- Subjects
- Digital Society
- Question bank
- 3.6 Artificial intelligence
3.6 Artificial intelligence questions for IB Digital Society
100 questions on Artificial intelligence, written by Augment in the style of IB exam papers. Six of them are below in full, to answer on paper.
- IB Digital Society
- Part of 3 Content
- SL and HL
Questions by paper and level
SL and HL questions are for students at either level. HL only questions cover what only HL students study.
| Paper | SL and HL | HL only | All |
|---|---|---|---|
| Paper 1 | 56 | 19 | 75 |
| Paper 2 | 11 | 0 | 11 |
| Paper 3HL only | 0 | 14 | 14 |
| Every paper | 67 | 33 | 100 |
Counted on
Questions to try
Written by Augment in the style of IB exams, not past papers. Answer them on paper, then check each one against its markscheme with a free account.
Satellite evidence of destroyed villages
In many civil wars, villages are burned or demolished in remote areas that investigators cannot safely visit, and the forces responsible often deny that the attacks took place.
Clearsight, an international human rights organization, uses an intervention to address this. It buys high-resolution commercial satellite images of conflict regions every few days. An artificial intelligence (AI) model, trained using supervised learning on thousands of labeled images, compares each new image with an older image of the same place and flags buildings that appear to have been destroyed. Analysts check every flagged site, compare it with witness statements and videos posted online, and publish reports that are sent to international courts and United Nations investigators.
It is claimed that this intervention will provide reliable evidence that brings those responsible for attacks on civilians to justice.
To what extent do you agree with this claim?
12 marks
Focus scores at Halvorsen School
Background
Halvorsen School, for students aged 11 to 16, is trialing FocusView in four classrooms. Cameras at the front of each room feed an AI system that gives every student, and the whole class, a "focus score" from 0 to 100, shown live on the teacher's tablet.
Source A: How FocusView works
Every 10 seconds the camera captures an image of the class. A face-detection algorithm finds each student's face. A machine learning model, trained on thousands of labeled classroom videos, then classifies each face and head position as focused, distracted or unclear.
Each student's results are turned into a score for the lesson. If the class score falls below 50, the teacher's tablet shows an alert. Images are deleted once they have been analyzed; scores are kept until the end of the school year.
Adapted from the FocusView information sheet for schools.
Source B: Average class focus score in one trial classroom

Halvorsen School trial report, average of four weeks.
Source C: A teacher's view
The alerts made me notice when I had been talking for too long, and I started breaking up my lessons. But students have worked out what the camera likes. They sit up and stare at the board while thinking about something else entirely. In group work, when students turn to face each other, the scores drop, so the system seems to reward the kind of lesson where I talk and they listen.
Daniel Okafor, science teacher, Halvorsen School.
Source D: Letter from the Parents' Association
We were told about the trial after it had started. Our children did not choose to be analyzed every 10 seconds, and there is no way to opt out. Several parents of autistic students report that their children, who often avoid eye contact, are labeled distracted when they are listening carefully. We also learned that scores may be included in end-of-year reports. We ask the school to pause the trial until parents and students have been consulted.
Halvorsen School Parents' Association, 2025.
- (a)
With reference to Source A, outline how supervised learning could be used to train FocusView's model.
2 marks
- (b)
With reference to Source B, suggest two reasons why the class focus score is lower in the afternoon.
4 marks
- (c)
Compare and contrast what Source C and Source D reveal about how FocusView affects students.
6 marks
- (d)
With reference to Sources A–D and your own knowledge, evaluate the use of attention-scoring cameras such as FocusView in classrooms.
12 marks
Credit for workers without bank accounts
This pre-released statement sets out the real-world nature of a challenge and two interventions related to the digital society higher level extension. Candidates should use it to plan and carry out extended inquiries before paper 3.
Challenge topic: Global well-being
Area for inquiry: Local and global inequalities
Candidates should also consider relevant connections to other digital society concepts, content and contexts. Teachers are advised to explore several scenarios within this context.
About the challenge
The World Bank's Global Findex Database 2021 reported that about 1.4 billion adults worldwide had no account at a bank or other financial institution. Many of them work in the informal economy as market traders, farmers or day laborers, and many are women. Without savings accounts or a credit history, they cannot borrow from banks to buy stock or equipment, so they turn to informal moneylenders who may charge very high interest. This keeps them in low-income work and widens the gap between those inside and outside the formal economy.
