2024 IB DP Digital society Practice Paper with Answers

To manage digital well-being, users can utilize specific app-based interventions:
1. App timers/limits: Restricts access once a limit is reached, promoting self-regulation.
2. Grayscale/Bedtime modes: Reduces the visual appeal of the screen and silences alerts to prevent late-night usage.

Award [1] mark for each valid digital well-being feature identified, up to [2] marks max.
- Acceptable features include: app timers, screen time dashboards, grayscale mode, notification muting, focus mode.
- Do not accept generic answers like 'turning off the phone'.

Electronic Health Records (EHRs) digitalize patient medical histories, improving clinical workflow and safety.
1. Accessibility: Different specialists within a hospital can immediately view past treatments and test results.
2. Error reduction: Software-based alerts warn clinicians if a newly prescribed medication conflicts with existing prescriptions or documented allergies.

Award [1] mark for each valid clinical, administrative, or safety benefit of EHRs identified, up to [2] marks max.
- Accept: instant retrieval of data, reduced administrative/paperwork errors, automated alerts, improved coordination of care.
- Do not accept extremely vague answers like 'makes doctors work faster'.

The transition from IPv4 to IPv6 was primarily driven by the exhaustion of available IP addresses. The two primary technical distinctions lie in their length and format:
- IPv4 addresses consist of four numbers separated by dots (e.g., 192.168.1.1) and are 32 bits long.
- IPv6 addresses consist of eight groups of hexadecimal digits separated by colons (e.g., 2001:0db8:85a3:0000:0000:8a2e:0370:7334) and are 128 bits long.

Award [1] mark for each valid technical difference identified, up to [2] marks max.
- [1] mark for address length (32-bit vs 128-bit).
- [1] mark for notation format (dotted-decimal vs hexadecimal colons).
- Accept other valid technical differences, such as the support for built-in security features (IPsec) or differences in packet header size.

AI bias generally arises from two primary sources:
1. Data bias: If the historical data used to train the machine learning model reflects past human prejudices, the AI will learn and perpetuate those same biases.
2. Algorithmic design bias: Developers may choose to prioritize variables or design weightings in a way that disproportionately favors or disadvantages specific groups.

Award [1] mark for each valid factor of bias identified, up to [2] marks max.
- Accept: skewed/unrepresentative training data, historical human prejudice embedded in data, designer bias, poorly defined feedback loops.
- Do not accept 'the computer is evil' or other non-technical explanations.

GIS technologies visualize geographic relationships by:
1. Data Layering: Superimposing different layers of environmental data over a geographical map to assess correlations (e.g., overlaying zoning laws with flood zones).
2. Vector Mapping: Representing features as points (e.g., hospital locations), lines (e.g., rivers), or polygons (e.g., municipal boundaries).

Award [1] mark for each valid representation method identified, up to [2] marks max.
- Accept: use of layers/layering, vector data (points/lines/polygons), raster data (grid cells/satellite imagery), 3D spatial modeling.
- Do not accept general terms like 'taking pictures'.

Metadata is data about data. In digital text or PDF files, standard metadata fields include details about the document's origin, identity, and version history, such as the author's identifier, creation timestamp, file size, and the application used to create the file.

Award [1] mark for each valid type of document metadata identified, up to [2] marks max.
- Accept: author/creator identity, creation date, modification date, file size, file format, word count, access permissions.
- Do not accept elements of the actual document content (e.g., the text paragraphs or embedded images).

A VPN protects data through several key mechanisms: 1. Encryption: Data packets transmitted from the employee's device are encrypted before leaving the device. Even if an attacker on the same public Wi-Fi intercepts the packets (packet sniffing), the data is unreadable without the decryption key. 2. Tunneling: The VPN encapsulates the encrypted data packets within another packet protocol, creating a secure virtual tunnel between the user's device and the company's VPN gateway. This masks the internal protocol structure. 3. Masking IP and Location: The user's external IP address is replaced by the VPN server's IP address, obscuring the actual location and origin of the traffic. 4. Authentication: Secure protocols (such as IPsec or OpenVPN) require robust user authentication before granting access to the corporate network, preventing unauthorized actors from spoofing the connection.

