Digital Entrepreneurship, AI and Technology-Enabled Ventures

Introduction

Digital technologies have altered not only how ventures operate but also what a venture is. A conventional enterprise typically creates value through physical assets, employees, locations, and linear transactions. A digital venture may create value through software, data, algorithms, online communities, cloud infrastructure, and digitally coordinated partners. Its products may be replicated at very low marginal cost, distributed globally, updated continuously, and improved through usage data.

Digital entrepreneurship therefore refers to the identification, development, and exploitation of opportunities in which digital technologies are central to value creation, delivery, or capture. It includes software startups, platform businesses, data-product firms, digitally enabled manufacturers, online marketplaces, creator businesses, and established firms transformed through automation and artificial intelligence (AI).

For graduate entrepreneurs and managers, the central question is not whether technology should be used. The more important questions are: Which technology creates defensible value? Which capabilities should be built internally? Which should be bought or obtained through partnerships? What data are required, and can they be used lawfully and ethically? How should AI-related risk be governed? How does digital technology alter the economics and organizational design of a venture?

This chapter addresses these questions by connecting digital business models with platform strategy, cloud services, data products, automation, generative AI, governance, and strategic technology choices.

Learning Objectives

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

1. Explain the characteristics and economics of digital entrepreneurship.
2. Distinguish platforms, digital products, data products, automation systems, and AI-enabled ventures.
3. Evaluate cloud, data, automation, and generative-AI opportunities.
4. Make reasoned build–buy–partner decisions.
5. Design a responsible data strategy and identify algorithmic risks.
6. Assess digitally transformed business models and relevant performance metrics.
7. Develop an implementation and governance approach for a technology-enabled venture.

1. The Nature of Digital Entrepreneurship

Digital entrepreneurship is broader than founding a software company. It occurs whenever digital technologies materially shape entrepreneurial opportunity, resources, processes, or outcomes. A restaurant using predictive demand data, a logistics venture coordinating independent drivers through an app, and a manufacturer selling equipment through remote monitoring are all examples of digital entrepreneurship.

Four characteristics are especially significant.

1.1 Low-cost replication and scalability

Once developed, many digital products can be reproduced at low additional cost. A software service can serve another customer without manufacturing another physical unit, although hosting, support, compliance, and customer-acquisition costs still rise. This creates the possibility of rapid scale, but not automatic profitability. A venture may have low marginal production cost while facing high costs for trust, sales, infrastructure, security, and regulation.

1.2 Intangible and cumulative assets

Code, data, workflows, models, brands, user networks, and organizational learning are often more important than physical equipment. These assets may become more valuable as they accumulate. However, data are not automatically an asset in the economic or legal sense. Their value depends on relevance, quality, permission, accessibility, security, and the ability to convert them into decisions or customer outcomes.

1.3 Continuous experimentation

Digital ventures can test interfaces, prices, recommendations, and communication rapidly. Experimentation can reduce uncertainty, but it can also encourage careless optimization. A statistically successful experiment may damage trust, exploit vulnerable users, or optimize a short-term metric at the expense of long-term value.

1.4 Interdependence with ecosystems

Digital ventures depend on app stores, payment providers, cloud platforms, identity services, data suppliers, application programming interfaces (APIs), regulators, and online communities. Entrepreneurs must therefore manage not only a firm but also a technical and institutional ecosystem.

2. Digital Business Models and Venture Economics

A business model explains how a venture creates, delivers, and captures value. Digital technology can alter each component.

A useful distinction is between digitized, digital, and digitally transformed ventures. Digitization converts analogue information into digital form, such as scanning paper records. Digitalization uses digital systems to improve an existing process, such as online scheduling. Digital transformation changes the underlying operating and business model, such as moving from selling industrial equipment once to providing equipment-as-a-service supported by sensors and predictive maintenance.

