2.2 Expression questions for IB Digital Society

34 questions on Expression, 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 2 Concepts
  • 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.

PaperSL and HLHL onlyAll
Paper 124327
Paper 2606
Paper 3HL only011
Every paper30434

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.

Question 1 · DS-00008Paper 1 · HL only · 12 marks

Hiding faces in protest photos

Photos and videos shared online have helped protest movements spread their message. However, in some countries police and others use facial recognition on images posted online to identify people who took part in protests. Those identified may be arrested, lose their jobs or be harassed.

VeilCam is a free app made by an activist collective in the city of Darrowgate. When a user takes or uploads a photo or video in the app, an artificial intelligence (AI) model detects every face in it and blurs each one before the file can be shared. The app also removes the file's metadata, such as the date, time and GPS location. All processing happens on the user's phone.

It is claimed that VeilCam will protect protesters from being identified and punished for taking part in protests.

To what extent do you agree with this claim?

12 marks

Question 2 · DS-00279Paper 2 · 24 marks

An encyclopedia in our language

Background

Velan is a regional language spoken by about 600,000 people. The Velan Wiki, an online encyclopedia written and edited by volunteers, had 3,000 articles in 2022. In 2023 a grant from a large technology company paid for about 40,000 articles to be machine-translated from English into Velan and added to the wiki.

Source A: How the new articles were made

The 40,000 most-viewed English articles were translated by a neural machine translation model. Models like this learn from parallel texts: the same sentences in two languages. Few texts have ever been published in Velan, so the model was trained on about 2 million sentence pairs, compared with hundreds of millions for widely spoken languages. Each article carries the label machine translated until a volunteer has checked it.

Adapted from the Velan Wiki project page.

Source B: Articles on the Velan Wiki

Stacked bar chart of articles on the Velan Wiki. 2022: 3,000 written by volunteers. 2024: 3,600 written by volunteers, 2,100 machine translated and checked, 37,900 machine translated and not yet checked; 43,600 in total.

Velan Wiki statistics page. About 40 volunteers edit the wiki each month.

Source C: Volunteer editors' views

Mira: When young people search in Velan, they now find an answer instead of switching to English. That matters for keeping the language alive.

Joran: The translations use words no Velan speaker would use. Other websites copy them, and soon they will be in the training data for the next AI model too.

Ilse: The article on the city of Paris is now longer than anything we have about our own history. The encyclopedia sees the world through English eyes.

Velan Wiki discussion page, 2024.

Source D: The technology company's announcement

At the speed volunteers write, 40,000 articles would have taken more than a century. We did it in six months. On a standard automatic test, our Velan translations scored above 85 % of the quality of a professional human translation. Every person deserves access to the world's knowledge in their own language, and this grant brings Velan speakers closer to that goal. The model is free for the community to use.

Technology company press release, 2023.

  1. (a)

    With reference to Source A, outline why the amount of training data affects the quality of a machine translation model.

    2 marks

  2. (b)

    With reference to Source B, explain two impacts of the machine-translated articles on the Velan Wiki's volunteer editors.

    4 marks

  3. (c)

    Compare and contrast what Source C and Source D reveal about how machine translation affects the knowledge available in Velan.

    6 marks

  4. (d)

    With reference to Sources A–D and your own knowledge, to what extent does machine translation help speakers of smaller languages to create and share knowledge?

    12 marks

Question 3 · DS-00233Paper 3 · HL only · 30 marks

False claims in referendum campaigns

This pre-released statement sets out the real-world nature of a challenge, and of two interventions, from the digital society higher level extension. Candidates should use it to plan and carry out extended inquiries before paper 3.

  • Challenge topic: Governance and human rights

  • Area for inquiry: Participation and representation

As well as the information given here, candidates should consider relevant connections to other digital society concepts, content and contexts. Teachers should explore several scenarios in this context.

About the challenge

A referendum lets citizens decide a single question directly, such as a change to their country's constitution. Referendum campaigns now take place largely online. Alongside honest debate, false and misleading claims spread quickly: invented statistics, edited videos, and synthetic audio or video (deepfakes) of public figures. Some false claims are shared by people who believe them (misinformation); others are spread deliberately (disinformation), sometimes by groups that hide who is paying for them. Much of the sharing happens in end-to-end encrypted messaging groups, where no platform or authority can see the content, and through paid adverts shown only to selected audiences.

Article 25 of the International Covenant on Civil and Political Rights (1966) protects the right of citizens to take part in public affairs and to vote in genuine elections that guarantee the free expression of their will. Voters need accurate information to use this right well. Yet efforts to counter false claims raise their own concerns: who decides what is false, and could rules meant to protect voters restrict freedom of expression?