Award [1 mark] for recognizing that public Wi-Fi is vulnerable to eavesdropping or packet sniffing. Award [1 to 2 marks] for explaining how encryption renders intercepted data unreadable to unauthorized parties. Award [1 to 2 marks] for explaining how tunneling protocols secure data in transit by encapsulating packets. Award [1 mark] for explaining how authentication checks protect the corporate network from unauthorized devices.

Persuasive design negatively impacts cognitive well-being in the following ways: 1. Infinite scroll removes natural stopping points (cognitive barriers), leading to passive consumption, decreased self-regulation, and decision fatigue, which diminishes overall mindfulness and mental focus. 2. Push notifications act as external triggers that constantly interrupt the user's current task. This leads to task-switching costs, higher cognitive load, reduced deep attention capacity, and heightened levels of stress or anxiety due to the fear of missing out (FOMO). To mitigate these effects, a digital intervention such as setting daily screen-time limits or configuring 'do not disturb' schedules can be used. These interventions re-introduce artificial constraints and conscious decision points, allowing the user to restore cognitive control, reduce interruptions, and protect dedicated periods of focus.

Award [1 to 2 marks] for explaining how infinite scroll disrupts self-regulation and cognitive attention. Award [1 to 2 marks] for explaining how push notifications interrupt focus, increase cognitive load, or induce anxiety. Award [1 to 2 marks] for explaining how a digital intervention (e.g., app timers, focus modes) establishes structural boundaries to restore cognitive well-being. Maximum 6 marks total.

Self-imposed 'digital detoxes' (temporarily abstaining from digital device usage) are often proposed as solutions to the negative mental health impacts of social media. On one hand, they restore individual agency, interrupt compulsive feedback loops (dopamine-driven cycles from infinite scroll), and improve sleep quality and real-life social interactions. On the other hand, their effectiveness is limited because they place the entire burden of change on the user, ignoring the systemic power imbalance. Persuasive design features are specifically engineered to exploit human cognitive vulnerabilities (FOMO, variable ratio rewards). For young people, a prolonged digital detox can lead to social exclusion, anxiety over missed connections, and eventual relapse into intensive use. Therefore, while digital detoxes provide short-term relief, they cannot fully mitigate the impacts of predatory algorithmic design. Long-term well-being requires systemic interventions, including 'design-by-default' regulations, legislative bans on certain persuasive features for minors, and ethical design frameworks implemented by technology platforms.

Level descriptors: [1-2 marks]: Minimal focus. Mentions digital detox or social media impacts superficially. [3-4 marks]: Explains the concept of digital detox and identifies some persuasive design features like infinite scroll, but lacks deep critical analysis. [5-6 marks]: Provides a balanced discussion analyzing both the positive aspects of individual agency (detoxes) and the powerful structural barriers of persuasive design. [7-8 marks]: Fully developed, critical evaluation that clearly contrasts individual responsibility with platform/systemic accountability. Demonstrates excellent understanding of digital well-being concepts and provides a clear, logical conclusion supported by evidence.

AI diagnostic systems offer immense promise in rural areas of developing economies where medical specialists are scarce. However, their deployment faces severe challenges. Ethically, AI models trained primarily on Western datasets may suffer from demographic bias, leading to inaccurate diagnoses for local populations. Issues of informed consent also arise when patients lack digital literacy to understand how their data is used. Practically, the digital divide poses a massive hurdle: unstable power grids, lack of high-speed internet, and a shortage of trained local staff to operate or maintain these systems. Furthermore, there are legal ambiguities regarding liability when an AI system misdiagnoses a patient in a setting with minimal regulatory oversight. Thus, while AI diagnostics can democratize health access, their success depends on addressing systemic infrastructural deficits and ensuring algorithmic models are culturally and biologically representative of the populations they serve.

Level descriptors: [1-2 marks]: Outlines a few general points about AI in healthcare without context. [3-4 marks]: Explains specific ethical or practical challenges but lacks balanced coverage of both. [5-6 marks]: Offers a balanced evaluation of both ethical challenges (bias, consent) and practical realities (infrastructure, literacy) within the specific context of rural developing economies. [7-8 marks]: Outstanding, highly structured evaluation. Demonstrates deep understanding of the intersection of technology, equity, and healthcare. Clearly argues the tensions between automated efficiency and systemic vulnerability with a well-justified conclusion.

