7AAVCD42 Artificial Intelligence and Society
7AAVCD42 Overview
- Postgraduate module
- Semester 1 2026/27
- Coursework assessed
- KEATS learning environment
Technology and society in one analytical frame
Artificial Intelligence and Society treats contemporary AI as both a set of technical mechanisms and a group of institutional choices.
- Trace the mechanism Separate inputs, representations, model operations and outputs before judging social consequences.
- Compare with discipline Use the same dimensions for both products, governance approaches or institutional cases.
- Bound every claim State what the evidence establishes, what remains inferred and what would change the conclusion.
What you get in the full 7AAVCD42 Study & Assessment Guide
- 10 chapters, 34 pages — every assessed topic, week by week
- 29 original practice questions, each with a model answer
- 10 worked examples, solved the way 7AAVCD42 marks
- 10 computed figures, redrawn from the course’s own numbers
- Before-you-submit checklist for the 1 x 2,100 word report (60%) — what markers reward, what costs marks
Free on this page: assessment table, chapter map, one worked example, key terms, FAQ. The rest is in the full study & assessment guide.
How 7AAVCD42 is assessed
| Component | Weight | Format |
|---|---|---|
| 1,400 reflective write up of your practical research project | 40% | Group presentation in seminar week 6 and individual reflection due 19 November 2026 before 2pm |
| 1 x 2,100 word report | 60% | Evaluate and compare two technical products or services, due 12 January 2027 before 2pm |
The module assignment summary publishes both weights and current deadlines; check KEATS for the latest full briefs and submission instructions.
Assessment structure
Both components are coursework; the captured assessment summary lists no final examination.
What other students are asking about 7AAVCD42
- Where can I find lecture notes or study guides for 7AAVCD42 at King's?
- The 7AAVCD42 study guide in AskSia Library covers Artificial Intelligence and Society in 10 chapters, with the assessment table, a chapter-by-chapter topic map, key terms and a worked example, free to read on this page.
- Is there a final exam in 7AAVCD42?
- No final exam is listed in the offering this AskSia Library guide is built from; 7AAVCD42 is assessed by 1,400 reflective write up of your practical research project 40%, 1 x 2,100 word report 60%.
- What does 7AAVCD42 cover on machine vision and model opacity?
- In AskSia Library's 7AAVCD42 guide, the Machine Vision and Model Opacity chapter explains trace image elements through learned features, layers, outputs and accountability.
- What does 7AAVCD42 cover on language models, prediction and hallucination?
- In AskSia Library's 7AAVCD42 guide, the Language Models, Prediction and Hallucination chapter explains compare next-word prediction with factual grounding, bias and responsible use.
- What does 7AAVCD42 cover on data lifecycles, post-training and safety?
- In AskSia Library's 7AAVCD42 guide, the Data Lifecycles, Post-Training and Safety chapter explains follow collection, labelling, handover and model steering into deployment.
Current 7AAVCD42 dates
| Date | Item | Control |
|---|---|---|
| 19 November 2026 | Individual reflection | Due before 2pm. |
| 12 January 2027 | Report | Due before 2pm. |
The module assignment summary publishes these dates; confirm current submission instructions on KEATS.
What 7AAVCD42 covers
Ten chapters connect technical mechanisms with evidence, institutions, social consequences and governance choices.
AI Histories and Social Frames
Use history to test present claims about intelligence, automation and progressWhy history belongs in AI analysis
The module begins with key events in AI history, the social dimension of intelligence and a first account of machine learning. This chapter turns chronology into an argument.
- Intelligence tests
- AI history
- Hype cycles
- Machine learning
- Social assumptions
Machine Vision and Model Opacity
Trace image elements through learned features, layers, outputs and accountabilityFrom image elements to institutional decisions
Machine vision begins with numerical image elements and builds learned features through network layers.
- Image Elements
- Learned Features
- Neural Networks
- Model Opacity
- Accountability
Language Models, Prediction and Hallucination
Compare next-word prediction with factual grounding, bias and responsible useFluency is one performance criterion
Transformer-based large language models calculate relations in context and generate plausible continuations. This chapter compares that prediction mechanism with factual grounding and human understanding.
- Transformer Architecture
- Next-Token Prediction
- Factual Grounding
- Hallucination
- Model Bias
Data Lifecycles, Post-Training and Safety
Follow collection, labelling, handover and model steering into deploymentFailures can begin before modelling
The module follows data through collection, cleaning, labelling, crowdsourcing and review, then examines human feedback, fine-tuning and safety rules after initial training.
- Data Collection
- Data Labelling
- Data Cascades
- Post-Training
- AI Safety
Comparing AI Governance Approaches
Compare rules, audits and oversight by force, actor, evidence and remedyGovernance mechanisms are not interchangeable
Ethical codes, company audits, technical standards, contracts, laws and international agreements bind different actors and offer different remedies.
