Chapter 15: Emerging Computational Paradigms and Future Directions

1. Introduction

Information technology is entering a period in which the dominant assumptions of classical computing are being reconsidered. For decades, progress was largely associated with faster processors, larger memory, improved networks, and increasingly sophisticated software. Contemporary research is expanding this model through quantum computing, neuromorphic architectures, autonomous agents, digital twins, spatial computing, decentralized systems, privacy-enhancing computation, and sustainable computing.

These paradigms are not merely new products. They represent alternative ways of representing information, performing computation, organizing trust, interacting with environments, and managing technological resources. Some remain experimental; others are already influencing industry. Their significance therefore cannot be assessed by technical novelty alone. A doctoral-level analysis must consider computational assumptions, institutional consequences, governance, economic feasibility, social risks, environmental costs, and research uncertainty.

This chapter critically investigates these paradigms and develops a framework for evaluating future information-technology developments. It emphasizes that the future is unlikely to be dominated by one universal architecture. Instead, heterogeneous and interconnected systems will combine classical, quantum, biological-inspired, distributed, immersive, and intelligent components.

2. Learning Objectives

By the end of this chapter, readers should be able to:

1. Explain the concepts and computational principles underlying emerging paradigms.

2. Compare quantum, neuromorphic, autonomous, spatial, decentralized, privacy-enhancing, and sustainable computing.

3. Evaluate their technical maturity, strategic value, limitations, and governance requirements.

4. Design research and implementation frameworks for responsible adoption.

5. Analyze how emerging paradigms may transform organizations, public institutions, and society.

6. Develop plausible scenarios and research agendas for future IT development.

3. Meaning and Concept

An emerging computational paradigm is a distinctive approach to computation that changes at least one foundational assumption about data, algorithms, hardware, interaction, or institutional control. Classical computing generally represents information as bits, executes instructions through deterministic or probabilistic electronic circuits, and separates computation from physical activity. Emerging paradigms challenge these boundaries.

Quantum computing uses quantum states and operations to process information. Neuromorphic computing imitates selected organizational principles of biological nervous systems. Autonomous agents perceive conditions, reason, plan, and act with varying degrees of independence. Digital twins create computational representations of physical or organizational systems and maintain a relationship between the model and its referent. Spatial computing integrates digital information with three-dimensional environments and embodied interaction. Decentralized technologies distribute records, decision rights, or execution across networks. Privacy-enhancing computation permits analysis while reducing exposure of sensitive data. Sustainable computing treats energy, materials, carbon, and lifecycle effects as design variables rather than externalities.

The term “paradigm” should not imply that these systems are automatically superior. A paradigm is useful only when its assumptions fit a problem. Quantum computing may be valuable for certain optimization or simulation tasks but unnecessary for ordinary administrative applications. Blockchain may support shared accountability among distrusting parties but introduce latency and governance complexity. An autonomous agent may improve responsiveness while creating unacceptable risks in a high-stakes setting. Critical evaluation is therefore essential.

4. Fundamental Principles

Several principles connect the paradigms discussed in this chapter.

Heterogeneity. Future systems will combine processors, accelerators, sensors, models, and networks with different capabilities. Programming and governance must accommodate this diversity.

Embodiment. Computation increasingly occurs in physical environments through robots, sensors, vehicles, wearables, and industrial equipment. Software behavior consequently has material consequences.

Distribution. Data, computation, identity, and authority are moving across organizational and geographic boundaries. Reliability and trust can no longer depend exclusively on a central infrastructure.

Adaptivity. Systems increasingly learn from changing data and modify their behavior. Adaptivity creates efficiency but also complicates testing, prediction, and accountability.

Privacy by design. Sensitive information should be protected during collection, storage, processing, and sharing, rather than only after a breach.

Lifecycle responsibility. Computational performance must be evaluated together with energy use, hardware sourcing, repairability, emissions, and disposal.

Human oversight. Autonomy should be proportional to the consequences of action. High-impact decisions require traceability, contestability, and meaningful human control.

5. Major Components / Elements

Quantum computing

A quantum computer uses qubits, which can occupy superposed states and become entangled. Quantum gates transform these states, and measurement produces classical outcomes. Quantum algorithms exploit interference to increase the probability of useful results. Quantum advantage is problem-dependent and should be distinguished from broad superiority over classical machines. Current systems are limited by noise, short coherence times, error-correction overhead, and scarce specialized skills.

