Die Druckversion wird nicht mehr unterstützt und kann Darstellungsfehler aufweisen. Bitte aktualisiere deine Browser-Lesezeichen und verwende stattdessen die Standard-Druckfunktion des Browsers.

aiMOOC

ABI LF - AI, inequality and power

QR-Code

ABI LF - AI, inequality and power

Mini-description: Compare sources on AI access, labour effects, platform power and unequal consequences. Develop an evidence-based analysis that distinguishes claims, data, methods, mechanisms and caveats.

Media entry – Look, think, question: What does this image make visible about AI infrastructure? What remains invisible: electricity, cooling, chips, data, labour, ownership, language, access? Write one observation, one inference and one question. Do not turn an inference into a fact.

Einleitung

AI can widen opportunities and redistribute power at the same time. The key question is not simply whether AI is “good” or “bad”, but who can access it, who controls essential inputs, whose work changes, who gains, who bears costs and how strong the evidence is.

This aiMOOC is designed for the Leistungsfach Englisch, Kursstufe Baden-Württemberg. It is English first: try context, word formation, visual clues, an English explanation and an English dictionary before using German help. The learning goal is independent source-based reasoning, not the reproduction of ready-made essays.

Core inquiry: How do unequal access, labour exposure, platform power and infrastructure interact to produce unequal consequences?

I-can goals

I can …	Evidence of learning
compare claims from several sources	I identify publisher, date, population, method, evidence and limitation.
distinguish exposure, automation and augmentation	I avoid equating “AI exposure” with “job loss”.
analyse platform power	I explain mechanisms such as control of compute, switching costs, vertical integration and network effects.
connect inequality to mechanisms	I build a causal chain and mark where the evidence becomes uncertain.
integrate visual and written material	I use a graph, image or data point as evidence rather than decoration.
communicate at LF level	I qualify claims, compare sources and respond to counter-evidence in writing and discussion.
audit AI-supported claims	I verify factual claims against accessible sources and state uncertainty.

Vorwissen / Prior knowledge

90-second prompt	Your first hypothesis	What evidence would you need?
Who can use advanced AI tools?	access	connectivity, cost, language, skills, compute
Which tasks may change first?	labour	task-level exposure, sector, occupation, country
Who can shape the market?	power	market structure, inputs, partnerships, switching costs
Who may gain or lose?	distribution	wages, productivity, prices, participation, public services

Rule: A plausible hypothesis is not yet an evidence-based conclusion.

Help ladder / Differentiation

Step	Use this before asking for more help
1 Context	Reread the sentence before and after the difficult expression.
2 Word formation	Split the word: unequal → equal → equality → inequality.
3 Visualise	Draw a two-arrow mechanism: input → effect → consequence.
4 English support	Use an English learner’s dictionary or write a simple English definition.
5 German rescue	Only now use a short German equivalent, then return to English.

German rescue line: access gap = Zugangslücke; bargaining power = Verhandlungsmacht; lock-in = starke Bindung an einen Anbieter; caveat = Einschränkung/Vorbehalt; exposure = Betroffenheit durch mögliche KI-Einflüsse.

Differentiation: Support learners may analyse two sources with sentence frames. Core learners connect three sources. LF challenge learners connect four or more sources, compare methods, qualify causal claims and address counter-evidence.

Prüfungsbezug Baden-Württemberg

Status check: 03 October 2026. For the 2027 Abitur, the official English LF rules refer to the Bildungsplan 2016 and specify stronger emphasis on AFB II and III. The written examination includes listening and a longer writing section with unseen literary or non-literary material and may include discontinuous material such as visuals, statistics or diagrams. The communication examination is a separate oral component.

This aiMOOC is therefore an Abi-style skills module. It does not claim that “AI, inequality and power” is itself the official 2027 Schwerpunkt. Always check the Facherlass for your own examination year.

Official references:

Written examination practice	Communication examination practice
aspect-focused summary, analysis, comparison, comment or target-text production; precise use of evidence; integration of visuals	coherent spoken position, interaction, clarification, response to counterarguments, spontaneous source-based discussion
evidence must be traceable in the material	evidence must be explained concisely and made usable in interaction
AFB II/III: analyse, compare, evaluate, connect	AFB II/III: explain, weigh, respond, qualify, synthesise orally

Medien-Einstieg

Quick visual task: The diagram shows a model structure, not a social system. Add four boxes around it: connectivity, compute, context, competency. Which boxes are necessary before a person can benefit from the model?

