must also apply to the institution itself. Historically, brilliant pedagogical experiments have remained trapped in single classrooms because institutional architecture was too rigid to adopt them broadly. By leveraging continuous awareness, a successful learning model can now be rapidly refined and scaled from an isolated prototype into a mass-scale feature across the entire ecosystem.

Institutions cannot expect teachers and students to carry this redesign alone. The responsibility sits squarely with rectors, vice-chancellors, deans and faculty leadership. Those that embed these shifts will prevail. Those that merely comply while preserving the old model will become beautifully efficient and redundant.

But if we redesign our universities from delivery machines into environments of human development, we are forced to confront a deeper question. If machines can handle routine cognitive work and master skills at scale, what kind of human beings are we trying to form? That is no longer just an institutional challenge. It is a civilisational choice.