Network effects

A network effect exists when the value of a product or service changes as participation changes. Direct network effects occur when users benefit from more users of the same service, as in communication networks. Indirect network effects occur when growth on one side attracts participants on another side, as in a marketplace connecting buyers and sellers.

Network effects can create defensibility, but they are not guaranteed. They may be weak, local, negative, or reversible. Congestion, fraud, poor quality, and privacy concerns can reduce value as participation grows. Entrepreneurs should specify which users benefit, through what mechanism, and under what conditions.

Digital unit economics

Digital ventures should distinguish accounting profit from unit economics. Unit economics examines the contribution of an individual customer, transaction, account, or usage unit.

A simple contribution calculation is:

Revenue and variable cost must be measured over the same period and include costs that genuinely vary with the customer. If a subscription generates $40 per month and variable service, payment, and support costs total $12, the contribution margin is $28 per month. Fixed costs such as executive salaries and long-term software development are not deducted in this calculation.

Customer acquisition cost (CAC) can be represented as:

The numerator and denominator should use the same acquisition period and attribution rule. A venture should avoid treating all marketing expenditure as attributable to customers if some expenditure builds a long-term brand.

Customer lifetime value (CLV) is not one universal number. Under a simplified constant-margin model, with stable retention and no discounting:

where m is contribution margin per period and r is the retention rate per period. The formula assumes constant contribution margin, constant retention, indefinite continuation, and no discounting. These assumptions are often unrealistic. A discounted formulation is more appropriate when cash flows vary over time:

where m_t is contribution margin in period t, p_t is the probability that the customer remains active in that period, i is the discount rate per period, and T is the planning horizon. The units of i must match the time period used for t.

These metrics support decisions but should not be treated as laws. High CLV relative to CAC may reflect strong economics, poor attribution, undercounted service costs, or temporary retention.

3. Platforms, Cloud Services, and Digital Infrastructure

3.1 Platform ventures

A platform facilitates interactions between two or more participant groups. Examples include buyers and sellers, hosts and guests, developers and users, or advertisers and audiences. The platform usually establishes rules, provides search and matching, manages identity and payments, and governs disputes.

A platform strategy requires attention to four design questions:

1. Participation: Which sides must be recruited first?
2. Matching: How are participants connected, and how is quality maintained?
3. Governance: What conduct, access, pricing, and data rules apply?
4. Monetization: Which side pays, and at what point in the interaction?

Early platform ventures commonly face a “cold-start” problem: buyers do not want an empty marketplace, while sellers do not want to join without buyers. Solutions include focusing on a narrow geographic or professional niche, subsidizing one side, manually creating initial supply, or integrating with existing communities.

Platform governance is a strategic capability. Rules concerning ranking, commissions, reviews, content moderation, account suspension, and data access influence fairness and trust. A platform that extracts value aggressively may encourage participants to multi-home, leave, or develop alternatives.

3.2 Cloud services

Cloud computing provides on-demand access to computing, storage, databases, networking, analytics, and AI capabilities. Its entrepreneurial importance lies in converting large capital expenditures into more flexible operating expenditure and enabling rapid experimentation.

Common service categories include:

• Infrastructure as a Service (IaaS): virtual machines, storage, and networking.
• Platform as a Service (PaaS): managed development and deployment environments.
• Software as a Service (SaaS): complete applications delivered through a network.
• Function or serverless services: execution of code in response to events without direct server management.

Cloud adoption does not remove technology risk. It can create vendor dependence, unexpected usage charges, data-residency issues, outages, cybersecurity exposure, and complexity in migrating data or applications. A sound venture evaluates portability, service-level commitments, recovery arrangements, access controls, and total cost of ownership.

3.3 APIs and composability

An API is a defined interface through which software components exchange requests and responses. APIs enable ventures to combine payments, maps, identity verification, messaging, analytics, and AI without developing every capability internally. Their advantages include speed and specialization; their disadvantages include dependency, changing terms, rate limits, privacy exposure, and cascading failure when a supplier becomes unavailable.