Neuromorphic computing

Neuromorphic systems use architectures inspired by neural networks, especially event-driven signaling, distributed memory, parallelism, and low-power adaptation. Spiking neural networks transmit discrete events rather than continuously updating every value. Such systems are promising for edge devices that must respond rapidly with limited energy, although programming methods, training procedures, and benchmarking remain less mature than those for conventional machine learning.

Autonomous agents

An autonomous agent observes an environment, maintains an internal state, selects actions, and evaluates outcomes. Modern agents may combine foundation models, planning modules, tools, retrieval systems, memory, and feedback. Autonomy ranges from recommendation to fully executed action. The central research problem is not simply producing intelligent outputs but ensuring reliable behavior under uncertainty, conflicting goals, adversarial conditions, and changing environments.

Digital twins

A digital twin is a continuously updated computational representation of a physical asset, process, or system. It normally includes sensors, data integration, simulation, analytics, and mechanisms for comparing predicted and observed states. A static digital model is not necessarily a digital twin; the defining feature is an ongoing relationship with the real-world referent.

Spatial computing

Spatial computing enables people and machines to understand and act within three-dimensional environments. It combines computer vision, spatial mapping, sensors, graphics, extended reality, and embodied interfaces. Applications include immersive training, design collaboration, maintenance assistance, medical visualization, and robotic coordination.

Decentralized technologies

Decentralized technologies distribute data, validation, identity, or execution across multiple participants. Distributed ledgers, peer-to-peer systems, decentralized identity, and smart contracts are examples. Their value lies in shared verification and reduced dependence on a single intermediary, but decentralization is multidimensional: a system may distribute data while retaining centralized governance or hardware control.

Privacy-enhancing computation

Privacy-enhancing computation includes methods such as homomorphic encryption, secure multiparty computation, differential privacy, trusted execution environments, and federated learning. These methods reduce the need to reveal raw data. Each involves trade-offs involving accuracy, computation, communication, trust assumptions, or implementation complexity.

Sustainable computing

Sustainable computing seeks to reduce the environmental and social impacts of information systems throughout their lifecycle. It addresses energy efficiency, carbon intensity, water use, material extraction, hardware longevity, data-center cooling, software efficiency, and responsible disposal. It also considers how computing can support sustainability in other sectors without shifting burdens elsewhere.

6. Process / Framework / Methodology

Organizations can evaluate an emerging paradigm through a seven-stage framework.

First, define the problem. Identify the operational, scientific, or social objective before selecting a technology. “Adopt artificial intelligence” is not a sufficiently precise goal; reducing diagnostic delay or improving equipment utilization is more useful.

Second, characterize the environment. Examine data quality, latency requirements, safety conditions, regulations, stakeholders, existing infrastructure, and failure consequences.

Third, match paradigm to requirement. Determine whether the problem benefits from quantum optimization, event-driven processing, autonomy, immersive interaction, distributed trust, protected computation, or lifecycle efficiency.

Fourth, establish a baseline. Compare the proposed system with the best practical classical or organizational alternative. Claims of innovation are meaningful only relative to cost, performance, reliability, and environmental impact.

Fifth, prototype safely. Use sandboxes, simulations, synthetic data, limited pilots, and human review. Autonomous or externally connected systems should begin with restricted permissions.

Sixth, evaluate multidimensionally. Measure accuracy, latency, robustness, explainability, security, privacy, energy, carbon, accessibility, total cost, and social effects.

Seventh, govern and scale. Define responsibility, audit procedures, escalation rules, update mechanisms, procurement standards, and exit plans before deployment at scale.

7. Important Theories and Models

The Church–Turing perspective provides a foundation for asking what can be computed, while quantum computing investigates whether different physical resources can provide substantial efficiency improvements for particular problems. Complexity theory remains important because a faster device does not make every computational problem tractable.

The No Free Lunch principle suggests that no optimization method is universally best across all problem classes. This supports careful matching of paradigms to tasks rather than assuming that quantum, neural, or decentralized methods are generally superior.

The agent-environment model represents an intelligent system as an agent receiving percepts and taking actions. It supports analysis of observability, determinism, episodic versus sequential decision-making, and the degree of autonomy required.

The cyber-physical systems model emphasizes feedback between computational processes and physical processes. It is fundamental to digital twins, robotics, smart infrastructure, and autonomous vehicles, where software errors can produce physical harm.