Mobile money, which lets people store and send money on an ordinary cell phone, has spread quickly in many countries since Kenya's M-Pesa was launched in 2007. It has made payments easier, but borrowing is still difficult because lenders have little information about people with no banking history. Digital lenders now use new sources of data to judge who should receive credit, while community groups are using digital tools to strengthen older ways of saving and lending together.
This statement introduces two digitally based interventions designed to give low-income informal workers fair access to credit. Each leads to a range of impacts and implications that should be evaluated against the criteria of equity, acceptability, cost, feasibility, innovation and ethics.
In the examination, candidates will be given a source booklet containing four sources about a scenario.
About the interventions
Intervention 1: A digital lender offers small loans through a smartphone app. A machine learning model calculates each applicant's credit score from alternative data collected from their phone.
Intervention 2: Community savings groups, whose members pool regular savings and lend to each other, keep their records in a shared digital ledger app that also builds a credit history for each member.
About the stakeholders
informal workers without bank accounts
savings group members and leaders
digital lending companies
banks and financial regulators
Additional terminology
Alternative data
Credit history
Credit score
Digital ledger
Economic stratification
Financial inclusion
Informal economy
Interest rate
Microloan
Mobile money
Predatory lending
Rotating savings and credit association (ROSCA)
Unbanked
Notes
It is not necessary for candidates to study the mathematics of credit-scoring models.
Source booklet
Source A: Tamu Market
Tamu Market is the largest open-air market in Kisanga, a fast-growing town in East Africa. The Tamu Market Traders' Savings Cooperative (TMTSC) has 640 members, 80 % of them women, who sell vegetables, clothes and household goods or run small repair stalls.
Almost every member has a mobile money account, but fewer than one in ten has a bank account. When traders need money for stock, many borrow from moneylenders in the market who charge 20 % interest a month.
Most TMTSC members also belong to small savings groups of 20 to 30 people. Each week, members pay a fixed amount into the group, and the group lends from the pooled money. Records are kept in paper notebooks by the group's treasurer.
Source B: Two proposals for TMTSC
QuickLend, an app from the digital lender Mavuno Digital:
with the applicant's permission, the app reads their mobile money history, airtime top-ups, contact list size and the phone's location history
a machine learning model trained on Mavuno's past borrowers gives a score within two minutes
first loans are up to USD 20 for 30 days, with a fee of 9 %; limits rise after each loan repaid on time
Kikundi Ledger, a free app from a local nonprofit:
the group treasurer records each member's weekly payments and loans on a phone; the app works offline and synchronizes when there is a signal
each member gets a text message receipt after every payment
loans come from the group's own pooled savings, decided by a vote of the members
after 12 months, a member can choose to share their record with a partner bank as a credit history
Source C: Grace, a tailor
I want to buy an electric sewing machine. It costs about USD 300. My savings group can only lend me USD 60 because the pot is small and others need loans too. QuickLend says the money arrives on my phone in minutes. But my phone is registered in my husband's name, so I don't know what it will say about me.
Source D: Fatuma, leader of a savings group
Our group knows who is reliable, because we see each other every day in the market. That is our strength. But last year our treasurer moved away and the notebook went with her; we lost two years of records. Some of our older members cannot read well, so they just trust whatever the treasurer writes.
- (a)(i)
Outline what is meant by financial inclusion.
2 marks
- (a)(ii)
Identify two other types of alternative data, apart from those listed in Source B, that a credit-scoring model could use to predict whether a borrower will repay a loan.
2 marks
- (b)(i)
Explain one reason why QuickLend's machine learning model may give women traders lower credit scores than men who are equally able to repay.
3 marks
- (b)(ii)
Explain one way in which Kikundi Ledger could help a trader like Grace (Source C) to obtain a larger loan in the future.
3 marks
- (c)
Digital lenders increasingly decide who can borrow by analyzing data collected from applicants' phones.
To what extent is it acceptable for lenders to make credit decisions in this way?
8 marks
- (d)
TMTSC has to choose between two interventions to address the challenge of giving its members fair access to credit:
Intervention 1: recommending QuickLend, the app-based lender that scores applicants from their phone data
Intervention 2: introducing Kikundi Ledger to all of its members' savings groups
Recommend which of the interventions TMTSC should choose.