This statement introduces two interventions designed to reduce the influence of false claims on referendum campaigns. Their impacts and implications should be evaluated against the six criteria of the HL evaluation framework: equity, acceptability, cost, feasibility, innovation and ethics. In the examination, candidates will be given a source booklet with four sources about one scenario.

About the interventions

Intervention 1: A fact-checking tip line. Citizens forward suspicious messages to a chatbot on a messaging app. The chatbot compares each claim with a database of published fact-checks and replies at once if it finds a match. New claims are passed to human fact-checkers, who reply within 24 hours and publish weekly reports on the most widely shared rumors.

Intervention 2: A political advertising transparency rule. Every paid online political advert must carry a digital imprint stating who paid for it. Platforms must add each of these adverts to a public, searchable ad library showing how much was spent and which audiences were targeted. Campaigners who break the rule are fined.

About the stakeholders

  • voters

  • campaign groups on each side of a referendum

  • social media and messaging companies

  • electoral bodies, journalists and fact-checking organizations

Additional terminology

Ad library
Chatbot
Deepfake
Digital imprint
Disinformation
End-to-end encryption
Fact-checking
Filter bubble
Misinformation
Synthetic media
Targeted advertising

Notes

Candidates do not need to study how deepfakes are created.

Source booklet

Source A: Rellmark's constitutional referendum

Rellmark, a country of 9 million people, will hold a referendum in eight months on a new constitution that would reduce its number of regions from 12 to 5. The Electoral Commission of Rellmark, an independent public body, runs the vote and enforces the rules on campaign spending.

After the last referendum, in 2021, a commission survey found that 58 % of voters had seen a claim about the referendum that they later learned was false. The most widely shared one, that pensions would be cut if the proposal passed, began in a messaging group and was later used in paid adverts by a page that never revealed who funded it. A fake audio clip of a government minister also circulated in the final week.

Source B: The commission's two proposals

Fact-checking tip line

  • Run with two independent fact-checking organizations employing 12 fact-checkers.

  • Available in both of Rellmark's official languages; costs USD 350 000 for the campaign period.

  • The chatbot cannot see any message unless a user forwards it.

Advertising transparency rule

  • Requires a new law, which parliament would have to pass within four months.

  • Applies to adverts on every platform with more than 100 000 users in Rellmark.

  • Fines of up to USD 50 000 per advert.

Source C: An entry from a trial version of the ad library

Ad library entry for advert 30417. Advertiser: Rellmark Forward. Paid for by: not declared. Shown 2–9 March. Spent USD 10 000–15 000. Shown 400 000–500 000 times. Audience: age 65+, Northmoor and Tarvel regions. Text: Vote No: the new constitution will cut your pension.
Figure 1: Ad library entry

Source D: Views on the proposals

"Most of what I read about the referendum comes from my family's group chat. I wouldn't know how to forward something to a fact-checker."

Elena, 71, retired teacher

"Our group spends USD 300 on adverts. Filling in forms for every advert will hurt small groups like ours more than big campaigns with lawyers."

Marco, volunteer for a small campaign group

"In 2021 we received more claims in the final week than we could check in a month. A tip line will only work if we are not overwhelmed."

Spokesperson for a fact-checking organization

"Why should a commission paid for by the government decide what is true? That's censorship."

Comment on a news website

  1. (a)(i)

    Outline the difference between misinformation and disinformation.

    2 marks

  2. (a)(ii)

    Identify two characteristics of a deepfake.

    2 marks

  3. (b)(i)

    Explain two reasons why false claims shared in end-to-end encrypted messaging groups are difficult for the commission to counter.

    4 marks

  4. (b)(ii)

    With reference to Source C, explain one way in which the ad library entry could help voters judge the advert.

    2 marks

  5. (c)

    Discuss whether an electoral commission should decide which claims made during a referendum campaign are false.

    8 marks

  6. (d)

    The Electoral Commission of Rellmark has to choose between two interventions to address the challenge of false claims in the coming referendum campaign:

    • Intervention 1: A fact-checking tip line, with a chatbot on a messaging app and human fact-checkers.

    • Intervention 2: A political advertising transparency rule, with digital imprints and a public ad library.

    Recommend which of the interventions the Electoral Commission of Rellmark should choose.

    12 marks

Question 4 · DS-00014Paper 1 · HL only · 12 marks

Recommending across faiths

In many countries, tensions between religious communities are made worse online, where recommendation algorithms tend to show people more of what they already watch.