The triple-bottom-line model evaluates economic, environmental, and social performance. For computing, it should be extended with governance and technical resilience because a low-energy system that violates privacy or fails unpredictably is not sustainable in a broader sense.

The socio-technical systems perspective treats technology, institutions, professional practices, incentives, and users as interdependent. It explains why technically successful pilots may fail when they conflict with organizational authority, worker expertise, or public expectations.

8. Practical Applications

Quantum methods may support molecular simulation, portfolio optimization, scheduling, cryptanalysis research, and logistics, although many applications remain exploratory. Neuromorphic processors are relevant to always-on sensing, robotics, prosthetics, anomaly detection, and low-power industrial monitoring.

Autonomous agents can support software operations, research assistance, customer service, supply-chain coordination, and adaptive tutoring. Their deployment should be risk-tiered: drafting and recommendation are lower risk than autonomous financial transactions, clinical decisions, or infrastructure control.

Digital twins assist predictive maintenance, production optimization, urban planning, energy management, and aerospace engineering. Spatial computing supports remote maintenance, surgical planning, immersive education, and collaborative product design. Decentralized systems can improve provenance, cross-organizational settlement, digital identity, and peer-to-peer coordination.

Privacy-enhancing computation is valuable in healthcare research, financial fraud detection, cross-border analytics, and public-sector data collaboration. Sustainable computing applies to cloud architecture, workload scheduling, efficient algorithms, device refurbishment, renewable-powered data centers, and carbon-aware software deployment.

9. Examples

A hospital might use federated learning to train a diagnostic model across institutions without transferring patient records. Differential privacy can reduce the risk of identifying individuals in published statistics, although excessive noise may reduce usefulness.

A manufacturer may combine a digital twin with edge sensors and an autonomous maintenance agent. The twin estimates equipment health, while the agent recommends or schedules intervention. A human engineer may retain approval authority for safety-critical actions.

A city could use spatial computing to visualize traffic, energy, and construction scenarios. If these data are placed on a decentralized platform, governance must still determine who validates records, corrects errors, and protects residents from surveillance.

10. Technology and Digital Applications

Relevant technologies include quantum processors and simulators; neuromorphic chips; edge artificial intelligence; foundation models and agent orchestration frameworks; Internet of Things sensors; digital-twin platforms; augmented, virtual, and mixed reality; distributed ledgers and decentralized identity; homomorphic encryption and secure multiparty computation; federated-learning systems; carbon-aware cloud schedulers; and lifecycle-monitoring tools.

These technologies should be integrated selectively. For example, an edge device may use a neuromorphic accelerator for event detection, transmit only protected summaries to a cloud service, update a digital twin, and allow an autonomous agent to recommend action. Such architectures illustrate convergence rather than isolated adoption.

11. Strategic / Managerial Importance

Managers must develop technological ambidexterity: the ability to maintain dependable current systems while exploring uncertain alternatives. Emerging paradigms create opportunities for new services, operational efficiency, resilience, and competitive differentiation. They also create risks of vendor dependence, skills shortages, regulatory exposure, stranded investments, and reputational damage.

Strategic decisions should be based on capabilities rather than hype. Organizations need architecture road maps, interdisciplinary teams, experimental budgets, data governance, workforce development, and measurable adoption criteria. Procurement should assess interoperability, auditability, energy performance, security, accessibility, and exit options. Boards and senior executives should ask not only what a system can do, but who controls it, how it fails, and who bears the consequences.

12. Detailed Case Study

Background

The following is a fictional illustrative case. Meridian Transit Authority operates buses, rail vehicles, stations, and a centralized control center. It wants to reduce delays, maintenance costs, and energy consumption while improving passenger information.

Problem

The authority has fragmented sensor data, aging maintenance practices, high peak-hour energy demand, and limited ability to test operational changes safely. It considers deploying a digital twin, autonomous agents, spatial visualization, and privacy-enhancing analytics.

Analysis

A technology-first approach would create significant risk. The data are inconsistent, many assets lack sensors, and operational decisions involve safety regulations and labor agreements. Full autonomy is inappropriate for dispatching and emergency response. Passenger-location data also create privacy concerns.

