Over the past two years, students have come to know AI through one interface: the Chatbot. You type a prompt, the system gives a response. It can summarise an article, explain a concept, draft an email, or generate a first version (or even the last) of an assignment. That was already very significant, with wide impact at several levels of student learning journeys.

But I think a more important shift is now taking place—even while most of us are still trying to adapt to the previous wave.

AI is becoming less like a clever reply machine and more like a system that can participate in execution.

That is why this is not only a technology story. It is also a work story, a learning story, and an institutional story.

AI is moving from Answering to Acting

This shift matters. A chatbot mainly responds. An agent can work towards an outcome. It can break a task into steps, use tools, retrieve information, check progress, and continue until a goal is reached.

It is moving from systems that mainly respond to questions, to systems that can increasingly pursue goals, use tools, and complete multi-step tasks.

This is what makes recent developments in “Agentic AI” so important. The key issue is no longer only whether AI can produce a fluent answer or if we can detect "Cognitive Offloading" in students' deliverables. The deeper issue is that AI can help execute parts of real work, or even the full work.

That distinction matters a great deal for universities. It goes much further than "teach them a few prompting tricks, and our job is done!" which seems to be the default response to these challenges.

Why this matters

For students, this means that the old idea of academic effort also shifts. The question is no longer only, “Can I produce an answer?” The right questions are “Can I define the problem well, guide the process well, and judge the result well?”

This is a much more demanding question, and much harder to teach "How to" do it well.

Why prompting is not enough

Many students assume that being “good at AI” means knowing a few clever prompting tricks. That view is understandable, but too shallow.

Prompting matters. Of course it does. A clearer prompt often leads to a better output. But prompting is only a surface skill. It is a simple technique, not a true capability.

A good prompt cannot compensate for weak judgment. It cannot replace domain understanding. It cannot detect every risk, omission, or misleading answer. It cannot decide whether an output fits the real problem context, a professional standard, or an ethical boundary.

So, we have to stop teaching prompting tricks and move on to the real thing!"

The right prompt becomes more or less obvious once we have a workable understanding of the problem we want to address.

Real AI-readiness is much deeper

It includes at least four things:

  • Problem framing: knowing what the real task is, what success looks like, and what constraints matter.
  • Judgment: knowing what to trust, what to question, and what to reject.
  • Oversight: knowing when AI can assist, when humans must intervene, and who remains accountable.
  • Application of Domain Expertise: knowing how AI should be used in a real field such as education, business, health, sport, law, finance, public policy, and so on. (Can you think of an area where these agents would not add value?)

For students, the future advantage will not go simply to those who can “talk to AI.” It will go to people who can direct it, evaluate it, and apply it wisely in their own domains to create real value.

What students should get from their "Education" now

If routine execution becomes cheaper, then higher-value human capabilities become more important.

This is where universities should be very careful not to send the wrong message. Students do not need an AI curriculum built around prompt gimmicks. They need a curriculum that strengthens the human capabilities that will always remain essential in a world with AI agents ready to be deployed everywhere.

Four stand out.

  • First, creative thinking. Students need to learn how to reframe a problem, not just respond to one. Better outputs often start with better questions.

  • Second, critical thinking. Students need to evaluate evidence, identify weak reasoning, notice what is missing, and resist the temptation to confuse fluency with truth.

  • Third, systems thinking. Real-world problems rarely sit inside one neat box. Students need to understand interdependence, trade-offs, second-order effects, and unintended consequences, often spanning different expertise domains.

  • Fourth, an interdisciplinary mindset. AI becomes genuinely useful when technical capability is combined with domain context. The most valuable graduates will likely not be those who know the most commands or more "prompting tricks", but those who can connect AI tools to the interconnected realities of a complex world.

The real challenge for universities

For higher education teaching and learning, the real issue is not whether students are using AI. They already are, all the time.

The deeper issue is whether the institution is preparing them to use AI well.

That means moving beyond defensive conversations about detection and misconduct, flagging AI usage scores. It means asking harder questions. Are we redesigning assessment to value judgment, process, and application rather than routine output alone? Are we helping staff understand what good AI-supported work looks like? Are we teaching students how to verify, supervise, and document their use of AI responsibly?

An institution does not become AI-ready because it purchases a platform licence or publishes a short guideline. It becomes AI-ready when it develops a culture of wise AI use.

As a famous management guru once said: "Culture eats Strategy for breakfast!"

Final thought

Agentic AI will make some forms of low-value execution cheaper. But it will not make human judgment less important. Quite the opposite. It raises the value of judgment, oversight, context, and responsibility.

So the question for students is not, “How do I become good at prompting?”

A better question is: “How do I become the kind of person who can use AI wisely and effectively in the real world?”

And the question for universities is closely related: "Are we still rewarding routine production, or are we preparing students for a world in which the premium lies in directing, evaluating, and applying intelligence well?

This is the real educational challenge now.

Let’s build education that is rigorous, human-centred, and genuinely future-ready.

Let’s build human-centred futures—together.

Thanks for reading.
— HM