4. Data Products and Data Strategy

A data product uses data as a central input to deliver information, prediction, recommendation, monitoring, or decision support. Examples include fraud alerts, demand forecasts, credit-risk assessments, route optimization, and industry benchmarks.

A robust data strategy addresses the complete data lifecycle:

1. Purpose: What decision or customer outcome will data improve?
2. Collection: Is collection necessary, lawful, proportionate, and consented where required?
3. Quality: Are data accurate, complete, timely, representative, and consistently defined?
4. Storage: Where are data stored, for how long, and under whose control?
5. Use: Who may access data, and for which approved purposes?
6. Sharing: Are data shared with partners under clear contractual and security conditions?
7. Retention and deletion: When should data be archived or destroyed?

The principle of data minimization is strategically useful: collecting more data is not always better. Excess data increases security obligations, legal exposure, storage costs, and the possibility of harmful secondary uses.

Data quality can be assessed through dimensions such as accuracy, completeness, consistency, timeliness, validity, uniqueness, and lineage. Data lineage records where data originated, how they were transformed, and where they were used. It is particularly important when an AI output influences a consequential decision.

A data product must also solve the “last-mile” problem. A highly accurate prediction has little value if a customer cannot interpret it, trust it, or act upon it. Value arises when data are embedded in a workflow with clear responsibility and feedback.

5. Automation, AI, and Generative AI

Automation uses technology to perform tasks with limited human intervention. It can be rule-based, such as automatically sending an invoice, or model-based, such as predicting which support ticket requires escalation.

AI systems infer patterns from data or use programmed methods to perform tasks associated with perception, prediction, language, planning, or decision support. Generative AI produces new text, images, audio, code, or other content in response to prompts or structured inputs. Large language models generate outputs by learning statistical relationships in extensive training data; they do not guarantee factual accuracy, legal ownership, or sound judgment.

Opportunity areas

Entrepreneurs can use AI to:

• reduce administrative work;
• improve search, recommendation, and personalization;
• assist customer support;
• summarize documents and generate drafts;
• detect anomalies and fraud;
• forecast demand;
• support product design and software development;
• create new forms of content and interaction.

The strongest opportunities typically combine AI with proprietary workflow knowledge, high-quality domain data, distribution, or integration into a customer’s operational system. Merely placing a generic model behind a new interface may be easy to copy.

AI system lifecycle

An AI-enabled venture should manage a lifecycle rather than treat a model as a one-time product:

1. Define the use case and unacceptable uses.
2. Identify data sources, rights, quality, and representativeness.
3. Select or build a model appropriate to the task.
4. Validate performance against relevant benchmarks.
5. Test safety, robustness, privacy, security, and bias.
6. Deploy with human oversight and access controls.
7. Monitor drift, incidents, user feedback, and changing conditions.
8. Retrain, restrict, replace, or retire the system when necessary.

Model performance and business performance

Technical accuracy is only one dimension of success. A model can be accurate but economically useless if it produces too many costly false positives, cannot be integrated into operations, or undermines customer trust.

For binary classification, precision is:

Recall is:

Precision answers: among items predicted positive, how many were actually positive? Recall answers: among actual positive items, how many were identified? Both measures assume clearly defined positive and negative classes and a chosen decision threshold. In fraud detection, for example, high recall may identify more fraud but generate more investigations; high precision may reduce wasted investigations but miss more fraud.

Generative AI introduces additional risks: hallucinated information, prompt injection, confidential-data leakage, copyright disputes, inconsistent outputs, hidden bias, and overreliance by users. Appropriate controls include retrieval from approved sources, output validation, restricted data handling, human review for high-impact decisions, logging, red-teaming, and clear disclosure of system limitations.

6. Build, Buy, or Partner?

Technology choices should be treated as strategic choices rather than purely technical preferences.