Strategy / Solution

Meridian first established a data-quality program and created a baseline for delays, energy use, and failures. It then developed a digital twin for a selected rail corridor. Edge devices detected equipment anomalies, while a cloud model simulated schedules and energy scenarios. An autonomous agent generated maintenance recommendations but required engineer approval. Federated learning allowed multiple depots to improve failure prediction without centralizing all detailed operational data. A spatial interface enabled engineers to inspect simulated failures in a three-dimensional station model. Carbon-aware scheduling shifted nonurgent model training to periods with lower grid emissions.

Outcome

In this illustrative scenario, the pilot reduced unplanned maintenance events and improved the accuracy of disruption forecasts. It did not eliminate delays, and integration costs were substantial. The authority retained human authorization for safety-critical actions, published a data-governance policy, and created an independent review committee. The most valuable result was not a single technology but a controlled architecture linking sensing, simulation, protected analytics, and accountable decision-making.

Learning Points

The case demonstrates that emerging paradigms are complementary. It also shows that organizational readiness, data quality, governance, and baseline measurement are prerequisites for technical value. A staged deployment produces stronger evidence than simultaneous large-scale transformation.

13. Advantages / Benefits

Emerging paradigms can provide substantial benefits:

• Quantum and specialized hardware may improve selected optimization, simulation, or sensing tasks.
• Neuromorphic and edge systems can reduce latency, bandwidth use, and energy consumption.
• Autonomous agents can extend organizational capacity and respond continuously.
• Digital twins can support experimentation without disrupting physical operations.
• Spatial computing can improve understanding of complex environments and collaborative work.
• Decentralization can reduce single points of failure and support shared verification.
• Privacy-enhancing computation can enable collaboration over sensitive data.
• Sustainable computing can reduce operating costs, emissions, and material waste.

These benefits are conditional. They depend on suitable problem selection, reliable data, competent governance, and realistic evaluation.

14. Limitations / Challenges

Technical limitations include immature hardware, difficult programming models, interoperability problems, uncertain benchmarks, adversarial attacks, model drift, and inadequate testing of adaptive systems. Quantum systems face error correction and scaling challenges; neuromorphic systems face training and software ecosystem limitations; autonomous agents may hallucinate, pursue poorly specified goals, or misuse tools.

Organizational limitations include skills shortages, legacy integration, unclear accountability, procurement constraints, and resistance to altered roles. Economic challenges include high capital expenditure, uncertain return on investment, and dependence on specialized suppliers. Environmental impacts may increase when computational demand grows faster than efficiency improvements.

Decentralization does not automatically create democracy or trust. Distributed systems can reproduce inequalities through token ownership, validator concentration, or inaccessible governance. Privacy technologies may protect data while making legitimate auditing more difficult. Spatial systems may exclude users with disabilities or normalize pervasive surveillance.

15. Ethical and Governance Considerations

Governance must address responsibility, fairness, privacy, safety, security, explainability, accessibility, and sustainability. Autonomous systems require explicit authority boundaries, logging, human escalation, and mechanisms for appeal. Organizations should distinguish assistance from delegation: a system that recommends an action is not equivalent to one that executes it.

Data governance should specify purpose limitation, retention, consent or lawful basis, access controls, provenance, and deletion or correction procedures. Privacy-enhancing methods should be selected according to threat models rather than used as symbolic guarantees. Decentralized systems require governance for software updates, dispute resolution, identity recovery, and malicious participants.

Environmental governance should measure lifecycle impacts, including embodied emissions from hardware, water consumption, energy sources, and electronic waste. Ethical review should include affected communities, workers, and nonusers, not only system owners. Independent audits, impact assessments, red-team exercises, and transparent incident reporting are essential for high-impact deployments.

16. Current and Emerging Trends

Important trends include hybrid quantum-classical workflows, quantum-safe cryptography, edge and federated intelligence, increasingly capable tool-using agents, multimodal foundation models, real-time digital twins, spatial collaboration, decentralized identity, confidential cloud computing, and carbon-aware workload orchestration.

Convergence is especially significant. Digital twins are becoming intelligent and spatial; autonomous agents are being embedded in physical systems; privacy technologies are moving toward hardware-software co-design; and sustainability metrics are entering cloud procurement. At the same time, regulation is shifting from general principles toward risk-based obligations, technical standards, and sector-specific controls.