Source warning: This Commons map uses older data and is useful as a historical visual baseline, not as current evidence. Compare it with the 2025 ITU figures below. Ask: What changed? What does the map hide? Which countries are grouped too broadly?

Source pack: four lenses

Use the source cards as a launch pad. Open the original source when you need a figure, definition, method or limitation. Do not cite this aiMOOC as the final evidence source.

Lens A: AI access

The digital divide is not only a question of being online. AI access can depend on broadband quality, affordability, devices, electricity, compute, language coverage, skills and institutional capacity.

Source A1 – ITU, Facts and Figures 2025: The ITU reports major gaps in internet use by income level, gender, age and urban/rural location. Its global headline data indicate that billions remain offline and that internet use is much higher in high-income than in low-income economies.

ITU Facts and Figures 2025

Source A2 – World Bank, Digital Progress and Trends Report 2025: The report frames AI readiness through four “Cs”: connectivity, compute, context and competency. It argues that advanced AI innovation, data-centre capacity and investment are heavily concentrated.

World Bank: Digital Progress and Trends Report 2025

Analysis prompt: Why can a discussion of “AI access” become misleading if it mentions software subscriptions but ignores chips, data centres, skills and local-language context?

Viewing – World Bank, “small AI”: Note one claim about why smaller or locally adapted AI systems may matter for developing economies. Then write one question the video does not answer.

Lens B: labour effects

Visual caution: A robot photograph or a delivery-rider photograph cannot by itself prove automation, job loss or algorithmic control. Use the image to generate questions; use labour evidence to answer them.

Source B1 – ILO, Generative AI and Jobs 2025: The ILO estimates that roughly one in four workers globally is in an occupation with some degree of generative-AI exposure. Its task-based approach emphasises that transformation of jobs is more likely than complete replacement in many cases.

ILO: Generative AI and Jobs – 2025 update

Source B2 – IMF research: IMF publications use a different concept of AI exposure and estimate a larger share of global employment as exposed, with higher exposure in advanced economies. Exposure can bring productivity gains or displacement pressure, so the same headline number cannot be read as a job-loss forecast.

IMF: AI and the global economy IMF Working Paper: The Global Impact of AI – Mind the Gap

Compare, do not collapse: If two organisations publish “25%” and “40%”, first compare definitions, task models, populations, dates and thresholds. Different measures are not automatically contradictory.

Listening/Viewing – ILO: Use an assigned excerpt. Record the speaker’s main claim, one piece of evidence and one caveat. Do not make a full transcript.

Lens C: platform power

Concept prompt: A network effect can make a service more useful as participation grows. It can also raise barriers to switching. Which additional mechanisms would you need to show before claiming market power?

Source C1 – OECD, Artificial Intelligence Markets, 2026: The OECD describes AI markets as dynamic while also identifying structural risks, including advantages linked to hardware, data, cloud infrastructure and vertical integration.

OECD: Artificial Intelligence Markets

Source C2 – UK Competition and Markets Authority, foundation models: The CMA has highlighted risks involving control of critical inputs, incumbent firms’ ability to shape routes to market and partnerships that may reinforce existing market power.

CMA: AI foundation models update

Mechanism vocabulary: gatekeeping, bundling, interoperability, switching costs, vertical integration, exclusive access, economies of scale, data feedback loops.

Viewing – OECD panel, 2026: Build a two-column note sheet: “evidence of competition” and “evidence of concentration”. Add a third column for what remains uncertain.

Lens D: unequal consequences

Visualisation task: Build a chain from digital demand to infrastructure, materials, energy, work and local effects. Mark each arrow as supported, plausible or unknown. The second image represents a broader material-infrastructure question; it is not proof that one specific mine exists “because of AI”.

Source D1 – UNCTAD, Technology and Innovation Report 2025: UN Trade and Development argues that AI can create development opportunities while capabilities, investment and market power remain unevenly distributed. It calls for stronger infrastructure, skills, data capacity and international cooperation.