Vidora, the most popular video-sharing platform in the fictional country of Kesh, where three large religious communities live side by side, has changed its recommendation algorithm. The algorithm estimates which community each viewer belongs to from their viewing history. Alongside each user's usual recommendations, a "Bridges" row now shows videos liked by viewers from more than one community, such as cooking for religious festivals, music that mixes traditions or stories from shared neighborhoods. Videos that receive many angry reactions are left out of the Bridges row. Users cannot turn the row off.

It is claimed that the Bridges feature will increase tolerance between religious communities.

To what extent do you agree with this claim?

12 marks

Question 5 · DS-00281Paper 2 · 24 marks

Requests to remove

Background

Clipwave is a video-sharing platform used by 30 million people in the Republic of Karst. A law passed in 2024 requires platforms to act on government requests to remove content within 24 hours, or pay fines of up to 5 % of their income in Karst. Clipwave publishes a report on the requests it receives.

Source A: How Clipwave handles a government request

  1. The request arrives from a government ministry, naming the video and the law it breaks.

  2. Clipwave's legal team checks the video against Karst's law and against Clipwave's own community rules.

  3. If the video breaks Clipwave's own rules, it is removed everywhere.

  4. If it breaks only Karst's law, it is restricted in Karst only: viewers in Karst cannot see it, but viewers in other countries can.

  5. If it breaks neither, the request is refused.

  6. The uploader is told, unless the law forbids it.

Adapted from Clipwave's transparency report.

Source B: Government requests to Clipwave in 2024

Reason given by the government

Requests

Removed everywhere

Restricted in Karst only

Refused

National security

120

30

70

20

Criticism of public officials

340

10

150

180

False information about elections

210

40

120

50

Hate speech

90

72

10

8

Copyright

60

55

0

5

Clipwave transparency report, 2024.

Source C: Government spokesperson

Foreign groups use platforms like Clipwave to spread lies and turn citizens against each other, especially before elections. Clipwave makes money from the people of Karst, so it must follow the laws of Karst, just as any shop or bank must. The law does not stop criticism. Anyone who believes a video was removed unfairly can go to court.

Government press briefing, 2025.

Source D: Karst Journalists' Association

Videos filmed by our members of police at last year's protests were restricted within hours, labeled as "false information". No judge looked at them first. Clipwave admits that most of the requests it accepted named no specific false claim, but a fine of 5 % of its income is a risk no company will take often. Requests are kept secret, so citizens cannot see what is being hidden. Many reporters now avoid posting anything about officials at all.

Karst Journalists' Association, 2025.

  1. (a)

    With reference to Source A, outline how Clipwave could make a video unavailable in Karst only.

    2 marks

  2. (b)(i)

    With reference to Source B, suggest one reason why Clipwave restricted many videos in Karst only rather than removing them everywhere.

    2 marks

  3. (b)(ii)

    With reference to Source B, suggest one reason why Clipwave refused a much larger share of requests about criticism of public officials than requests about hate speech.

    2 marks

  4. (c)

    Compare and contrast what Source C and Source D reveal about who decides what can be published online in Karst.

    6 marks

  5. (d)

    With reference to Sources A–D and your own knowledge, to what extent should platforms such as Clipwave comply with government requests to remove content?

    12 marks

Question 6 · DS-00033Paper 1 · HL only · 12 marks

Labeling climate misinformation

False and misleading claims about climate change, such as that recent warming is entirely natural or that renewable energy cannot work, spread widely on video-sharing platforms.

Streamly, a global video-sharing platform, has introduced an intervention. A machine learning classifier, trained on videos already reviewed by professional fact-checkers, analyzes the speech, captions and descriptions of every new video that mentions the climate. Videos that the classifier scores as likely to contain misinformation are shown with a label: "Some claims in this video contradict the scientific evidence", followed by a link to a summary written by climate scientists. Channels that repeatedly post labeled videos can no longer earn money from advertising.

  1. (a)

    It is claimed that this intervention will increase public support for action on climate change.

    To what extent do you agree with this claim?

    12 marks

Practice all 34 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.

The notes on 2.2 Expression

More questions in Concepts

  1. 2.1Change
  2. 2.2ExpressionThis page
  3. 2.3Identity
  4. 2.4Power
  5. 2.5Space
  6. 2.6Systems
  7. 2.7Values and ethics

Check your answers in Augment

A free account opens all 34 questions on Expression with their markschemes, and keeps a record of what you have marked.