### Model Response Outline

**Introduction**
- **Definitions:** Define algorithmic management (the use of algorithms, data tracking, and automated decision-making to coordinate workforce activities) and the gig economy.
- **Significance:** Link this to the HL extension topic of global well-being, specifically focusing on psychological well-being (stress, agency, anxiety) and social well-being (isolation, community support).
- **Thesis Statement:** Although algorithmic management offers workers schedule flexibility and direct access to labor markets, its reliance on pervasive surveillance, opaque decision-making, and isolation significantly undermines worker well-being globally.

**Arguments for Negative Impacts (To a Great Extent)**
- **Constant Surveillance and Stress:** Algorithms track real-time locations, acceptance rates, and speed. This relentless data-gathering creates high-stress environments, leading to 'technostress' and anxiety.
- **Insecurity and Fear of Deactivation:** The threat of sudden, automated 'deactivation' (termination) by an algorithm without human recourse creates persistent job insecurity and psychological distress.
- **Social Isolation:** Gig workers operate in physical isolation, lacking a physical workplace, colleagues, and traditional support networks. This erosion of social capital leads to feelings of alienation.
- **Asymmetry of Power:** The platform holds all data and controls the parameters of work, leaving workers with little agency or bargaining power (concepts of Power and Control).

**Arguments for Positive/Mitigating Impacts (To a Lesser Extent)**
- **Autonomy and Flexibility:** Algorithms allow workers to choose their own hours, potentially supporting work-life balance and individual well-being.
- **Reduction of Human Bias:** Automated dispatching and rating systems can theoretically reduce direct face-to-face discrimination or harassment from human supervisors.
- **Platform-Led Interventions:** Some modern platforms are introducing in-app well-being resources, peer forums, or clearer earnings calculators to mitigate stressors.

**Conclusion and Synthesis**
- **Weighted Judgment:** Conclude that while the technological flexibility is beneficial, the current paradigm of algorithmic management prioritizes corporate efficiency over human dignity. Thus, to a highly significant extent, these systems degrade psychological and social well-being unless governed by human-in-the-loop interventions and strong labor regulations.

### Marking Rubric (12 Marks Total)

**[10–12 Marks]**
- **Focus & Analysis:** Excellent, balanced evaluation of "to what extent" the systems impact well-being. Detailed focus on both psychological and social dimensions.
- **Digital Society Concepts:** Confident and appropriate integration of key concepts (e.g., Power, Identity, Change, Well-being) and digital systems terminology (e.g., data analytics, information asymmetry, automated decision-making).
- **Structure & Evidence:** Highly structured response with clear transitions. Incorporates real-world context (e.g., ride-hailing or delivery apps) to back up claims. Concludes with a nuanced, well-supported synthesis.

**[7–9 Marks]**
- **Focus & Analysis:** Good, balanced discussion of the positive and negative aspects of algorithmic management on well-being.
- **Digital Society Concepts:** Clear understanding of how the digital system operates and its general socio-psychological consequences. Relevant terms are used correctly.
- **Structure & Evidence:** Well-structured argument, although the evaluation or conclusion may lack some depth or fail to fully address the global perspective.

**[4–6 Marks]**
- **Focus & Analysis:** Descriptively explains how gig apps work, with some link to stress or isolation. The analysis is largely one-sided or superficial.
- **Digital Society Concepts:** Limited use of precise terminology; relies on general descriptions of computers or apps rather than critical analysis of algorithmic management.
- **Structure & Evidence:** Lacks clear structure or a cohesive, balanced argument.

**[1–3 Marks]**
- **Focus & Analysis:** Minimal understanding of the prompt. Offers a generalized opinion on smartphones, apps, or gig work with no analytical depth.
- **Digital Society Concepts:** Minimal or incorrect use of terminology.
- **Structure & Evidence:** Fragmented or unstructured response.

Based on the deployment of digital health technologies in remote regions:

1. Overcoming geographical barriers: Telehealth platforms allow patients in rural areas to connect with specialist doctors in urban centers via video conferencing, reducing travel costs and time.
2. Remote monitoring: Wearable or home-based digital health devices track vital signs and automatically transmit data to healthcare providers, facilitating proactive management without requiring frequent physical appointments.

Award [1] for each valid way identified, up to [2].