A build decision is more attractive when the capability is strategically distinctive, changes rapidly with the venture’s learning, requires deep integration, or creates proprietary advantage. Buying is attractive when the capability is standardized, mature, non-differentiating, and available from reliable suppliers. Partnership is useful when complementary assets are required, such as a startup’s analytics combined with a hospital’s clinical distribution or a manufacturer’s equipment combined with a software venture’s monitoring system.

A practical decision matrix can score alternatives against weighted criteria. Let s_j be the score of option j on criterion k, and w_k be the weight of criterion k, with weights summing to 1:

This is a prioritization tool, not an objective truth. Scores should be documented, sensitivity-tested, and supplemented by legal, security, and scenario analysis.

Important criteria include strategic uniqueness, time to market, total cost, scalability, security, data control, interoperability, regulatory exposure, reversibility, and organizational learning. Entrepreneurs should also distinguish switching costs from irreversible dependence. A cheap vendor can become expensive if migration is technically or contractually difficult.

7. AI Governance and Algorithmic Risk

AI governance is the system of policies, roles, processes, controls, and evidence used to ensure that AI is lawful, safe, explainable enough for its context, secure, and aligned with organizational purpose. Governance should be proportionate to risk rather than identical for every application.

7.1 A risk-based approach

A customer-service drafting tool and an automated employment-screening system should not receive identical scrutiny. Risk assessment should consider:

• the severity of possible harm;
• the number and vulnerability of affected people;
• whether decisions are reversible;
• the degree of human control;
• the sensitivity of data;
• the likelihood of misuse;
• the explainability and auditability required;
• applicable sectoral and jurisdictional rules.

7.2 Principal algorithmic risks

Bias and discrimination may arise from historical data, sampling, labels, features, proxy variables, or unequal error costs. Removing a sensitive attribute does not necessarily remove discrimination because other variables may act as proxies.

Privacy risk includes unauthorized collection, re-identification, inference of sensitive attributes, and use beyond the original purpose.

Security risk includes data poisoning, model theft, adversarial manipulation, prompt injection, and unauthorized access.

Reliability risk occurs when systems fail under conditions different from training or validation conditions. Data drift and concept drift can reduce performance over time.

Opacity and contestability become problematic when affected people cannot understand, challenge, or correct an important decision.

Accountability gaps arise when responsibility is distributed among model providers, developers, deployers, managers, and users without clear ownership.

7.3 Governance controls

Effective controls include an inventory of AI systems, named system owners, documented intended use, data and model cards, approval gates, access controls, audit logs, incident reporting, monitoring, impact assessments, vendor due diligence, and retirement procedures. Human oversight must be meaningful: a person who cannot understand, question, or override an output is merely a ceremonial reviewer.

Governance should also address intellectual property and contracting. Agreements with model or data suppliers should clarify permitted uses, confidentiality, retention, training rights, security responsibilities, liability, service continuity, audit access, and termination assistance.

8. Designing and Scaling a Technology-Enabled Venture

A disciplined venture process links customer discovery with technical feasibility and governance.

Stage 1: Define the problem

Start with a costly, frequent, and sufficiently important customer problem. Avoid beginning with a technology looking for a use. Identify the user, decision, workflow, current workaround, and measurable outcome.

Stage 2: Test the value proposition

Use interviews, observation, prototypes, concierge services, and manually delivered solutions. Early manual work is not a failure of automation; it can reveal what should and should not be automated.

Stage 3: Establish a minimum viable architecture

Select the smallest reliable architecture that can test the critical assumptions. Separate core intellectual property from replaceable services. Establish identity management, backups, logging, security, and data-retention rules earlier than founders often expect.

Stage 4: Validate economically and technically

Measure activation, retention, conversion, contribution margin, service quality, latency, error rates, and customer outcomes. Test adverse cases, not only average performance. A venture should know what happens when data are missing, traffic spikes, a supplier fails, or a model is confidently wrong.