12 marks
Drones for clearing landmines
Long after wars end, landmines and unexploded bombs stay buried in fields and roads. Around the world they kill or injure thousands of people every year and stop families from returning to farm their land.
After a ten-year conflict, the government of Arelia and a demining non-governmental organization (NGO) have started using autonomous drones to survey farmland. The drones fly pre-set routes a few meters above the ground carrying cameras, thermal sensors and ground-penetrating radar. A machine learning model analyzes the data and marks likely mines on a digital map. Human deminers then remove each marked item. Before the drones, deminers searched every square meter by hand with metal detectors.
It is claimed that this intervention will make the land safe for families to return to quickly.
To what extent do you agree with this claim?
12 marks
Counting birds with Fieldnote
Background
Fieldnote is a free app run by a university wildlife research center. Volunteers photograph or record birds, and an image- and sound-recognition model suggests the species. Each sighting is uploaded with its location, date and time to a national database used by researchers.
Source A: A Fieldnote sighting
When a volunteer takes a photo or records a song, the app suggests up to three species with a confidence score, for example Song thrush 91 %, Mistle thrush 7 %. The volunteer picks one and the sighting is uploaded with the phone's GPS location and a time stamp. Rare species, or sightings with a confidence score under 60 %, are sent to a volunteer expert to check.
Adapted from the Fieldnote help pages.
Source B: Sightings checked by experts in 2024

Fieldnote annual report. A random sample of 5,000 sightings was checked by experts.
Source C: Volunteers' views
Priya, retired teacher: I never knew one bird from another. Now I record what visits my garden every morning and feel part of something bigger.
Dev, aged 15: I use it on the way to school. Last spring I logged a hoopoe and an expert messaged me to say it was the first one in our county for six years.
Mark, farmer: I stopped using it when I saw that my sightings showed exactly where the owls nest on my land.
Fieldnote community newsletter, 2024.
Source D: A researcher's blog
Fieldnote has given us more records in three years than volunteers collected on paper in the previous thirty. But the data have limits. Most sightings come from towns and gardens, so we know far more about where people live than about remote hills. Common birds are recorded again and again while shy species are missed. And when the app is unsure, beginners tend to pick the first suggestion. We correct for these biases with statistics, but we cannot fix data that was never collected.
Blog of a researcher at the wildlife research center, 2025.
- (a)(i)
With reference to Source A, identify one item of metadata uploaded with each sighting.
1 mark
- (a)(ii)
With reference to Source A, identify one type of machine learning the app's recognition model could have been trained with.
1 mark
- (b)
With reference to Source B, explain two reasons why some groups of birds are identified less accurately than others.
4 marks
- (c)
Compare and contrast what Source C and Source D reveal about Fieldnote's value to volunteers and to researchers.
6 marks
- (d)
With reference to Sources A–D and your own knowledge, evaluate the use of apps such as Fieldnote to collect data for scientific research.
12 marks
Newcomers and the services they need
This pre-released statement sets out the real-world nature of a challenge and two interventions related to the digital society higher level extension. Candidates should use it to plan and carry out extended inquiries before paper 3.
Challenge topic: Global well-being
Area for inquiry: Changing populations
Candidates should also consider relevant connections to other digital society concepts, content and contexts. Teachers are advised to explore several scenarios within this context.
About the challenge
People move for many reasons: to find work, to join family, or to escape war, persecution and disasters. In May 2022 the United Nations Refugee Agency (UNHCR) reported that the number of people forcibly displaced worldwide had passed 100 million for the first time. Many more move between countries as migrant workers or students.
When newcomers arrive in a city, they need shelter, health care, schooling for their children, legal advice and work. Getting these services is often difficult. Many arrive without identity documents or proof of their qualifications, so they cannot show who they are or what they can do. Procedures are complex and written in an unfamiliar language, and the organizations that help are often overwhelmed. False information spread by smugglers or on social media can lead people to make dangerous decisions.
Digital systems can help newcomers prove who they are and find accurate information, but newcomers are also among the people least able to refuse when they are asked for their personal data.
This statement introduces two digitally based interventions designed to help newcomers gain access to the services they need. Each leads to a range of impacts and implications that should be evaluated against the criteria of equity, acceptability, cost, feasibility, innovation and ethics.