UNCTAD: Technology and Innovation Report 2025

Source D2 – World Bank: The four-C framework helps separate several bottlenecks that are often hidden inside the phrase “AI divide”.

Viewing – UN Trade and Development: Identify one opportunity claim, one inequality claim and one proposed response. For each, ask what evidence would strengthen the claim.

Listening / Viewing

Choose two videos from the course and work in three passes.

Pass	Focus	Output
1 Gist	What is the central message?	one sentence
2 Evidence	Which data, examples or expert claims support it?	three notes with timestamps
3 Caveat	What is missing, uncertain or defined differently elsewhere?	one limitation and one follow-up question

LF challenge: Compare how an international organisation, a competition authority or an expert panel frames the same problem. Consider mandate and purpose without assuming that institutional status automatically guarantees neutrality.

Reading

Use a source passport for every report or article.

Field	Question
Publisher	Who produced the source and for what institutional purpose?
Date	Is the evidence current enough for the claim?
Population / scope	Which countries, workers, firms or technologies are included?
Definition	What exactly counts as access, exposure, adoption, concentration or harm?
Method	Survey, model, administrative data, expert judgement, case study or market analysis?
Evidence	Which figure or passage directly supports the claim?
Limitation	What cannot be concluded?
Comparison	Does another source use the same concept and denominator?

Reading rule: Never write “Source X proves …” when the source only estimates, models, suggests or reports an association.

Kontextwortschatz und Wortbildung

Word family	English explanation	Use in analysis
equal → equality → unequal → inequality	differences in resources, opportunities or outcomes	unequal access; income inequality
access → accessible → accessibility	the ability to reach, afford and effectively use something	access is broader than availability
expose → exposed → exposure	being in a position where tasks may be affected	exposure is not the same as displacement
automate → automation → automated	performing tasks with reduced direct human execution	automation can affect parts of a job
augment → augmentation → augmented	supporting or extending human capability	AI may augment rather than replace a task
compete → competition → competitive	firms or systems contest users, resources or markets	competition can coexist with concentration
concentrate → concentration → concentrated	resources or market activity cluster among few actors	concentrated compute capacity
distribute → distribution → distributive	how benefits or costs are spread	distributive consequences
depend → dependence → dependency	reliance on another actor or infrastructure	cloud dependency
regulate → regulation → regulatory	rules and institutions governing activity	regulatory response

Dictionary routine: infer from context → analyse word family → use the visual/mechanism → check an English dictionary → use German only as a final rescue.

Analyse-Werkzeugkasten

Use C-E-R-C: Claim – Evidence – Reasoning – Caveat.

Move	Sentence frame
Claim	The source suggests that …
Evidence	This is supported by …
Reasoning	This matters because the mechanism could be …
Comparison	In contrast, Source B measures …
Qualification	The figures are not directly comparable because …
Caveat	This does not establish that …
Synthesis	Taken together, the sources indicate … while leaving … uncertain.

Source comparison test: A stronger paragraph does not merely contain more citations. It explains why the sources agree, differ or cannot be directly compared.

Visual analysis

For photographs, diagrams and statistics, move from description → function → evidence value → limitation.

Ask	Example
What is shown?	visible infrastructure, worker, interface, trend or relationship
How is it framed?	scale, angle, labels, colour, omitted context
What can it support?	a descriptive point or analytical question
What can it not support?	a causal claim without additional evidence

Fictional data lab

All values below are fictional classroom data. They must never be cited as real-world evidence.

Fictional region	Effective AI-tool access	Advanced digital skills	Local-language support index	Jobs with some AI-task exposure
Aster	82%	68%	76/100	38%
Birch	61%	41%	54/100	29%
Cedar	29%	18%	31/100	21%
Delta	14%	7%	22/100	15%

Task: Create one graph. Then write one descriptive claim and one cautious explanatory hypothesis. State why the table cannot establish causality.

LF extension: Add one fictional policy intervention and predict two possible effects plus one unintended consequence. Clearly label all additions as hypothetical.

Speaking

Evidence hearing: In groups, assign roles such as labour researcher, platform analyst, development economist, worker representative and consumer advocate. Each speaker must use one traceable source and one caveat.