- Regulatory Force
- Governed Actors
- AI Audits
- Public AI
- Accountability
Facial Recognition and Biometric Power
Separate face detection, faceprints, comparison and institutional decisionsA faceprint is a representation
The captured teaching sequence identifies face detection, feature extraction, faceprint creation and comparison as distinct stages.
- Face Detection
- Feature Extraction
- Faceprint Creation
- Biometric Comparison
- Surveillance Power
Automated Decisions in Social Services
Test benefit claims against privacy, trust, inequality and appeal rightsBenefits and risks share mechanisms
AI in education, healthcare, social care and public services is often justified through accessibility, personalisation and cost.
- Decision Support
- Personalisation
- Public Services
- Unequal Error
- Appeal Rights
Physical AI, Sensing and Responsibility
Reconstruct sense-model-act loops, accidents and distributed responsibilityThe street model is not the street
Physical AI senses an environment through cameras, radar, lidar or other devices, builds a working model and selects an action.
- Sensor Input
- Street Model
- Action Planning
- Accident Chain
- Distributed Responsibility
AI Agents, Evidence and Risk
Distinguish observed agent behaviour from wider claims and risk framesObserved behaviour needs a transfer warrant
AI agents can use tools, act over time and interact with other agents. This chapter separates behaviour observed in a designed environment from broader claims about deployment or future danger.
- AI Agents
- Tool Use
- Goal Drift
- Risk Frames
- Evidence Transfer
AI, Labour and the Future of Work
Analyse task change, surveillance, job quality and unequal distributionWork changes through tasks and control
AI can create coding and labelling work, assist existing tasks, monitor workers and displace activities. This chapter maps those effects without treating an occupation as a single automated block.
- Task Automation
- Worker Augmentation
- Workplace Surveillance
- Job Quality
- Distributional Effects
The module moves from AI history into machine learning, machine vision and language models, then follows data collection, labelling, post-training and safety. Later topics place governance, facial recognition, automated public decisions, physical AI, agents and work inside concrete social settings.
The most useful study habit is to keep the computational representation separate from the person, object or street it represents, then show how an organisation turns the model output into a consequential decision.
Two forms of coursework
The published assessment summary contains an individual reflection on a group project worth 40% and a comparative report worth 60%.
The reflection grows from group research and presentation work. It should use specific episodes to show how evidence, discussion or task decisions changed your understanding. The report evaluates and compares two technical products or services within an approved topic area. It needs a common evaluative question, matched dimensions and a conclusion limited to the users, setting and evidence examined.
The two numeric weights total 100%.
A method for every case
Begin by defining the decision or claim. Reconstruct the technical route from data or sensor input to representation, model operation and output. Identify the institution using the result, the people exposed to it and the route for challenge or correction. Compare an alternative explanation, second product or governance mechanism.
Finish by stating the condition that bounds the conclusion. This approach prevents technical description from floating free of social analysis and prevents ethical criticism from ignoring how the system actually works.
Evidence and current information
Product claims, policy summaries and model outputs require different checks.
Prefer current authoritative material for regulation and operational facts, independent research for contested effects and transparent documentation for technical features. Record dates when systems or rules may change. The module's example submissions can illustrate broad response moves, but they are not a rubric and should never be copied.
Full briefs and submission instructions remain on KEATS.
Use of generative AI
The module permits level 2 selective use to understand and learn technical terms and concepts, while requiring submitted words to be the student's own. Use any permitted tool to support learning rather than replace authorship. Verify factual claims, keep a source trail and follow the latest module-specific instructions.
The chapter practices in this guide are newly written study exercises with AskSia allocations, not official King's marking schemes.
Worked example · free
Compare two AI services with one evidence frame
- 2Define one common evaluative question.
- 3Apply matched evidence dimensions to both cases.
- 3Reach a bounded comparative judgement.
Key terms
- Machine Learning
- Machine Learning uses data-driven optimisation to build patterns or decision rules rather than relying only on explicitly programmed instructions.
- Neural Network
- Neural Network names a layered computational structure that transforms inputs into learned features and outputs.
- Model Opacity
- Model Opacity is difficulty in seeing or explaining the data, internal relations or institutional process behind an output.
- Transformer
- Transformer is the network architecture used by large language models to process relations across a context.
- Hallucination
- Hallucination is generated content presented fluently even though the claimed fact or source lacks reliable grounding.
- Data Cascade
- Data Cascade is the propagation of an unresolved collection, labelling or handover compromise into later system stages.
- Post-Training
- Post-Training further shapes a trained model through feedback, fine-tuning and behavioural rules.
- AI Governance
- AI Governance comprises rules, standards, audits and institutions used to direct and hold AI development or deployment accountable.
