AI Applications — Module Assessment

Question 1

In most approved medical AI systems, what role does the AI play in a diagnosis?

Accepted Answer

It supports a clinician by flagging or prioritizing findings for review.

Answer

It replaces the clinician and issues the final diagnosis.

Answer

It prescribes treatment directly to the patient.

Answer

It has no contact with real patient data.

Question 2

Why can a medical AI model that performs well in testing still fail in a new clinic?

Accepted Answer

Because different equipment or patient populations can differ from its training data.

Answer

Because mathematics stops working outside a laboratory.

Answer

Because AI models physically wear out over time.

Answer

Because clinicians deliberately feed it wrong images.

Question 3

What does a recommendation system primarily optimize for?

Accepted Answer

A defined business objective such as engagement or sales.

Answer

The long-term wellbeing of each individual user.

Answer

Complete fairness across all customer groups.

Answer

The lowest possible price for the customer.

Question 4

How can a machine-learning fraud-detection system end up being unfair?

Accepted Answer

It can inherit bias from historical data and penalize certain groups.

Answer

It cannot be unfair because it uses only numbers.

Answer

It randomly declines transactions with no pattern at all.

Answer

It only ever produces false alarms, never real detections.

Question 5

In the standard autonomy loop, what is the job of the planning stage?

Accepted Answer

Predicting how others will move and choosing a safe path to follow

Answer

Turning camera and lidar data into a labeled model of the surroundings

Answer

Converting a chosen path into steering, throttle, and braking commands

Answer

Combining several sensors so their strengths cover each other's gaps

Question 6

What did the DARPA Grand Challenge demonstrate about autonomous driving?

Accepted Answer

That the field went from zero finishers in 2004 to five finishers in 2005, proving the approach could work

Answer

That full Level 5 autonomy was achieved and ready for public roads by 2005

Answer

That no vehicle could ever complete an off-road autonomous course

Answer

That human drivers are unnecessary in all modern vehicles

Question 7

Why do we say AI complements rather than replaces the scientific method?

Accepted Answer

Because AI proposes and prioritizes candidates, while experiment and peer review still confirm them

Answer

Because AI can prove new findings on its own without any experiments

Answer

Because AI is too slow to be useful in modern research

Answer

Because scientists are no longer needed once a model is trained

Question 8

What did DeepMind's AlphaFold achieve at CASP14 in 2020?

Accepted Answer

It predicted protein 3-D structures with accuracy near experimental quality, largely solving a long-standing problem

Answer

It designed entirely new living organisms from scratch

Answer

It replaced all laboratory experiments in biology

Answer

It proved that protein shape does not affect protein function

Question 9

What is the most defensible summary of the evidence on intelligent tutoring systems?

Accepted Answer

Well-designed systems can produce real but modest and uneven gains.

Answer

They reliably match a one-to-one human tutor in every setting.

Answer

They have been proven useless in controlled studies.

Answer

They only work for mathematics and never for other subjects.

Question 10

Why might the same adaptive-learning platform help one school yet widen inequality at another?

Accepted Answer

Differences in devices, connectivity, and adult support change what the tool can deliver.

Answer

The software secretly performs better for wealthier users by design.

Answer

Adaptive platforms only function in certain countries.

Answer

The tool refuses to run without a human tutor present.

Question 11

Which statement best reflects the current status of copyright for purely AI-generated work in the United States?

Accepted Answer

Copyright offices have questioned protection because works are expected to have human authorship.

Answer

It is fully protected the moment the model produces it.

Answer

It is illegal to create such work at all.

Answer

Ownership always belongs to the company that built the model.

Question 12

Why can a shift to AI-generated content help a studio's budget yet still harm individual creators?

Accepted Answer

Because it can cut costs while displacing workers and using styles or likenesses without consent.

Answer

Because AI tools are always more expensive than hiring people.

Answer

Because generated work is guaranteed to be lower quality.

Answer

Because studios are legally required to keep all staff.

Question 13

According to the lesson, AI applications tend to deliver the most durable value when the system:

Accepted Answer

Augments skilled humans rather than replacing them

Answer

Fully replaces human experts to cut costs

Answer

Reaches the highest possible score on a benchmark

Answer

Operates with no human oversight at all

Question 14

A headline claims an AI 'matches doctors at diagnosis.' Which question from the framework best exposes its likely limits?

Accepted Answer

What happens on the hard, rare, or high-stakes cases, and who is accountable when it is wrong?

Answer

How fast does the model run on modern hardware?

Answer

How large was the training dataset?

Answer

Which company built the model?