18 - From data-driven to AI-driven Flashcards
What is the final stage of data transformation in a business?
To become a leader and innovator, capable of systematically outperforming competitors.
How does artificial intelligence improve decision-making?
AI can improve decision-making beyond the capabilities of the people in the business.
What does the data team focus on in the final stage of transformation?
Providing data products that the business uses to thrive.
What are the three activities in the VBI model for this wave?
- Value
- Build
- Improve
What does the ‘Value’ aspect of the VBI model focus on?
Expressing the value of data in innovation and creativity.
What is emphasized in the ‘Build’ aspect of the VBI model?
Applying AI to strategic, future-facing processes.
What is the key focus of the ‘Improve’ aspect in the VBI model?
Creating mechanisms for innovation to become business as usual.
Fill in the blank: Artificial intelligence may create profound insights from data and exceed the limitations and bias of the _______.
[human brain]
What is required for AI applications to be effective?
Strong foundations, embedded in business processes, with a clear business case.
What is the definition of artificial intelligence (AI)?
The ability of a digital computer to perform tasks commonly associated with intelligent beings.
What is machine learning (ML)?
An AI discipline that implements computer software that can learn autonomously.
What does a single supplier view (SSV) entail?
Collecting data from disparate sources to form a single, accurate record for each supplier.
What does the Gartner Hype Cycle illustrate?
The evolution of technology perception through phases from innovation trigger to plateau of productivity.
True or False: AI can solve problems without the need for good underlying data quality.
False
What are two intractable problems that AI can solve for a data-driven business?
- Speed
- Complexity
What is forecasting in the context of data?
Estimating future events based on past and present data through trend analysis.
What distinguishes predictive modelling from forecasting?
Predictive modelling estimates the likelihood of specific outcomes, while forecasting gives a general judgement.
What are the two types of AI applications used in policing?
- Location-based algorithms
- Predictive tools focusing on individuals
What is the importance of a clear business case in AI implementation?
It aligns with the data and analytics strategy, ensuring effective AI application.
What does the AI maturity model illustrate?
The journey in AI implementation from basic decision support processes to complex decision-making.
Fill in the blank: The chief data officer becomes the focus of future _______.
[competitive advantage]
What is the role of AI in decision support?
To improve the options and alternatives available to human decision-makers.
What is the significance of a ‘single customer view’ (SCV) mentioned in relation to SSV?
It serves as a model for creating a unified view of suppliers for better data management.
What is a major challenge for businesses using AI?
Navigating the balance between technology capabilities and the quality of data.
What is anomaly detection?
The identification of unusual patterns that do not conform to expected behavior.
What is the purpose of fraud detection?
To identify and prevent fraudulent activities within a system.
What are examples of process automation?
- Automation of approvals
- Automation of claims
What role do chatbots play in business?
They automate customer service interactions and provide information to users.
What is robotics in the context of business?
The use of automated machines to perform tasks traditionally done by humans.
What is clustering in customer analysis?
Grouping customers based on shared characteristics or behaviors.
What is sentiment analysis?
The use of AI to determine the emotional tone behind words.
What is the goal of risk management?
To identify, assess, and prioritize risks followed by coordinated efforts to minimize, monitor, and control the probability or impact of unfortunate events.
What technologies are used in facial recognition?
AI algorithms that analyze facial features and patterns.
What is voice recognition?
The ability of a machine to identify and process human voice input.
What does predictive maintenance entail?
Using data analysis tools and techniques to detect anomalies in equipment operation and potential defects.
What is guided decision-making?
The process of using data-driven insights to inform business decisions.
Fill in the blank: Robotic Process Automation (RPA) is a form of business process automation that uses ________ to learn and complete data processing applications.
[AI]
What is the significance of data governance in fraud detection?
It ensures data quality and helps expose low-level fraudulent behavior.
What percentage of frauds benefited from weaknesses in internal controls, according to a 2016 survey by KPMG?