Communication-practice format, not an official timing prescription: Give a short source-based opening statement, then discuss a contested claim with a partner. Respond to evidence rather than repeating prepared speeches.

Useful moves:

Refer to evidence: “The ILO estimate is useful here because …”
Qualify: “That conclusion depends on how exposure is defined.”
Build on a point: “I agree with the mechanism, but the distributional effect may differ because …”
Respond: “A counterpoint is …; however, Source C suggests …”

Privacy: Audio or video recording is optional and should remain private unless you explicitly consent to another use. No upload is required.

Writing

Choose one route. Your teacher may adjust length and timing.

Route	Task	Main focus
AFB I	Summarise the relevant findings of one source for the question “Who can benefit from AI?”	selection and precision
AFB II	Compare the labour-effects claims of the ILO and IMF and explain why their headline percentages are not directly interchangeable.	method-aware comparison
AFB II/III	Analyse how platform power may amplify unequal access. Integrate at least three different source types.	mechanism and synthesis
AFB III	Write a comment, article or speech manuscript evaluating the claim: “Equal access to AI tools is enough to make AI socially inclusive.”	judgement based on evidence and counter-evidence

No replacement essays: Use feedback, source checking and language tools to improve your own draft. Do not submit a generated full essay as your work.

AI claim protocol: If AI support is permitted by your teacher, record what you used it for, verify factual claims against accessible sources, check dates and definitions, and rewrite the reasoning yourself. This aiMOOC does not require any external AI service.

Mediation

Fictional school note – German source: „Eine Schule möchte KI-Werkzeuge erproben. Nicht alle Lernenden verfügen zu Hause über gleich schnelle Geräte oder kostenpflichtige Zugänge. Deshalb sollen Aufgaben auch ohne private KI-Nutzung bearbeitbar bleiben. Die Schule plant Schulgeräte, gemeinsame Übungszeiten und eine kurze Einführung in Quellenprüfung. Persönliche Texte sollen nicht ungefragt an externe Dienste übertragen werden.“

Task: Write a concise English message for an international partner school explaining the relevant policy idea. Do not translate sentence by sentence. Select what the target audience needs, preserve the distinction between access and effective use, and avoid adding facts.

Hilfen für anspruchsvolle Analyse

If you get stuck …	Try …
Your paragraph is descriptive	add “This matters because …” and name a mechanism.
You have many numbers	choose the figure that directly answers your claim.
Sources disagree	compare definitions, denominator, method, date and population.
Your argument is too certain	use may, can, is associated with, suggests, depends on.
Your vocabulary is repetitive	build a word family before searching for a synonym.
You cannot connect sources	write one bridge sentence: “Source B complicates Source A because …”

Feedback und Fehleranalyse

Use feedback in this order: argument → evidence → source handling → structure → language.

Code	Meaning	Revision move
C	claim too broad	narrow the claim
E	evidence missing or weak	add precise support
M	mechanism missing	explain how or why
S	source problem	check date, method, scope or wording
Q	qualification missing	add a caveat
L	language obscures meaning	simplify syntax and repair accuracy
R	register / target format	adapt tone and genre

Error log: Keep three columns: recurring problem, corrected example, rule for next time. Do not only collect corrections; formulate the rule.

Transfer

Use the framework beyond AI. Test it on another technology or platform: access → inputs → market structure → labour effects → distribution → public response. Ask which links are evidenced and which are assumptions.

For school-level transfer, use only anonymous or fictional data unless a teacher-approved method exists. Do not collect sensitive personal data, do not upload classmates’ work without permission and do not send texts to external AI services unless this is explicitly authorised.

Abi-Check

Written examination – can you …

Check	Yes / not yet
select information for an aspect-focused summary?
analyse argument, structure, language and communicative strategies in a non-literary text?
integrate a visual, statistic or diagram accurately?
compare sources with different methods without forcing agreement?
move from evidence to an AFB II/III judgement with caveats?
produce the requested target format rather than a generic essay?

Communication examination – can you …

Check	Yes / not yet
present a concise source-based position?
react to a partner rather than deliver a memorised monologue?
clarify, qualify and challenge claims politely?
use evidence orally without drowning the discussion in detail?
reformulate when vocabulary is missing?

