Computer ScienceFoundation18 min read

Applications of Artificial Intelligence

Where AI is genuinely used, what it is worth, and what it costs

This topic appears in:

01

What AI is actually good at

AI is not general cleverness. It is very good at a specific kind of task: finding patterns in large amounts of data, quickly, consistently and without getting tired. Every successful application below is that same capability pointed at a different problem.

It is correspondingly poor at anything requiring genuine understanding of an unfamiliar situation, at explaining itself, and at knowing when it is wrong — which is why the useful applications pair it with a human rather than replacing one.

FieldWhat AI doesWhat it does not do
Healthcareflags suspicious regions on a scandecide treatment or tell the patient
Agricultureidentifies crop disease from a photographknow that the field is waterlogged
Financeflags unusual transactions as possible fraudjudge whether a person is trustworthy
Transportrecognises lanes, signs and obstacleshandle a situation never seen in training
Educationadapts practice questions to a student's levelnotice that a child is upset
Languagetranslates and transcribes speechgrasp context, sarcasm or intent reliably
02

Applications that matter in Pakistan

The syllabus asks for real uses rather than science fiction, and the strongest answers are local.

In agriculture, a farmer photographs a leaf and a model identifies the disease and suggests treatment — reaching farms an agricultural extension officer could never visit often enough. In healthcare, AI reading chest X-rays for tuberculosis lets a small number of radiologists cover a large population by flagging the scans that need attention. In banking, fraud detection watches transaction patterns continuously in a way no human team could. And in language, translation and speech recognition make services usable by people who do not read English or Urdu comfortably.

Why triage is the pattern that works

Notice that the medical example does not diagnose. It sorts — putting the scans most likely to show a problem at the top of the radiologist's list. The human still decides, but spends their time where it matters most. Applications built this way succeed because the AI's mistakes are caught, and because the value comes from speed rather than from replacing judgement.

03

The costs, honestly stated

Every exam question about AI applications wants the drawbacks as well, and vague worry scores poorly. The specific costs are these.

Employment. Automation removes some roles and creates others, but rarely for the same people in the same place — a data-entry clerk does not become a machine-learning engineer. Privacy. These systems need large amounts of data, often personal. Accountability. When an automated decision is wrong, it is frequently unclear who is responsible. Environmental cost. Training a large model consumes a great deal of electricity. Dependence. A skill that is always automated is eventually lost, which matters when the system fails.

Worked example

A hospital proposes replacing its radiologists with an AI system that reads scans more accurately on average. Evaluate the proposal.

  1. Note the benefit: it is faster, works continuously, and does not tire or become less consistent late in a shift.Consistency is a genuine advantage — human accuracy falls over a long day.
  2. But "more accurate on average" hides the distribution of its errors.A system that is better overall may fail badly on an unusual case, which is exactly where expertise matters most.
  3. It cannot explain its reasoning, so a patient cannot be told why, and a wrong result cannot be argued with.Accountability for a medical decision cannot rest on an unexplainable output.
  4. It was trained on a particular population and may perform worse on patients unlike that group.A model trained abroad may not transfer to a different population.
  5. The sound proposal is AI as triage, with radiologists reviewing — the same accuracy gain, with the failures caught.Replacing the human removes the safeguard that makes the accuracy gain safe to use.

Use it to prioritise scans, not to replace radiologists — the accuracy gain is real, but its errors need a human to catch them.

04

Evaluating a proposal

Questions in this chapter usually describe a proposed use and ask whether it is a good idea. Four questions produce a full answer every time.

  • What does it replace? A tedious task nobody wants, or a judgement someone is accountable for?
  • What happens when it is wrong? An irritating recommendation, or a person denied medical care?
  • Whose data trained it, and does that match who will be affected?
  • Can the person affected find out why? If not, the decision should not be automated alone.

Before you leave this chapter

  1. AI finds patterns in large data quickly and consistently. That is the whole capability.
  2. The applications that work pair it with a human — triage rather than replacement.
  3. Name specific costs: employment, privacy, accountability, energy, deskilling.
  4. Ask what happens when it is wrong. The higher the stakes, the more human oversight is needed.
  5. A model trained on one population may not transfer to another.
05

Why an accurate model can still fail in the field

A model reported as 95% accurate has been measured on some test set. Whether that number survives contact with real use depends on whether the people it will be applied to resemble the people it learned from.

A disease-detection model trained on scans from one hospital may meet different equipment, different patient demographics and different image quality elsewhere, and quietly perform far worse. The accuracy figure was never wrong — it was answering a narrower question than the one that matters.

  • Ask what the accuracy was measured on. A number with no stated test population means very little.
  • Ask how the errors are distributed. A model 95% accurate overall may be 99% accurate for one group and 70% for another.
  • Ask what a wrong answer costs. The same accuracy is fine for a recommendation and unacceptable for a diagnosis.
  • Ask whether it can be monitored. A model whose performance is never re-checked after deployment will drift as the world changes.