17. Best Practices

1. Begin with a clearly defined problem and measurable baseline.

2. Use the least complex paradigm that satisfies the requirement.

3. Separate experimentation from production and restrict early permissions.

4. Design for interoperability, portability, and graceful failure.

5. Apply threat modeling, privacy impact assessment, and lifecycle assessment.

6. Maintain human oversight proportionate to risk.

7. Record data provenance, model versions, decisions, and incidents.

8. Test under distribution shift, adversarial conditions, and degraded connectivity.

9. Include users, workers, affected communities, and domain experts in design.

10. Establish monitoring, retraining, updating, rollback, and retirement procedures.

11. Measure total cost, energy, carbon, accessibility, and social outcomes.

12. Develop workforce skills in systems thinking, ethics, security, and domain practice.

18. Future Scope

Future research should investigate hybrid architectures that allocate tasks dynamically across classical, quantum, neuromorphic, edge, and cloud resources. Important questions concern formal verification of autonomous agents, reliable memory and planning, safe human-agent collaboration, and evaluation methods that measure social as well as technical performance.

Digital-twin research should improve semantic interoperability, uncertainty representation, causal modeling, and privacy-preserving synchronization. Spatial computing requires better accessibility, shared standards, identity protection, and methods for reducing motion sickness and cognitive overload. Decentralized technologies need scalable governance, energy-efficient consensus, usable recovery mechanisms, and evidence about real-world institutional value.

Privacy-enhancing computation requires practical combinations of methods with acceptable accuracy and performance. Sustainable computing needs standardized lifecycle metrics, repairable hardware, efficient software, renewable-aware scheduling, and research into rebound effects. A broad research agenda should also examine geopolitical dependence, digital inequality, labor transformation, and the distribution of benefits and risks.

Three scenarios are useful for planning. In a distributed intelligence scenario, edge devices and autonomous agents coordinate locally while sharing protected summaries. In a quantum-augmented enterprise scenario, quantum services are accessed selectively through hybrid cloud workflows for specialized optimization. In a regulated immersive infrastructure scenario, spatial systems become normal in education, industry, and public administration but operate under strong identity, accessibility, and surveillance controls. Scenario planning should test resilience rather than predict a single future.

19. Chapter Summary

Emerging computational paradigms expand the foundations of information technology beyond conventional centralized, binary, and screen-based computing. Quantum and neuromorphic systems offer alternative computational mechanisms; autonomous agents introduce adaptive action; digital twins connect models to physical reality; spatial computing changes interaction; decentralized technologies redistribute trust; privacy-enhancing computation protects sensitive collaboration; and sustainable computing integrates lifecycle responsibility.

Their value is contextual and conditional. Effective adoption requires problem definition, baseline comparison, controlled experimentation, multidimensional evaluation, and governance proportional to risk. The future of IT will most likely consist of hybrid socio-technical ecosystems rather than a single dominant paradigm. Doctoral research should therefore combine technical innovation with critical analysis of institutions, ethics, sustainability, and public value.

20. Key Terms

1. Autonomous agent: A computational system that perceives conditions and selects actions with limited direct human control.

2. Carbon-aware computing: Scheduling or designing computation according to carbon intensity and environmental conditions.

3. Decentralized technology: Technology that distributes data, authority, validation, or execution across participants.

4. Digital twin: A continuously updated computational representation of a physical or organizational system.

5. Differential privacy: A method that limits the privacy loss caused by including an individual's data in an analysis.

6. Edge computing: Processing data near its source rather than exclusively in a distant cloud.

7. Federated learning: Collaborative model training in which data remain at participating sites.

8. Homomorphic encryption: Encryption that permits computation on protected data.

9. Neuromorphic computing: Computing inspired by neural organization and often based on event-driven processing.

10. Privacy-enhancing computation: Methods that enable useful computation while reducing exposure of sensitive information.

11. Quantum computing: Computation using quantum states, gates, and measurement.

12. Quantum bit (qubit): The basic unit of quantum information.

13. Spatial computing: Interaction with digital information through spatially aware, three-dimensional environments.

14. Spiking neural network: A neural model that communicates through discrete temporal events or spikes.

15. Sustainable computing: Design and use of computing that minimizes lifecycle environmental and social harm.

16. Trusted execution environment: A protected hardware-supported area for processing sensitive code and data.

17. Digital provenance: Documented history showing the origin, transformations, and custody of data or artifacts.

18. Human oversight: Deliberate human monitoring, authorization, intervention, or review of system behavior.

21. Review Questions

Short Answer Questions

1. What distinguishes a digital twin from a static digital model?

2. Define privacy-enhancing computation and name two techniques.

3. Why is quantum advantage problem-dependent?

4. What is the role of a baseline in evaluating emerging technology?

5. Identify two lifecycle concerns in sustainable computing.

Long Answer Questions

1. Critically compare quantum and neuromorphic computing in terms of representation, performance, maturity, and potential applications.