Stage 5: Scale responsibly

Scaling requires process standardization, observability, incident response, documentation, hiring, vendor management, and governance. Technical debt—the future cost created by expedient technical choices—should be tracked like financial debt. Some technical debt is rational during experimentation; unmanaged debt can prevent reliable growth.

9. Illustrative Case: CareRoute Analytics

The following case is fictional and is intended for educational analysis.

Background

CareRoute Analytics is a startup offering scheduling software to home-care agencies. Agencies coordinate caregivers, patients, travel times, qualifications, and last-minute absences. Existing scheduling is largely manual and generates overtime, missed visits, and caregiver dissatisfaction.

Problem

The founders propose a generative-AI assistant that automatically creates schedules and sends messages to caregivers. They have limited historical data, inconsistent records across agencies, and no established process for handling patient-sensitive information.

Analysis

The opportunity is not simply “use AI in healthcare.” The customer problem is coordination under operational constraints. Some elements are suitable for deterministic optimization and rules: caregiver qualifications, availability, working-hour limits, and travel-time constraints. Generative AI may be useful for drafting explanations or messages, but it should not independently make high-impact decisions without controls.

The founders compare three options:

They decide to build the scheduling logic and workflow knowledge, buy mapping services and foundational AI capabilities, and partner with an established healthcare software provider for distribution and integration.

Approach and solution

The startup establishes a data inventory and separates personally identifiable information from development datasets where possible. It obtains contractual permissions, limits retention, encrypts data, and creates role-based access. The scheduling engine produces candidate schedules subject to explicit constraints. A human coordinator approves schedules before release. The language model drafts messages from approved templates and structured facts; it cannot invent appointment details or alter clinical instructions.

The pilot measures travel reduction, missed visits, overtime, coordinator workload, caregiver acceptance, and error incidents. It also monitors performance by agency, geography, caregiver type, and workload level to identify uneven outcomes.

Outcome

In the pilot, the venture finds that the largest value comes from reliable data cleaning and constraint handling rather than from sophisticated language generation. Coordinators accept the system because they can inspect the reason for a recommendation and override it. The company delays full automation of schedule release until it has stronger evidence and a tested incident process.

Learning points

1. The valuable innovation is a redesigned workflow, not the presence of AI alone.
2. Build–buy–partner decisions should follow strategic differentiation and risk.
3. Data quality and integration may matter more than model novelty.
4. Human oversight must include authority, information, and time to intervene.
5. Pilot metrics should include customer outcomes and harms, not only model accuracy.

10. Contemporary Developments and Strategic Implications

Several developments are shaping digital entrepreneurship.

Generative-AI commoditization is lowering the cost of basic content and coding capabilities. Competitive advantage is consequently shifting toward proprietary workflows, trusted distribution, specialized data, evaluation systems, and customer relationships.

Edge computing and connected devices allow data processing closer to where data are generated. This can reduce latency and bandwidth use but may complicate device management and security.

Digital public infrastructure and open standards can reduce barriers to identity, payments, and data exchange, while also creating new responsibilities concerning inclusion and governance.

Privacy-enhancing technologies, including encryption, secure computation, and controlled data environments, may enable useful analysis while reducing exposure, although they do not eliminate all risks.

Regulatory convergence and divergence are both occurring. General principles such as accountability, privacy, safety, and transparency are increasingly prominent, but obligations differ by country and sector. Ventures should obtain current legal advice rather than assume that a technology is lawful because a competitor uses it.

Sustainability is becoming a technology strategy issue. Digital services consume energy and require hardware, data centers, and network infrastructure. Entrepreneurs should assess the environmental cost of computation, storage, model training, and device replacement alongside financial cost.

Challenges and Limitations

Digital entrepreneurship can produce concentration of power, surveillance, exclusion, labor displacement, misinformation, cybersecurity harm, and environmental costs. Digital access is unequal, and algorithmic systems may reproduce social inequalities. Platform scale can make markets less contestable, while dependence on a few cloud or AI suppliers can create systemic vulnerability.

