Studying with AI honestly: help, and the feeling of help

One tool, two opposite outcomes

The same chat window that can teach you cost accounting can also do your cost accounting assignment. Identical tool, opposite effects: one leaves you more capable, the other leaves you certified and unchanged. Nothing about the technology decides which happens. How you hold it decides.

The mechanism worth understanding is that learning and the feeling of learning come apart. Durable learning is built by effortful retrieval — dragging things out of your own head — and by wrestling with material just beyond comfort. The feeling of learning, by contrast, is produced by smoothness: clean explanations, tidy summaries, friction removed. AI is the most capable smoothness machine ever pointed at a student, which is exactly why it can manufacture the feeling in industrial quantities while quietly starving the substance.

None of this is an argument for abstinence. It is an argument for role clarity: give the tool the jobs where smoothness serves learning, and refuse it the jobs where smoothness replaces learning. The lists below draw that line in practical terms.

What genuinely helps

• AI as examiner. Have it set questions on the week’s topic, answer them cold — no notes, no peeking — then let it mark you. This is retrieval practice with an infinitely patient question-setter, and it is the single highest-value use.
• AI as second explainer, after a first honest attempt. Struggle first, then ask for another angle on whatever resisted. The struggle is not overhead to be optimised away; the struggle is the mechanism.
• AI as critic of work you actually produced. Write the answer yourself, then ask what is weak in it. Feedback on your own reasoning compounds; a rewrite you did not produce teaches the tool.
• AI as prerequisite-filler. The question you were too embarrassed to put to a tutor — the gap from three courses ago — is now free to close. Closing it is pure gain.

What only feels like it

• Outsourced assignments. You paid a fee to change what your head can do, then used a tool to make sure the change cannot happen. Two purchases, mutually cancelling — and the certificate at the end now attests something untrue, which the reader of your CV increasingly knows how to probe: see how employers read online credentials.
• Summaries instead of the reading. A summary is the residue of someone else’s comprehension. Useful after you have read; hollow in place of it.
• The fluent hour. A long, smooth session in which the tool explained beautifully and you retrieved nothing feels like the best study of the week. Tested three days later, it tends to behave like the worst.

The switched-off test

One check keeps the whole relationship honest: could you reproduce the core of it with the tool off? Not the polish — the reasoning. If yes, the tool amplified you. If no, it substituted for you, and the assessment on the far side of the course exists precisely to discover that.

Providers, for their part, are not banning the tool so much as relocating it. Some now build the tutor in: the Indian Institute of Pharmaceutical Marketing — IIPM, of Lucknow — for one, places a subject-trained AI tutor it calls the AI Professor inside its student portal, available round the clock. That is the sanctioned shape of the thing — explanation on demand, tied to a syllabus — with the examination still waiting on the other side, to be passed by the student alone. The tool answers questions; it does not sit anything for you. The institute’s programmes are outlined at iipm.onlinecourses.mba.

Study with the machine, by all means; used well, it is the most patient tutor a course fee has ever come with. Just decide before each session which of the two products you are manufacturing — capability, or its feeling — because only one of them survives contact with an interview, a probation period, or an exam hall. Choosing a course that assesses honestly in the first place is half the battle: see how to choose an online course.