Operator reminder: Use the current official operator list for your examination year. A strong response follows the operator, not merely the topic.

Folgekurs

ABI LF - AI, democracy and public discourse: information power, synthetic media and public debate
ABI LF - AI, work and the future economy: productivity, occupations and distribution
ABI LF - AI, migration and borders: automated decision-making, mobility and rights
ABI LF - Digital identities and surveillance: data, platforms and autonomy

Interaktive Aufgaben

Quiz: Teste Dein Wissen

What is the safest interpretation of a large international “AI exposure” estimate? (It indicates that tasks in many jobs may be affected, not that the same share of jobs will disappear.) (!It proves that the same share of workers will lose their jobs.) (!It measures only the number of people already using paid AI tools.) (!It is identical across all studies because exposure has one universal definition.)

Why should an ITU internet-use figure and a World Bank AI-readiness figure not be treated as identical? (They may measure different concepts, populations, dates and dimensions of access.) (!Every international organisation uses the same indicators.) (!AI readiness is only another name for internet speed.) (!Differences between figures always mean one source is false.)

Which statement best distinguishes automation from augmentation? (Automation shifts task execution to a system, while augmentation supports or extends human performance.) (!Automation and augmentation both mean complete job loss.) (!Augmentation refers only to faster internet access.) (!Automation is a type of market concentration.)

Which mechanism is directly relevant to platform power? (Control of critical inputs combined with high switching costs can strengthen gatekeeping power.) (!A colourful user interface automatically creates monopoly power.) (!A large number of users proves that prices are unfair.) (!Every network effect eliminates competition.)

What makes a source comparison analytical rather than descriptive? (It explains how differences in definition, method or scope affect the conclusions.) (!It lists the titles of several reports.) (!It repeats every number from each source.) (!It assumes the newest source must always be correct.)

How should the older Commons internet-penetration map be used in this course? (As a historical visual baseline that must be checked against newer evidence.) (!As definitive evidence for internet access in 2026.) (!As proof of the causes of digital inequality.) (!As a replacement for current data.)

Which feature is relevant to current English LF written-exam preparation in Baden-Württemberg? (Analysing unseen material and integrating evidence, potentially including discontinuous material.) (!Memorising a single model essay for every topic.) (!Using only literary texts and never visuals.) (!Avoiding AFB III tasks.)

Why are AFB II and III especially important in the Leistungsfach? (The LF requires substantial analysis, comparison, evaluation and independent material linkage.) (!The LF assesses vocabulary lists only.) (!The LF avoids source comparison.) (!The LF replaces all writing with multiple-choice questions.)

What is a central skill for the communication examination? (Responding to another speaker while developing a coherent evidence-based position.) (!Reading a prepared essay without interaction.) (!Quoting long passages from memory.) (!Using as many statistics as possible without explanation.)

What should you do with a factual claim suggested by an AI tool? (Verify it against an accessible source and check date, definition, scope and evidence.) (!Use it if it sounds plausible.) (!Treat it as verified because the wording is confident.) (!Replace the source citation with the AI tool name.)

Memory

Exposure	Tasks may be affected by AI
Augmentation	Technology supports human performance
Automation	A system executes parts of a task
Gatekeeping	Control over access to an important route or input
Lock-in	Switching provider becomes costly or difficult
Compute	Processing capacity used to train or run models
Caveat	A limitation that qualifies a claim
Interoperability	Systems can work or exchange information across boundaries

Drag and Drop

Ordne die richtigen Begriffe zu.	Thema
Connectivity	Reliable digital connection and network access
Compute	Processing infrastructure for training or running AI
Context	Relevant data, language and local applicability
Competency	Skills needed to use and evaluate AI effectively
Concentration	Resources or market activity clustered among relatively few actors

Kreuzworträtsel

Compute	What word names the processing capacity needed to train or run AI systems?
Exposure	What term describes the extent to which tasks may be affected by AI?
Gatekeeping	What term describes control over access to an important input or route to users?
Augmentation	What term describes technology extending human performance rather than fully replacing it?
Infrastructure	What term covers the underlying physical and digital systems that make AI services possible?
Caveat	What word names a limitation that qualifies an evidence-based claim?