Entrepreneurs should also resist technological determinism. Technology does not independently produce innovation or social value. Outcomes depend on institutions, incentives, capabilities, user practices, regulation, and power relationships. A technically feasible product may be commercially weak, ethically unacceptable, or impossible to govern responsibly.

Best-Practice Checklist

A graduate entrepreneur evaluating a digital venture should ask:

• What specific customer problem is being solved?
• What part of the value proposition is genuinely digital?
• Which assumptions about adoption, data, and economics are untested?
• What should be built, bought, or obtained through partnership?
• Who owns and controls the data and intellectual property?
• What happens when the system is wrong, unavailable, manipulated, or misused?
• Can users understand, challenge, and correct important outputs?
• Are performance, safety, equity, and financial metrics monitored together?
• Can the venture migrate away from critical suppliers?
• Is there a documented process for incidents, updates, and retirement?

Chapter Summary

Digital entrepreneurship uses digital technologies as central mechanisms of opportunity creation, value delivery, or value capture. Digital ventures benefit from scalability, software-based replication, data accumulation, experimentation, and ecosystem participation, but they also face platform dependence, cybersecurity exposure, technical debt, regulatory obligations, and trust challenges.

Platforms coordinate participant groups and require deliberate choices about participation, matching, governance, and monetization. Cloud services and APIs accelerate experimentation but introduce supplier dependence and operational risk. Data products create value only when data are relevant, lawful, high quality, secure, and embedded in actionable workflows. Automation and AI can improve prediction, personalization, and productivity, while generative AI creates new capabilities and new risks such as hallucination, privacy leakage, bias, and prompt injection.

Build–buy–partner decisions should reflect strategic distinctiveness, speed, cost, control, learning, security, interoperability, and reversibility. AI governance should be risk-based and include clear ownership, data controls, validation, monitoring, meaningful human oversight, incident management, and retirement procedures. The strongest technology-enabled ventures integrate technology with customer insight, operational capability, trusted distribution, and responsible governance.

Key Terms

Digital entrepreneurship: Entrepreneurship in which digital technologies materially shape opportunity, value creation, delivery, or capture.

Digital transformation: Fundamental change to a business model or operating model through digital technologies.

Platform: A system that facilitates interactions between two or more participant groups.

Network effect: Change in a product’s value resulting from changes in participation.

Cloud computing: On-demand access to computing and digital infrastructure over a network.

API: A defined interface enabling software systems to exchange requests and responses.

Data product: A product whose central value depends on data, analysis, prediction, or information.

Data lineage: A record of data origin, transformation, movement, and use.

Automation: Performance of tasks with limited human intervention.

Artificial intelligence: Computational systems performing functions associated with perception, prediction, language, planning, or decision support.

Generative AI: AI that produces new content such as text, images, audio, or code.

Hallucination: An inaccurate or unsupported output presented by a generative system.

Build–buy–partner decision: A choice among internal development, external acquisition, and collaborative provision of a capability.

Customer acquisition cost: The attributable cost of acquiring a new customer.

Customer lifetime value: The expected discounted contribution from a customer over a defined horizon.

Precision: The proportion of predicted positive cases that are actually positive.

Recall: The proportion of actual positive cases that are correctly identified.

Algorithmic risk: The possibility that an algorithmic system causes unfair, unsafe, inaccurate, unlawful, or otherwise harmful outcomes.

AI governance: Policies, roles, controls, and processes for responsible development and use of AI.

Technical debt: Future cost created by expedient technical choices that require later correction or replacement.

Review Questions

Short Answer Questions

1. What distinguishes digitalization from digital transformation?
2. Explain one direct and one indirect network effect.
3. Why is data quality important to a data product?
4. State the difference between precision and recall.
5. Identify two risks associated with cloud-service dependence.