2. Explain how autonomous agents can be governed in high-impact organizational settings.

3. Discuss the relationship among digital twins, spatial computing, and cyber-physical systems.

4. Evaluate the claim that decentralization automatically increases trust and fairness.

5. Develop a governance framework for privacy-enhancing computation in healthcare research.

Analytical / Application Questions

1. A logistics firm wants to use an autonomous agent to reroute vehicles, access supplier systems, and approve emergency purchases. Design a staged evaluation and governance plan.

2. A university plans an immersive spatial-learning platform that collects behavioral and biometric data. Analyze the privacy, accessibility, sustainability, and educational-value issues.

3. An industrial company proposes a quantum-computing investment because competitors are doing so. Construct a decision analysis that compares a quantum pilot with classical optimization and neuromorphic edge processing.

22. Multiple Choice Questions with Answers

Question 1

Which feature most clearly distinguishes a digital twin from a static digital model?

A. It uses a three-dimensional interface in every case.

B. It maintains an ongoing relationship with a physical or organizational referent.

C. It is always hosted on a decentralized ledger.

D. It eliminates the need for physical sensors.

Correct Answer: B. It maintains an ongoing relationship with a physical or organizational referent.

Question 2

Why should an organization establish a baseline before adopting an emerging computational paradigm?

A. To ensure that the newest technology is selected.

B. To avoid measuring environmental effects.

C. To compare the proposed system with the best practical alternative.

D. To transfer accountability to the technology vendor.

Correct Answer: C. To compare the proposed system with the best practical alternative.

Question 3

Which application is most naturally associated with neuromorphic computing?

A. Always-on, low-power event detection at the edge.

B. Replacing all database management systems.

C. Guaranteeing legally binding identity recovery.

D. Performing every type of optimization faster than classical computers.

Correct Answer: A. Always-on, low-power event detection at the edge.

Question 4

What is a central governance requirement for a high-impact autonomous agent?

A. Removing all human intervention to maximize efficiency.

B. Allowing the agent unrestricted access to organizational systems.

C. Defining authority boundaries, logging, escalation, and review procedures.

D. Using only publicly available training data.

Correct Answer: C. Defining authority boundaries, logging, escalation, and review procedures.

Question 5

Which technique enables collaborative model training while keeping data at participating institutions?

A. Federated learning

B. Quantum measurement

C. Spatial mapping

D. Carbon-aware scheduling

Correct Answer: A. Federated learning

Question 6

What does sustainable computing evaluate in addition to operational energy use?

A. Only processor clock speed.

B. Hardware lifecycle, materials, water, emissions, and disposal impacts.

C. Only the number of software features.

D. The number of users who prefer immersive interfaces.

Correct Answer: B. Hardware lifecycle, materials, water, emissions, and disposal impacts.

Question 7

Which statement best characterizes decentralized technology?

A. All decentralized systems have no governance structure.

B. Decentralization may distribute validation or authority without eliminating central control in every layer.

C. Decentralized systems cannot experience concentration of power.

D. Decentralization guarantees privacy by itself.

Correct Answer: B. Decentralization may distribute validation or authority without eliminating central control in every layer.

Question 8

Why is quantum computing not universally superior to classical computing?

A. Qubits cannot represent information.

B. Quantum algorithms and advantages are problem-dependent and subject to hardware limitations.

C. Quantum systems cannot use mathematical algorithms.

D. Classical computers cannot perform optimization.

Correct Answer: B. Quantum algorithms and advantages are problem-dependent and subject to hardware limitations.

Question 9

Which principle is most appropriate when assigning autonomy to a safety-critical system?

A. Autonomy should be proportional to the consequences of action.

B. Autonomy should always be maximized.

C. Human oversight should be removed after deployment.

D. The system should conceal uncertainty from users.

Correct Answer: A. Autonomy should be proportional to the consequences of action.

Question 10

A company wants to process sensitive data across competing hospitals without pooling raw records. Which approach is most relevant?

A. Privacy-enhancing computation combined with federated methods.

B. Unrestricted centralized data collection.

C. A spatial display without access controls.

D. Replacing all data with synthetic images without validation.

Correct Answer: A. Privacy-enhancing computation combined with federated methods.