In the examination, candidates will be given a source booklet containing four sources about a scenario.
About the interventions
Intervention 1: A portable digital identity wallet. Newcomers store digitally signed credentials, such as identity, vaccination and education records, in an app and choose which ones to share with service providers.
Intervention 2: A multilingual AI chatbot. Newcomers ask questions in their own language on a popular messaging app and receive answers about procedures and services, with complex cases passed to human caseworkers.
About the stakeholders
migrants, refugees and asylum seekers
nongovernmental organizations (NGOs) and their caseworkers
government agencies responsible for migration and public services
employers, schools and health care providers
Additional terminology
Asylum seeker
Biometric data
Digital identity
Digital wallet
Function creep
Hallucination (in AI)
Internally displaced person
Large language model
Machine translation
Misinformation
Refugee
Verifiable credential
Notes
It is not necessary for candidates to study the technical details of digital signatures.
Source booklet
Source A: Harborgate Welcome Center
Harborgate is a port city in southern Europe. The Harborgate Welcome Center (HWC) is an NGO that helped 14 000 newcomers from more than 30 countries last year. About a third arrived with no identity documents.
Newcomers wait an average of 11 days for a first meeting with a caseworker. HWC's 22 caseworkers speak six languages between them; newcomers speak more than 25.
Source B: The two proposals
PassKey wallet, a free app from the technology company Nordline Systems:
HWC checks any documents a newcomer has and issues digitally signed credentials to their phone
to recover a lost phone, each user's fingerprints and a photo of their face are stored in a central database run by Nordline
users choose what to share; for example, they can prove they are over 18 without showing their date of birth
the national migration agency has asked for access to the database "to speed up asylum checks"
AskHarbor, a chatbot built by HWC on a commercial large language model:
available 24 hours a day on a popular messaging app, by text or voice, in 12 languages
answers only from HWC's own verified information; hands over to a caseworker when it cannot answer or detects an emergency
conversations are stored for 30 days to improve answers
Source C: Lena, an HWC caseworker
In our trial, AskHarbor answered about 1800 questions a week, and seven in ten answers were fully correct the first time. That frees us for the hard cases. But it once gave a family the wrong deadline for an appeal. If I hadn't spotted it, they could have lost their case.
Source D: Two newcomers
Samir, 34: I'm an engineer, but my diploma was lost on the journey, and nobody here believes me. A wallet with proof would change everything. But my fingerprints in a database? Where I come from, lists like that were used to find people like my brother.
Halima, 19: I ask my phone everything. A chatbot at 2 a.m. is better than waiting 11 days.
- (a)(i)
Outline the difference between a refugee and an internally displaced person.
2 marks
- (a)(ii)
Identify two reasons why newcomers may arrive in a city without identity documents.
2 marks
- (b)(i)
Explain one way in which the PassKey wallet could let a newcomer prove something about themselves without revealing more personal data than is needed.
3 marks
- (b)(ii)
Explain one reason why AskHarbor may give a newcomer incorrect information.
3 marks
- (c)
The national migration agency has asked for access to the PassKey database.
To what extent should newcomers be required to provide biometric data in order to receive help?
8 marks
- (d)
HWC has to choose between two interventions to address the challenge of helping newcomers gain access to the services they need:
Intervention 1: the PassKey digital identity wallet
Intervention 2: the AskHarbor multilingual chatbot
Recommend which of the interventions HWC should choose.
12 marks
Practice all 100 in Augment
Free with an account. Every question has its markscheme, and Augment keeps the ones you got wrong so you can come back to them.
More questions in Content
- 3.1Data
- 3.2Algorithms
- 3.3Computers
- 3.4Networks and the internet
- 3.5Media
- 3.6Artificial intelligenceThis page
- 3.7Robots and autonomous technologies
More on Artificial intelligence
- 3.6 Artificial intelligence notesWhat Artificial intelligence covers, explained, with its key terms.
- IB Digital Society question bankEvery Digital Society topic counted, and one question from each with its markscheme.
- Digital Society exam formatEach Digital Society paper, how long it is and its share of the grade.
Check your answers in Augment
A free account opens all 100 questions on Artificial intelligence with their markschemes, and keeps a record of what you have marked.