The final assessment, grounded in the modules

01M01

Rules-based AI has largely been replaced by…

Probabilistic AI models

Modern AI typically learns weighted patterns from data instead of relying only on fixed “if this, then that” rules.

Module 1 · Rules to probability

02M01

Which statement best describes General AI?

AI that can perform well on many problems, just like humans

Artificial General Intelligence refers to broad, human-like capability across many different tasks rather than excellence in one narrow domain.

Module 1 · AI types

03M01

Which is not a type of machine learning: supervised, unsupervised, or semi-supervised?

They are all types of machine learning

The three choices describe established learning approaches distinguished by how much labeled outcome data is available.

Module 1 · Machine learning types

04M01

Which statement about machine learning is true?

All of the above

Deep learning is a subset of machine learning; ML can learn from examples with correct answers; and generative AI is an application of ML, not the category that contains all ML.

Module 1 · Machine learning

05M01

Which statement is true of generative AI such as ChatGPT?

It can create new content on a subject you suggest

Generative models produce new text, images, audio, code, or other content in response to instructions and context.

Module 1 · Generative AI

06M02

Which is an example of work data scientists perform?

Feature analysis

Data scientists examine possible input features to determine which variables carry useful predictive signal for a model.

Module 2 · Data science

07M02

Which is an example of outcome data for machine learning?

Both human judgment and objective metrics

An outcome can be established through human agreement, such as sentiment, or through an objective measure, such as a completed sale.

Module 2 · ML fit

08M02

What is the best process for developing AI capabilities?

Agile

AI development requires iterative testing, learning, and adjustment, making an agile approach better suited than a fixed waterfall sequence.

Module 2 · Development process

09M02

What do we call the correct answers used to train AI models?

Gold data

Gold data is a trusted set of verified answers used to train a model and test how accurately it performs on held-out examples.

Module 2 · Measuring accuracy

10M02

Which is a quality of Big Data: velocity, veracity, or variety?

All of the above

The classic four qualities are Volume, Velocity, Variety, and Veracity, with veracity emphasizing trustworthiness.

Module 2 · The four Vs

11M03

Which was not a historical theme in process optimization?

Globalization of production networks

The module’s progression centers on production efficiency, continual quality and process improvement, and customer or user experience.

Module 3 · Optimization lineage

12M03

For an accounts-payable team studying automation opportunities, which research technique is least fruitful at the discovery stage?

Demo a new solution

The team should first understand the real workflow through interviews, cycle-time analysis, and process mining before evaluating a particular solution.

Module 3 · Identify before selecting

13M03

True or false: AI-driven process optimization does not require human oversight.

False

AI systems meet messy real-world conditions, so organizations should monitor performance, retain human judgment, and design intervention paths.

Module 3 · Human-in-the-loop

14M03

What technology extracts text from scanned PDFs, faxes, and images?

Optical Character Recognition (OCR)

OCR converts visible characters in image-based documents into machine-readable text that software can search, validate, and process.

Module 3 · Task automation

15M03

What is Deming’s four-step model for continually improving business processes?

Plan, Do, Check, Act

PDCA is a repeatable cycle: plan a change, test it, study the result, and act on what was learned before repeating.

Module 3 · PDCA

16M05

Which statement best describes the state of AI in sales and marketing?

Many companies use it, with substantial opportunities still ahead

Marketing and sales is a leading AI application, but underused tools, fragmented data, and uneven customer experiences leave significant room for innovation.

Module 5 · Current adoption

17M05

Why does AI in sales and marketing matter?

It helps companies create customers and acquire them at lower cost

AI supports the enduring purpose of marketing: finding, winning, and retaining customers while improving the efficiency of existing efforts.

Module 5 · Create a customer

18M05

What is usually the biggest challenge in putting AI to work for sales and marketing?

Getting your data into systems that can put it to work

First-party data is the fuel and differentiator, but integration, quality, ownership, and permission are often harder than the model itself.

Module 5 · Data is the Crown Jewels

19M05

How can AI improve sales and marketing?

All of the above

AI can identify target segments, create content and copy for those segments, and reduce customer-acquisition cost through better prediction and optimization.

Module 5 · Sales and marketing use cases

20M05

What is typically the fastest way to integrate AI into sales or marketing?

Audit and test capabilities in current automation tools

Many existing marketing and sales platforms already contain underused AI capabilities, making an audit faster than building or procuring from scratch.

Module 5 · Adoption checklist

21M06

What rough ratio do customers use when selecting service channels?

Roughly equal across self-service, remote human, and traditional

Customers move among all three channel types, and many who begin with a chatbot still need a well-designed handoff to a human agent.

Module 6 · Channel complexity

22M06

How do you protect the customer experience when AI cannot meet a customer’s needs?

Fail gracefully and shift the experience to the right channel

A clean escalation preserves context and trust by moving the customer to a human or another appropriate channel without forcing a restart.

Module 6 · Be useful

23M06

Who sets the standard for what customers expect from service?

Customers, based on the best experiences in their lives

Expectations travel across industries; customers compare each interaction with the easiest and most useful experiences they receive anywhere.

Module 6 · Customer expectations

24M06

What framework matters most when connecting with customers in practice?

Be there, be useful, be quick

Effective service is reachable, actually resolves the problem, and respects the customer’s time.

Module 6 · Service mantra

25M06

Where can you get data to train AI for customer-service experiences?

All of the above

Site search, social listening, and website or app behavior all reveal customer intent, friction, language, and recurring service needs.

Module 6 · Customer experience data

26M09

Which are real worries about AI: bias, job displacement, or intelligent weapons?