Accuracy is misleading for rare events

If one transaction in a thousand is fraudulent, a model that simply answers "not fraud" every single time is 99.9% accurate and completely useless. That is why fraud and disease detection are judged on how many real cases they catch and how many false alarms they raise, not on raw accuracy. It is the single most useful thing to know about evaluating these systems.

Practice questions

6 questions · 20 marks · full working on every one

Try each one on paper first, then open the working. The marks are shown where they are actually awarded, because that is where they are actually lost.

Short questions

3 · 6 marks

Two marks each, in the style of the short-question section of the paper. Answer in two or three lines.

SQ1[2 marks]
Give one application of AI in healthcare and state its main benefit.
Model answer

AI can analyse medical images such as chest X-rays and flag those most likely to show disease, so a radiologist reviews the urgent cases first. The benefit is speed and consistency: it works continuously without fatigue, letting a small number of specialists cover a large population.

Examiner tip. Choose "flags for review" rather than "diagnoses". The triage framing is more accurate and shows you know where the human fits.

SQ2[2 marks]
State two disadvantages of using AI in decision-making.
Model answer

It usually cannot explain how it reached a decision, so a person affected cannot be told why. And it reproduces the biases in its training data while appearing objective, which can make unfair outcomes harder to challenge.

Examiner tip. Two distinct disadvantages, each with a consequence. "It might be wrong" is true of humans too and earns little.

SQ3[2 marks]
Explain one way AI can help farmers.
Model answer

A farmer can photograph an affected leaf with a phone and a model identifies the crop disease and suggests treatment. This reaches farms that an agricultural extension officer could not visit often enough, and gives an answer within seconds rather than days.

Examiner tip. Say why it is better than the existing alternative. "It helps farmers" without the comparison rarely scores both marks.

Solved numericals

2 · 8 marks

Full working, one step per line, with the marks shown where they are awarded.

N1[4 marks]
Describe two benefits and two risks of self-driving vehicles.
Full working
  1. Benefit: they do not tire, become distracted or drive under the influence, and human error causes most collisions[1]
  2. Benefit: they can give independent mobility to people unable to drive, and can coordinate to reduce congestion[1]
  3. Risk: an unfamiliar situation not represented in training may be handled badly, and roads present many such situations[1]
  4. Risk: responsibility for a collision is unclear — the owner, the manufacturer or the software developer — and the vehicle is a target for interferenceaccept job losses among professional drivers[1]

Benefits: no fatigue or distraction, and mobility for non-drivers. Risks: unfamiliar situations, and unclear liability.

Examiner tip. The liability point is the strongest risk to raise, because it is a problem no improvement in the technology solves on its own.

N2[4 marks]
A bank introduces an AI system to detect fraudulent transactions. Explain why this suits AI, and state one problem it may cause customers.
Full working
  1. There are millions of transactions, far more than any human team could examine[1]
  2. Fraud shows as an unusual pattern, and finding patterns in large volumes of data at speed is exactly what these systems do well[1]
  3. Problem: false positives — a legitimate but unusual transaction is blocked, such as a card used abroad on holiday[1]
  4. The customer is inconvenienced and may not be told why, since the system cannot explain the pattern it reacted tothe inability to explain is the sharper point[1]

The volume and the pattern-matching suit AI; the cost is blocked legitimate transactions with no explanation.

Examiner tip. False positives are the standard cost of any detection system. Naming a concrete example — a card blocked abroad — makes the point specific.

Long questions

1 · 6 marks

Theory and numerical together, as they appear in the long-question section.

LQ1[6 marks]
A government department proposes using AI to decide which applicants receive financial assistance, replacing the current human assessors.
  1. Give two arguments in favour of the proposal.
  2. Give two arguments against.
  3. Recommend how the system should be used, with justification.
Mark scheme
  1. In favour: decisions would be made far faster and applicants would wait less[1]
  2. In favour: the same rules would be applied to every applicant, removing the inconsistency between individual assessors[1]
  3. Against: trained on past decisions, it would reproduce any historical unfairness while appearing objective[1]
  4. Against: an applicant refused assistance could not be told why, and would have no meaningful basis for appealthe accountability point[1]
  5. Recommendation: use the system to sort and prioritise applications, with a human assessor making every final decision[1]
  6. Justified because the speed and consistency are gained while accountability and the right to an explanation are preserved, and unusual cases still reach a personthe justification must address both sides[1]

(a) speed and consistency (b) inherited bias and no explanation (c) AI prioritises, humans decide — keeping the benefit and the accountability

Examiner tip. A recommendation that simply picks a side scores poorly. The mark is for a design that keeps the benefit while addressing the objection you have just raised.