Suitable responses may include:
- Eliminates the need to travel long distances for specialist consultations.
- Allows for real-time remote monitoring of chronic conditions.
- Decreases costs associated with seeking medical care (e.g., travel, accommodation).
- Enhances the speed of receiving medical advice or preliminary triage.
- Provides access to healthcare services for individuals with limited physical mobility.

Note: Reject vague answers such as "makes healthcare better" or "cures diseases faster" without a clear link to accessibility.

Challenge 1: Lack of reliable high-speed internet infrastructure in remote rural areas. Wearable cardiac sensors require steady network connectivity to transmit health metrics to the clinic. Poor connectivity can cause data packets to be lost, resulting in delayed interventions for critical health events. Challenge 2: Digital literacy barriers among rural populations, which often trend older. These patients may struggle to set up the wearable sensors, sync them with smartphone applications, or troubleshoot technical issues, leading to incorrect usage or complete abandonment of the system.

For each of the two suggested challenges: Award [1] for identifying a valid digital equity challenge and [1] for explaining/developing it in the context of the scenario. Maximum [2] marks per challenge. Max total [4] marks. Valid points include: Lack of telecommunications infrastructure (e.g., lack of 4G/5G or fiber broadband); Cost barriers (unaffordability of required secondary devices or data plans); Digital literacy limitations (inability to configure or troubleshoot the device); Language or accessibility barriers in app design.

To gain full marks, the response should clearly outline at least two similarities and two differences, or provide a deeply analytical synthesis of three of each. Comparing (Similarities): 1. Accessibility and Anonymity: Both digital interventions reduce the barrier to entry for teenagers seeking help by offering a degree of anonymity/pseudonymity, which helps destigmatize mental health support compared to face-to-face sessions. 2. Data Privacy and Security Risks: Both systems handle highly sensitive student data (conversational inputs, emotional states). Both require robust encryption and strict compliance with data protection laws (e.g., GDPR) to prevent unauthorized access. Contrasting (Differences): 1. Human Empathy vs. Algorithmic Automation: MindCare uses automated NLP algorithms, meaning interactions are standardized and lack genuine emotional empathy, though they are highly consistent. BuddyConnect relies on genuine human-to-human interaction, which fosters a stronger sense of belonging and community, although quality of support can vary. 2. Scalability and Availability: MindCare is highly scalable and available 24/7 without marginal costs, as the chatbot can handle unlimited concurrent users. BuddyConnect's availability is restricted by the schedules and capacity of human peer mentors and professional school moderators. 3. Risk Management: In a crisis (e.g., self-harm), MindCare is limited by its programming and must rely on automated redirection triggers. BuddyConnect, being human-moderated, can facilitate immediate escalation to crisis teams, though it carries a risk of negative peer influence or misinformation spreading before moderators can intervene.

[1-2 marks]: The response is mainly descriptive, identifying basic features of one or both systems with little explicit comparison or contrast. [3-4 marks]: The response compares and contrasts the two interventions, but the analysis may be unbalanced (e.g., mostly differences, or lacking technical/social depth). Some digital society concepts are integrated. [5-6 marks]: Balanced and detailed comparison and contrast of both interventions. Explicitly addresses digital well-being, privacy, scalability, and the nature of human versus automated interaction with clear, structured points.

### Exemplar Response

**Introduction**
The implementation of the AI-driven 'WellMind' platform by Genovia’s Ministry of Education represents a modern tension in digital society: leveraging advanced digital interventions to enhance global well-being versus protecting fundamental human rights like privacy and autonomy. While the platform aims to proactively address youth mental health crises, its reliance on continuous surveillance, algorithmic classification, and non-consensual data mining presents significant ethical challenges. This essay evaluates the implementation of 'WellMind' by balancing its potential health benefits against its societal costs.

**Arguments in Favor of 'WellMind' (Digital Interventions for Well-Being)**
From an interventionist perspective, 'WellMind' addresses a critical challenge in global well-being: the escalating youth mental health crisis. Traditional school counseling frameworks are often reactive, relying on students to self-report or teachers to notice behavioral changes, which frequently happens too late.
* **Proactive Support:** By analyzing real-time data such as search queries and attendance, 'WellMind' can detect early warning signs of depression or self-harm that humans might miss.
* **Equity in Support:** It standardizes mental health monitoring across all socioeconomic brackets, ensuring that marginalized students who may lack access to external healthcare receive attention.
* **Efficiency:** Algorithmic triaging allows scarce school counseling resources to be directed to where they are most urgently needed.