- Faceprint
- Faceprint is a mathematical representation extracted from an image for biometric comparison, not the person's face itself.
- Automated Decision
- Automated Decision is a consequential judgement made or materially shaped by a computational system.
- Physical AI
- Physical AI senses an environment, builds a working model and acts through a machine in the physical world.
- AI Agent
- AI Agent is a model-based system able to take actions, use tools and pursue tasks beyond a single conversational reply.
7AAVCD42 FAQ
How is this module assessed?
The captured Semester 1 assessment summary lists a 40% individual reflection connected to a group project and presentation, plus a 60% comparative report. The numeric weights total 100%. Full briefs and current submission instructions should be checked on KEATS before work is submitted.
What should the reflective assignment analyse?
Use specific moments from the group investigation to show how your assumptions, evidence choices or research practice changed. Connect each episode to a course concept and finish with a precise future action. A chronological diary of meetings does not by itself demonstrate reflective analysis.
How should the comparative report be structured?
Choose one evaluative question and apply the same technical, data, social and governance dimensions to both products or services. Distinguish company claims from independent evidence. The conclusion should specify which case is preferable for identified users, settings and conditions.
Do I need technical detail in an AI and society essay?
Yes, but only the detail needed to support the social claim. Explain the input, representation, model operation and output, then show how an institution uses the result. Technical vocabulary earns its place when it changes the evaluation of power, risk or accountability.
How can I compare AI governance approaches?
Compare force, governed actor, required evidence, oversight and remedy. An ethics code, company audit, contract and law are not interchangeable. Keep the dimensions constant across jurisdictions and verify current legal obligations against authoritative material rather than a model-generated summary.
What should I check immediately before submission?
Reopen the current KEATS brief, confirm the word limit and operational instructions, verify APA 7 references and links for the report, and ensure the submitted wording is your own. Read paragraph openings as an argument, then check each evidence sentence against its source.
Questions students ask in each 7AAVCD42 chapter
- Why study AI history in this module?
- History reveals how ideas about intelligence, automation and social value were formed. It helps test whether a current claim is technically new, institutionally different or simply repeating an older promise. Chapter 01 · AI Histories and Social Frames
- How does a model see an image?
- It receives numerical image elements and transforms them through learned features and layers. The output is a model representation and classification, not direct human-like perception of the original object. Chapter 02 · Machine Vision and Model Opacity
- Why can fluent text still be wrong?
- Language models optimise plausible continuation from learned patterns and context. Grammar and coherence can succeed even when a date, quotation, source or causal claim lacks grounding. Chapter 03 · Language Models, Prediction and Hallucination
- What is a data cascade?
- It is the propagation of an unresolved collection, labelling or handover problem into training, evaluation and deployment. The concept directs attention to the earliest repairable stage. Chapter 04 · Data Lifecycles, Post-Training and Safety
- How should governance mechanisms be compared?
- Use the same dimensions for each mechanism: force, actor, conduct, evidence, oversight and remedy. This reveals operational differences hidden by broad labels such as ethics or regulation. Chapter 05 · Comparing AI Governance Approaches
- What are the main facial-recognition stages?
- The course identifies face detection, feature extraction, faceprint creation and comparison. Each stage can introduce uncertainty before an institution decides what to do with the output. Chapter 06 · Facial Recognition and Biometric Power
- How is support different from substitution?
- Support provides evidence or recommendations that a responsible person can inspect and override. Substitution occurs when the system effectively makes the consequential decision or removes meaningful judgement. Chapter 07 · Automated Decisions in Social Services
- What is the sense-model-act loop?
- Sensors collect signals, software constructs a representation, planning selects an action and the action changes the next environment observed. Technical failure can enter at any stage. Chapter 08 · Physical AI, Sensing and Responsibility
- How are agents different from chatbots?
- Agents can act through tools, persist across tasks and interact with other systems or agents. The exact distinction depends on permissions, persistent memory and environment rather than the label alone. Chapter 09 · AI Agents, Evidence and Risk
- Why analyse tasks instead of whole jobs?
- Jobs contain different activities, relationships and responsibilities. A system may automate one task, create review work, change skill needs and increase monitoring without eliminating the occupation. Chapter 10 · AI, Labour and the Future of Work
How to prepare for the assessments
Study in paired passes. First reconstruct the system's data, representation, model operation and output without notes. Then add the institution, affected groups, evidence limit and governance route. For coursework, maintain a live evidence table with columns for claim, source type, case, comparison role and qualification.
Rehearse each chapter by answering its practice card, changing one condition and identifying the first sentence that must change. Keep a decision log for the group project so the reflection can analyse genuine developments rather than reconstructing them after the presentation. Before drafting the report, test whether every planned section compares both products through the same question.