61 percent
True or False: AI in fraud detection attempts to remediate problems and provide explanations.
False
What is the main function of customer service bots?
To assist customers by answering questions and providing information.
What is Natural Language Processing (NLP)?
A field of AI that helps computers recognize the meaning of human language.
What was the challenge addressed by the Billy Bot case study?
Finding good barristers’ clerks.
What was the outcome of implementing Billy Bot?
A permanent increase in productivity, saving 200 hours of work a month.
What do meaningful clusters in customer segmentation allow businesses to do?
Target sales and marketing efforts more effectively.
What is a significant risk of using biased data in AI applications?
The AI may perpetuate or exacerbate existing biases.
Fill in the blank: An AI application knows only about the data it is ________ to.
[exposed]
What is statistical parity in measuring bias in machine learning?
A test to evaluate if two groups have the same probability of a positive outcome despite differing in a protected attribute.
What is the effect of removing bias from AI algorithms?
It may make the algorithm less profitable.
What is the primary task of AI applications that support complex analysis and prediction?
To provide insights and forecasts that the business could not easily find otherwise.
What is one way to validate AI applications that could have been accomplished by skilled staff?
By conducting a human check using a sample of outputs.
What is positive sentiment in AI applications?
The process of assigning positive sentiment to words and phrases in messages similar to previously identified positive messages.
This involves using examples and human checks to validate sentiment analysis.
What benefits do cloud-based services provide for AI applications?
They aggregate data from multiple sources, enabling the application to learn from a broader range of experiences and build trust through demonstrated success.
This is particularly useful in areas like sentiment analysis and attack detection.
What is the ‘black box’ problem in AI?
A situation where the internal workings of an AI system are not transparent, making it difficult to understand how inputs lead to outputs.
This is especially problematic in regulated areas where decision-making transparency is required.
What are the four ‘V’s of big data?
- Variety
- Volume
- Velocity
- Veracity
These components characterize the complexity and challenges of processing big data.
Define ‘big data’.
Data that is high volume, with lots of variety (structured and unstructured) and high velocity, making it complex and hard to process.
This complexity can lead to challenges in data analysis and decision-making.
What is an ‘explanation by design’ in AI?
A method where the data team documents the data fed into the AI, the assumptions made, and the logic behind outcomes to build trust in the AI’s decisions.
It helps convey the reasoning behind decisions without revealing the full decision-making process.
What are common issues that can lead to AI disappointment?
- Missing data
- Data leakage
- Data quality issues
These problems can hinder the AI’s ability to generate valuable insights.
True or False: AI can fully replace human decision-making.
False.
Human interpretation and context are essential in decision-making, especially when AI guidance is poor.
What two types of trust are necessary for AI to be effectively integrated into decision-making?
- Employees must trust AI decisions
- Customers must trust AI protections and fairness
Building this trust is crucial when using AI for recommendations or decision-making.
How can bias in training data affect AI applications?
It can result in skewed outcomes that do not represent all demographics accurately, leading to unfair or unsafe decisions.
An example is car safety features designed primarily based on male crash test dummies.
What is an unexpected value in AI recommendations?
The ability of AI to provide insights or recommendations that are not obvious or standard, potentially leading to more valuable business outcomes.
This contrasts with safe but uncreative recommendations that may not engage users.
What should be considered when aligning AI with data and analytics strategy?
- Real-time data needs
- Impact on customer relationships
These considerations help ensure that AI is used effectively and adds competitive advantage.
What characterizes a data-driven model?
Ensuring that everyone in the organization has the best data in the right format when making decisions.
However, this does not address issues like information overload or bias.
What is essential for AI to transition from a simple tool to a creative catalyst?
AI must be allowed to innovate processes for the business, potentially with little human involvement.
This requires careful management of expectations and capabilities.
What is the main takeaway regarding AI and data quality?
AI cannot compensate for poor data engineering and requires careful introduction and management.
Utilizing external cloud-based applications can enhance learning about AI capabilities.