**Arguments Against 'WellMind' (Challenges and Negative Consequences)**
Conversely, the pervasive nature of 'WellMind' presents severe ethical and societal risks that may actually undermine student well-being.
* **Surveillance and the Chilling Effect:** Constant monitoring of student emails and searches creates a panoptic environment. Students, aware they are being watched, may alter their behavior, self-censor, or refrain from searching for help on sensitive topics (such as LGBTQ+ identity, reproductive health, or abuse), thereby exacerbating their isolation.
* **Algorithmic Bias and Errors:** AI models trained on historical data often exhibit bias. 'WellMind' may disproportionately flag students from diverse cultural backgrounds due to linguistic variations in expressing distress (false positives), or fail to flag quiet, high-achieving students who mask their depression (false negatives).
* **Consent and Power Asymmetry:** As minors in a public school system, students have no meaningful way to opt-out. This represents a stark power asymmetry between the state (the Ministry) and the individual, compromising the ethical principle of informed consent.

**Synthesis and Evaluation**
When evaluating 'WellMind', we must consider whether the intervention aligns with holistic well-being. Real well-being cannot coexist with systemic paranoia. The platform conflates 'mental health monitoring' with 'digital surveillance'. While the policy goal is noble, the mechanism undermines trust—a foundational element of effective mental health support. If students do not trust the digital medium through which they learn and communicate, the long-term psychological impacts of surveillance may outweigh the short-term benefits of early crisis detection.

**Conclusion**
Ultimately, the blanket implementation of 'WellMind' in its current form is deeply flawed. To ethically justify such an intervention, the Ministry of Education must transition from a model of passive surveillance to one of active digital empowerment. This would involve securing explicit, parental and student consent, ensuring complete transparency in how the algorithm operates, and incorporating a strict 'human-in-the-loop' safeguard to prevent automated labeling. Without these protections, 'WellMind' risks transforming schools from safe spaces of learning into environments of digital containment.

The essay is assessed using the following 12-mark rubric criteria:

### Criterion A: Knowledge and understanding (3 marks)
* **3 marks**: Demonstrates detailed, accurate, and highly relevant knowledge of digital society concepts (such as digital well-being, surveillance, algorithmic bias, and consent) and their application to the scenario.
* **2 marks**: Demonstrates good knowledge of digital society concepts, though some points may lack depth or specific relevance to the scenario.
* **1 mark**: Demonstrates limited or superficial knowledge of digital society concepts.

### Criterion B: Application and analysis (3 marks)
* **3 marks**: Applies digital society concepts seamlessly to the 'WellMind' scenario. Analyzes the tensions between different stakeholders (e.g., students vs. Ministry of Education) with clear depth.
* **2 marks**: Applies concepts to the scenario, but the analysis is somewhat descriptive rather than deeply analytical.
* **1 mark**: Superficially applies concepts; relies heavily on repeating facts from the prompt.

### Criterion C: Evaluation and synthesis (4 marks)
* **4 marks**: Offers a highly balanced and critical evaluation of both sides of the intervention. Synthesizes perspectives to form a highly reasoned, cohesive conclusion that directly addresses the prompt.
* **3 marks**: Offers a balanced evaluation with clear arguments for and against, leading to a logical conclusion, though some synthesis of ideas could be stronger.
* **2 marks**: Presents a mainly one-sided evaluation, or the arguments are not well-integrated into a final conclusion.
* **1 mark**: Offers little to no evaluation or logical conclusion.

### Criterion D: Structure and presentation (2 marks)
* **2 marks**: The essay is well-structured, coherent, and uses appropriate digital society terminology throughout.
* **1 mark**: The essay has some structure but lacks overall coherence or appropriate terminology.

To assess the negative impacts of digital dependency, researchers can monitor:
1. Sleep quality/duration: Measuring how late-night screen activity or blue light exposure disrupts sleep cycle metrics.
2. Psychological distress levels: Tracking self-reported shifts in anxiety, depression, or feelings of isolation correlated with high social media usage.

Award [1] mark for each valid indicator identified up to a maximum of [2] marks.