All of the above

Each concern creates real ethical, social, or governance questions that leaders must evaluate rather than treating AI as purely technical.

Module 9 · Ethics and responsibility

27M07

Which is not a finance function?

Sales Operations

The module’s finance functions include FP&A, Accounting and Reporting, Treasury, Risk and Governance, and Investor Relations.

Module 7 · Finance functions

28M07

Which is not a good problem statement for a new technology?

“Our sales platform is better than anything in use today…”

That sentence promotes and compares a proposed solution; it does not describe the customer’s underlying pain, constraint, or measurable problem.

Module 7 · Problem to solution framing

29M07

Which predictive-analytics processes can benefit from AI?

All of the above

Sales and demand forecasting, budgeting and resource allocation, and market-trend analysis all benefit from models that identify patterns and project outcomes.

Module 7 · Predictive analytics

30M09

What work of fiction predicted intelligent machines but missed the timing by about 80 years?

The Wreck of a World

The course’s AI-history timeline begins with this 1889 dystopian-fiction prediction of intelligent machines.

Module 9 · AI history timeline

31M04

In which ways will AI impact the future of work?

All of the above

AI will create new professions, transform existing roles and activities, and eliminate some tasks and roles while shifting demand toward new skills.

Module 4 · Future of work

32M04

Which is not a new employer expectation arising from AI collaboration?

Uniform use of the same AI tools in the same way

The course anticipates higher productivity, personalized workflows, and a shift toward outcomes—not identical tool use across different people and roles.

Module 4 · AI as coworker

33M04

In which way will AI not support us as a “new coworker”?

As an Ally that replaces human intelligence and people skills

The four roles are Assistant, Amplifier, Automator, and Advisor; they augment human work rather than replace the human capabilities that matter most.

Module 4 · Four roles of AI

34M04

Which skillsets matter for success in an AI-accelerated world?

All of the above

Workers need traditional, generative, and agentic AI skills alongside critical human skills and the learning agility to keep adapting.

Module 4 · Skills for the future

35M04

What is the best practice when integrating AI into one’s work?

Identify tasks, then choose the right AI support for each

Start with the work rather than the tool: audit job tasks, decide the role AI should play, and select a capability matched to the need.

Module 4 · Task-to-role mapping

36M09

How has the United States government’s approach to AI development been described?

Free-market oriented and primarily regulatory

The module contrasts the U.S. regulatory posture with China’s authoritarian model and the European Union’s fine-driven approach.

Module 9 · Government postures

37M04

Which is considered a perceived risk of adopting AI?

Loss of jobs

Job displacement is a widely perceived risk, even though the course also projects substantial job creation and emphasizes transformation and reskilling.

Module 4 · Workforce impact

38M09

Engineering teams are less likely to develop biased AI when members vary in what ways?

All of the above

Diversity in genders, demographics, and training helps a team challenge assumptions and detect errors that homogeneous groups may overlook.

Module 9 · Diversity and bias

39M09

How can a company reduce bias in autonomous AI systems?

Continuously evaluate programming and AI decisions

Ongoing monitoring, proxy review, diverse oversight, and feedback loops help detect and correct bias throughout the system lifecycle.

Module 9 · Bias controls

40M09

Which is not an ethical issue codified in current law: disability rights, minimum wage, or environmental protection?

None of the above

All three represent ethical concerns that societies have translated into legal rights, protections, duties, or standards.

Module 9 · Ethics and law

41M02

True or false: Machine learning is a black box that cannot accept feedback after creation.

False

Models can be evaluated, monitored, retrained, and improved with new data and human feedback; AI should never be treated as “set and forget.”

Module 2 · Measure and improve

42M10

Using ChatGPT or Gemini to evaluate a business use case belongs to which category?

Exploration

Standalone tools are a low-cost, low-risk way to test value and learn before committing to deeper integration or proprietary development.

Module 10 · AI sourcing spectrum

43M02

True or false: Business teams are unnecessary because building ML models is purely technical.

False

Useful models require domain knowledge, process context, outcome definitions, and stakeholder judgment alongside mathematics and computing.

Module 2 · Data science and domain knowledge

44M01

Is fixed account eligibility based on location, education, and birth year necessarily an ML problem?

No

When eligibility follows explicit, stable criteria, a deterministic rules engine is usually more transparent and appropriate than machine learning.

Module 1 · Rules-based vs probabilistic

45M10

Is build versus buy an important long-term AI strategy decision?

Yes

The choice shapes cost, differentiation, skills, maintenance, risk, ownership, speed, and the organization’s ability to scale.

Module 10 · Build and buy spectrum

46M08

Which company was not mentioned as a use case in the “ROI of AI” module?

Coca-Cola

Golden Chick, Carhartt, and Aerotech appear in the module’s case material; Coca-Cola is not part of that reviewed case set.

Module 8 · Case studies

47M08

What percentage represents the current ROI success rate for AI initiatives?

25%

The module reports that only about one-quarter of AI initiatives deliver ROI, while an even smaller share successfully scales.

Module 8 · State of AI ROI

48M08

Which is not an AI initiative goal?

Create new tasks

AI initiatives should connect to business value through new revenue, lower costs, or saved time—not create work for its own sake.

Module 8 · Goal categories

49M08

What is a key reason for the current state of AI ROI?

Decision makers rush in because they fear being left behind

FOMO leads organizations to start with technology instead of a defined problem, aligned team, reliable data, and measurable business outcome.

Module 8 · ROI-killing mistakes

50M08

What is the first question when selecting the right AI tool for a task?

What is the pain point?

Tool selection begins with the problem, followed by the workflow, data and security needs, a pilot, and total cost.

Module 8 · Tool-selection sequence