Long Answer Questions

1. Explain how digital technologies alter the economics and structure of entrepreneurial ventures.
2. Compare platform businesses with conventional linear businesses, including governance challenges.
3. Discuss the strategic criteria that should guide build–buy–partner decisions.
4. Explain the major stages of an AI system lifecycle and the controls appropriate at each stage.
5. Evaluate the claim that generative AI is a source of sustainable competitive advantage for startups.

Analytical / Application Questions

1. A marketplace startup has many buyers but few reliable sellers. Diagnose the cold-start problem and propose a staged platform strategy.
2. Design a data-governance plan for a venture using customer behavior data to personalize financial education. Address consent, minimization, security, access, retention, and user control.
3. A company proposes an AI system to rank job applicants. Develop a risk assessment covering data quality, bias, privacy, explainability, human oversight, monitoring, and the conditions under which deployment should be refused.

Multiple Choice Questions with Answers

Question 1

Which characteristic most directly distinguishes a digital product from a physical product?

A. It always has zero production cost.

B. It can often be replicated at low marginal cost.

C. It never requires customer support.

D. It is exempt from regulation.

Correct Answer: B. It can often be replicated at low marginal cost.

Question 2

An indirect network effect occurs when:

A. A user benefits from using a product privately.

B. A product becomes cheaper to manufacture.

C. Growth on one participant side increases value for another side.

D. A firm reduces its fixed costs.

Correct Answer: C. Growth on one participant side increases value for another side.

Question 3

Which capability is generally the strongest candidate to build internally?

A. A standardized payment gateway

B. A commodity email service

C. A strategically distinctive workflow embedded in the venture’s core value proposition

D. A generic cloud storage service

Correct Answer: C. A strategically distinctive workflow embedded in the venture’s core value proposition

Question 4

In the simplified formula CLV = m ÷ (1 − r), what does r represent?

A. The discount rate

B. The retention rate per period

C. The revenue growth rate

D. The error rate

Correct Answer: B. The retention rate per period

Question 5

If a classifier produces 80 true positives and 20 false positives, its precision is:

A. 20%

B. 25%

C. 75%

D. 80%

Correct Answer: D. 80%

Question 6

Which statement best describes data minimization?

A. Collecting every available data point for future use

B. Collecting only data necessary for a specified legitimate purpose

C. Deleting all data immediately after collection

D. Allowing unrestricted employee access to data

Correct Answer: B. Collecting only data necessary for a specified legitimate purpose

Question 7

Why can a technically accurate AI model still have low business value?

A. Accuracy is unrelated to customer outcomes, workflow integration, and cost.

B. Accurate models cannot be monitored.

C. Accuracy automatically creates regulatory violations.

D. Accurate models cannot operate in the cloud.

Correct Answer: A. Accuracy is unrelated to customer outcomes, workflow integration, and cost.

Question 8

Which is an example of meaningful human oversight?

A. A reviewer automatically approves every model output.

B. A reviewer is named but cannot access the evidence behind an output.

C. A qualified reviewer can inspect, question, and override a consequential output.

D. A user is told that the system is infallible.

Correct Answer: C. A qualified reviewer can inspect, question, and override a consequential output.

Question 9

What is a principal risk of dependence on a single cloud or AI supplier?

A. Guaranteed improvement in interoperability

B. Vendor lock-in and service-continuity vulnerability

C. Elimination of cybersecurity obligations

D. Permanent reduction in operating costs

Correct Answer: B. Vendor lock-in and service-continuity vulnerability

Question 10

In the CareRoute case, why was generative AI restricted to drafting messages?

A. The company had no customers.

B. Generative AI was used where it added assistance while high-impact scheduling remained constrained and reviewable.

C. Generative AI cannot process language.

D. Cloud services were unavailable.

Correct Answer: B. Generative AI was used where it added assistance while high-impact scheduling remained constrained and reviewable.