Suitable indicators include:
- Sleep patterns or sleep hygiene disruption.
- Levels of anxiety, depression, or loneliness.
- Decline in physical activity / sedentary behavior patterns.
- Decreased offline social interaction hours.
- Deterioration in academic focus or performance.

Unchecked mass facial recognition technology directly impacts human rights by:
1. Violating the right to privacy (Article 12) through continuous, non-consensual surveillance and tracking of individuals in public places.
2. Chilling the right to freedom of peaceful assembly and association (Article 20), as individuals may avoid participating in protests or public gatherings out of fear of state identification and profiling.

Award [1] mark for each correctly identified UDHR right, up to a maximum of [2] marks.

Acceptable answers include:
- Right to privacy / freedom from arbitrary interference with privacy (Article 12).
- Right to freedom of peaceful assembly and association (Article 20).
- Right to freedom of movement (Article 13).
- Right to equality and non-discrimination / freedom from discrimination (Article 2 / Article 7) due to biased demographic algorithms.
- Right to freedom of opinion and expression (Article 19).

1. **Definition of intervention:** The 'right to disconnect' legally establishes boundaries preventing employers from penalizing staff who ignore out-of-hours digital communications.
2. **Contextual challenge:** Remote work relying heavily on mobile/digital collaboration tools leads to constant availability pressure and the erosion of personal/professional boundaries.
3. **Well-being outcome:** Protecting non-work hours reduces prolonged screen time, cognitive fatigue, and stress, directly improving workers' overall psychological well-being.

Award [1 mark] for defining/identifying the 'right to disconnect' as establishing clear boundaries around out-of-hours communication.
Award [1 mark] for explaining the challenge of remote work (e.g., blurred boundaries, digital fatigue, expectations of constant availability).
Award [1 mark] for linking this intervention to positive digital well-being outcomes (e.g., reduced burnout, lower stress, protected recovery time).

*Note: Accept references to organizational-level policy as well as national legislation.*

1. **Definition of DPP:** DPPs act as digital records accessible via QR codes or RFID, storing data on material composition, manufacturing origin, and disassembly instructions.
2. **Lifecycle facilitation:** This transparency helps recyclers identify hazardous or valuable elements safely and efficiently, or allows third-party repairers to source compatible parts.
3. **Sustainable development impact:** By promoting repairability and efficient material recovery, DPPs decrease the volume of electronic waste (e-waste) and reduce the need for raw material extraction, supporting circular economy goals.

Award [1 mark] for explaining the function/definition of a digital product passport (e.g., a digital identity storing raw material data, repair instructions, or recyclability info).
Award [1 mark] for explaining how this data helps actors in the lifecycle (e.g., enabling easier repair by consumers/independent shops, or accurate sorting by recycling facilities).
Award [1 mark] for connecting the intervention to a sustainable development goal (e.g., reduction of e-waste, promoting circular economy, reducing mining of critical minerals).

*Note: Accept explanations that focus on consumer transparency regarding carbon footprint as long as it links to sustainable development.*

An outstanding response should discuss both the positive and negative implications of deploying mental health chatbots in rural, marginalized areas: Opportunities: 1. Accessibility and Scalability: In remote regions with a critical shortage of psychiatrists or psychologists, AI chatbots (such as Woebot or Wysa) provide immediate, 24/7 access to cognitive behavioral therapy (CBT) techniques without waiting lists. 2. Reduced Stigma and Anonymity: Marginalized communities often have strong cultural stigmas surrounding mental illness. Chatbots offer a low-barrier, highly private environment for users to seek help without fear of local judgment. 3. Affordability: Once deployed, digital interventions scale at near-zero marginal cost, making them highly cost-effective for underfunded local clinics. Challenges: 1. Lack of Therapeutic Alliance: AI cannot genuinely empathize or form a therapeutic bond, which clinical research shows is a critical factor in successful mental health outcomes. 2. Safety and Crisis Management: Chatbots may struggle to accurately diagnose complex psychiatric conditions or recognize acute self-harm/suicidal ideation, posing severe clinical risks if escalation protocols fail. 3. Data Privacy and Cultural Bias: AI models trained on Western, educated cohorts may lack cultural sensitivity for marginalized indigenous or rural populations. Additionally, sensitive medical conversations stored in the cloud raise deep privacy concerns regarding corporate data-sharing or potential breaches. Conclusion: While AI chatbots can serve as valuable supplementary tools to bridge the immediate gap, they should not replace human practitioners entirely. Interventions should combine AI triage with localized human-in-the-loop support to protect patient well-being.