LearningApps

Lückentext

A study of AI inequality begins by defining what counts as

. A worker can be affected by AI even if an entire job is not

. When technology supports a worker’s performance, this can be described as

. A headline percentage needs a clearly stated population and

. Control over essential inputs can create forms of platform

. High switching costs may contribute to provider

. A careful analyst adds a

when the evidence does not support a strong causal conclusion. In the Leistungsfach, sources should be connected through independent

rather than merely listed.

Offene Aufgaben

Leicht

Source passport: Choose one course source and complete publisher, date, scope, definition, method, evidence and limitation in no more than 120 words.
Visual question map: Use one Commons image from the course and create five analytical questions that the image raises but cannot answer by itself.
Word family map: Build a one-page visual around equal, access, expose and concentrate, including noun, verb or adjective forms where possible and one analytical sentence for each family.
Claim audit: Take three claims from your notes and label each as supported, overstated or not yet verifiable. Repair the wording without adding new evidence.

Standard

Listening evidence log: Use one embedded video excerpt and record gist, evidence, one timestamped example and one caveat. Compare your notes with a partner without creating a full transcript.
Labour comparison: Compare the ILO and IMF approaches to AI exposure. Explain at least two reasons why headline percentages should not be treated as directly interchangeable.
Platform mechanism diagram: Draw a mechanism from compute or cloud access to switching costs, gatekeeping and possible unequal outcomes. Mark every arrow as evidenced, plausible or uncertain.
Private interview option: Ask a consenting adult how AI has changed one task at work or study. Notes are enough; recording is optional and private. Compare the account with one labour source and state why one interview cannot represent a population.

Schwer

Evidence synthesis: Develop a 400–600 word analysis connecting access, labour and platform power with at least four sources. Include one methodological disagreement and one counterargument.
Fictional data visualisation: Visualise the fictional data table, formulate two hypotheses and design a real-data research plan that could test them without collecting sensitive personal data.
Source challenge: Find a recent credible source that complicates one course claim. Explain whether it changes the conclusion, the mechanism or only the degree of certainty.
Communication simulation: Conduct a source-based discussion on the claim “Equal access to AI tools is enough to make AI inclusive.” Use evidence from at least three sources, respond to a partner and finish with a qualified synthesis.

Lernkontrolle

Causal chain: Explain how unequal connectivity could interact with skills and compute to influence who benefits from AI. Identify one point where the chain may fail.
Method transfer: Two reports give different percentages for AI labour exposure. Develop a procedure for deciding whether they disagree substantively or merely measure different things.
Platform power case: Imagine that one provider controls a popular model, cloud infrastructure and a major distribution channel. Analyse two possible efficiencies and two possible competition risks without assuming the final outcome.
Unequal consequences: Use one access source, one labour source and one market-power source to explain why the same AI innovation could create different effects in two countries or social groups.
Visual evidence: Evaluate the evidential value of one infrastructure image in the course. State one claim it can support and two claims it cannot establish.
Abi transfer: Turn the course question into an AFB III task and write a thesis plus a counter-thesis. For each, identify the evidence you would need before reaching a judgement.

Lernnachweis

For a successful Lernnachweis, you should be able to show:

Source literacy: identify purpose, date, scope, method and limitation of a source.
Evidence use: support claims with precise, traceable material.
Method awareness: explain why apparently conflicting figures may not be directly comparable.
Analysis: connect access, labour effects, platform power and unequal consequences through explicit mechanisms.
Language: use accurate analytical vocabulary and suitable qualification.
Writing: produce an independent, well-structured source-based response in the requested target format.
Speaking: develop and defend a position while responding to another speaker.
Integrity: distinguish your reasoning from external assistance and verify AI-supported factual claims.
Privacy: use fictional or anonymised classroom data where appropriate and avoid unrequested uploads or external AI access.

OERs zum Thema

Additional open or public-interest source starting points:

Verknüpfte Lernbereiche

ABI LF - AI, inequality and power

aiMOOC-Projekte

Frag mich zum aktuellen Lerninhalt oder wähle eine kurze Hilfe.

Textmodus bereit

Sprache