Markband Descriptors: * 6 to 8 marks: The response shows a deep understanding of the global well-being context. It offers a balanced, well-structured discussion examining both benefits and limitations of AI mental health chatbots. Refers to specific digital systems (e.g., NLP, LLMs, CBT apps) and provides a relevant real-world example (e.g., Woebot, Wysa, or NHS digital pilots). Demonstrates clear critical thinking and draws a well-justified conclusion. * 3 to 5 marks: Explains some benefits and drawbacks of using AI chatbots for mental health, but the discussion may be unbalanced or lack technical depth. Mentions some stakeholders or digital systems. There may be a generic reference to a real-world example without detailed integration. * 1 to 2 marks: Minimal understanding shown. The response is highly descriptive, listing general facts about chatbots with little or no connection to global well-being or clinical challenges. No relevant examples are used.

### Model Response Outline

#### 1. Introduction & Recommendation
* **Recommendation**: Implementation of a municipal-level **Algorithmic Guardrails and Transparency API (AGTA)**.
* **How it works**: Platform companies operating within the municipality must integrate their dispatch and driver-tracking systems with the municipal AGTA. The API enforces strict digital limits: maximum continuous shifts (e.g., 4 hours without a 15-minute lockout break), dynamic adjustment of delivery deadlines during extreme weather, and direct, unalterable access for riders to their personalized algorithmic performance data.

#### 2. Analysis of Impacts
* **On Delivery Riders**:
* *Positive*: Significant reduction in physical fatigue and mental stress; decreased rate of traffic accidents due to the removal of algorithmic penalties for delayed deliveries; enhanced sense of digital agency and transparency over how their work is evaluated.
* *Negative*: Possible reduction in short-term daily earning potential if riders are forced to take breaks during high-demand surge pricing windows.
* **On Platform Companies**:
* *Positive*: Reduced driver turnover rates and improved brand reputation; mitigation of legal liabilities stemming from rider accidents.
* *Negative*: Increased technical compliance costs; potential drop in immediate delivery fulfillment rates during peak hours, necessitating larger active driver pools.

#### 3. Addressing the Core Tension (Economic Productivity vs. Well-being)
* The AGTA directly addresses this tension by asserting that sustainable economic development cannot be built on labor exploitation. By embedding well-being standards directly into the *code* and algorithms of the platforms, it shifts the definition of 'efficiency' to include human safety parameters. It uses digital sovereignty to regulate capitalistic algorithmic optimization, proving that policy-driven digital architecture can protect human rights without entirely dismantling the convenience of the gig economy.

### Marking Rubric (Total 12 Marks)

#### **[10–12 Marks]**
* **Recommendation**: Formulates a highly appropriate, specific, and realistic digital intervention or policy framework that directly addresses the well-being of gig-economy riders.
* **Analysis**: Offers a balanced, highly detailed analysis of both positive and negative impacts on multiple stakeholders (riders, platforms, and the city). Uses accurate Digital Society terminology.
* **Synthesis/Tension**: Provides a sophisticated discussion of the tension between economic productivity and human well-being, demonstrating a deep understanding of HL Global Well-being concepts.

#### **[7–9 Marks]**
* **Recommendation**: Formulates a clear digital intervention, though some technical details of implementation may be slightly vague.
* **Analysis**: Analyzes impacts on riders and platforms, but the balance between positive and negative perspectives may be uneven.
* **Synthesis/Tension**: Explicitly addresses the tension between productivity and well-being, though the argument may rely on superficial assertions rather than deep conceptual integration.

#### **[4–6 Marks]**
* **Recommendation**: Proposes a general solution rather than a specific digital intervention or policy framework.
* **Analysis**: Descriptive rather than analytical; focus is primarily on one stakeholder (e.g., only the riders) with limited recognition of wider systemic impacts.
* **Synthesis/Tension**: Mentions well-being and productivity but fails to clearly link them to the proposed digital solution.

#### **[1–3 Marks]**
* **Recommendation**: Minimal or unrealistic proposal.
* **Analysis**: Fragmented, disjointed, or highly subjective opinions with no grounding in Digital Society concepts.
* **Synthesis/Tension**: Absent or misunderstood.