ARTIFICIAL INTELLIGENCE & MACHINE LEARNING

1. What do you understand by Artificial Intelligence?

…..a field which tries to build machines / software that can think like humans, entities that
can not just replicate the thought processes and reasoning of humans but also behave like
humans
……. Can exhibit the following components of intelligence, namely: learning, reasoning,
problem solving, perception and understanding natural language.
(Russell, S.J. and Norvig.P 2015).

It is simulated intelligence in machines designed to mimic or suprpass human action,

thought and behaviour
Includes:
- Natural Language Processing – Getting Computers to read and hear like humans
- Vision Systems – Getting computers to see like humans
- Robotics- Getting computers to do stuff like humans
- Machine Learning – Using data to get computers to learn like humans
- Cognitive Systems -Getting computers to solve problems like humans

2. Where is AI used in organizations today? Give examples.

Refer to the statistics given below ( source: https://hbr.org/2017/04/how-companies-are-
already-using-ai quoting statistics based on findings by TCS)

AI in HR
Ai can be used in HR across the entire talent life cycle:
- Attract- Specialized Chatbots deployed during candidate attraction offer candidates the
opportunity to ask questions that are interpreted and responded to using natural
language processing (NLP), allowing prospective applicants to learn more about the
organization before they actually apply. E.g . The AI solution IBM developed to address
this challenge is called Watson Candidate Assistant (WCA). As Carrie Altieri, Vice
 President HR Communications, IBM, put it, “IBM gets 7,000 resumes per day and
surfacing the right candidate in a reasonable time is like finding a needle in a haystack.
Since implementing WCA, we have dramatically cut time-to-hire, doubled NPS, and
vastly improved the matching of candidates to jobs.”
- Hire – AI is used to predict how long a job requisition will take to fill based on historical
data, allowing recruiters to reprioritize as needed. Also to determine the match between
a candidate’s resume and the job requisition, and to make accurate predictions of future
performance based on information about the candidate collected in the job application
process.
- Engage- Ai is used for engagement analysis i.e analyze unstructured content from
annual surveys and pulse surveys, as well as social media chatter.
- Retain - AI-based compensation support, machine learning aided decision advice that
examines many more variables than were previously considered.
- Develop - AI tagging of learning content/ resources such as images and documents that
are uploaded to learning management systems are enriched with metadata through AI
technologies, and this helps learners locate the training content and use it again and
again more efficiently.
- Grow- Career coaching is a powerful way to create more meaningful work experiences
for employees and can lead to greater productivity and stronger succession pipelines for
businesses.
- Serve- Chatbots are a perfect example of where AI can improve the employee
experience, because they provide real-time answers at any time, day or night. Chatbots
also constantly learn from feedback, and improve the answers given.

AI in SCM :
- Chatbots for Operational Procurement
- Machine Learning (ML) for Supply Chain Planning (SCP) for forecasting and optimizing
supply chain decision-making.
- Machine Learning for Warehouse Management- ML provides an endless loop of
forecasting, which bears a constantly self-improving output. This kind of capabilities
could reshape warehouse management as we know today. “A forecasting engine with
machine learning, just keeps looking to see which combinations of algorithms and data
streams have the most predictive power for the different forecasting hierarchies”
(forbes.com 2017).
- Autonomous Vehicles for Logistics and Shipping
- Natural Language Processing (NLP) for Data Cleansing and Building Data Robustness
- ML and Predictive Analytics for Supplier Selection and Supplier Relationship
Management (SRM)

AI in Marketing
 Chatbots for Customer Service
 Recommendations/ Content Curation : Recommendation is one of the best
examples of AI in marketing. E-commerce websites, blogs, and many social
networking and media websites use artificial intelligence to analyse customer
activities on the internet and recommend products and content for better
conversions and maximized engagement
  Dynamic Pricing: Dynamic pricing (often referred to as personalized pricing) is a
pricing strategy where the price is determined depending upon the demand, the
availability and the profile of the customer. AI technologies are even used to decide
and design personalized offers based on customer profiles.
 Ad Targeting
 Machine Learning for Segmentation and Classification
 Forecasting
 Content Generation: Many tools, like wordsmith, are used by marketers to convert
data into sensible text. This auto content generation technology involves artificial
intelligence and saves their lot of time. The best part about this AI-generated
content is that readers are not able to discern automated content from content
written by a human. According to Gartner “By 2018, 20% of all business content will
be authored by machines.”
 Searches
 Business Forecasting: One of the challenges marketers face is how to predict the
exact future of the business or a product using a complex set of inputs under tight
time constraints. Many machine learning processes have been designed to
automatically learn patterns in the numerous data inputs and help marketers to
predict the future of anything they want to.

3. Name examples of where we encounter AI in our day-to-day Life

Siri, Cortana, Alexa,… Virtual Assistants, Driverless Cars, Drones, Ecommerce Bots,
Intelligent agents, Chatbots, Voice Recognition systems, Image recognition systems,
Navigation systems, Travel, Games, cognitive RPA, Recommendation Systems on Amazon,
Netflix, Social Media Networks…

4. What is Machine Learning?

 An application of artificial intelligence (AI) that provides systems the ability to
automatically learn and improve from experience without being explicitly
programmed.
 Machine Learning (ML) encompasses a lot of things. The field is vast and is
expanding rapidly. It is a branch of Artificial Intelligence (AI). Loosely speaking, ML is
the field of study that gives computer algorithms the ability to learn without being
explicitly programmed. The outcome we want from our computer algorithm is
PREDICTION. This is different from our previous problems where we wanted the
algorithm to solve a specific problem such as finding the best web page for our
search, sorting a list of items, or generating a secure means to computing a shared
secret in cryptography. What are we going to use to predict

5. How does Machine Learning differ from Artificial Intelligence?

 Machine Learning is a subset of AI
 It is what gives the machines the ability to learn by itself

6. Explain the difference between Supervised and Unsupervised Machine Learning.

 Supervised learning: discover patterns in the data that relate data attributes with a
target (class) attribute. These patterns are then utilized to predict the values of the
target attribute in future data instances. These techniques apply learning from the past
to predict. They start from analysis of a known training dataset, predicted output is
compared with the intended output and to improve the algorithm.
 Examples, Fraud/ Non Fraud transactions, Inventory Mgmt, Sentiment Analysis,
Credit Scoring
 Techniques : Classification, Regression, Decision Trees, Logistics

Unsupervised learning: The data have no target attribute. We want to explore the data to
find some intrinsic structures in them. The system explores the data and can draw nferences
from datasets to describe hidden structures from unlabeled data.
• Techniques- Clustering – Hierarchical, K Means, 2-Step Clustering, Association Rule

7. Give some practical applications of Supervised learning techniques

 Regression: Financial forecasting, trend analysis, marketing, time series prediction ,
drug response modeling fraud detection, credit scoring, clinical trials
 Decision tree: Customer Churn Analysis, Energy Consumption Patterns, , Fraudulent
Practice Analysis, Sentiment Analysis, Investment Solutions
 Naive Bayes Classification: Text analysis, email spam detection, sentiment analysis,
medical diagnosis
 Artificial Neural networks: Image processing, text classification, information
extraction, semantic parsing, question answering, paraphrase detection, language
generation, multi-document summarization, machine translation, and speech and
character recognition

8. Give some practical applications of Unsupervised learning techniques

Market research, pattern recognition, data analysis, image processing, document

identification, Identifying location of retail/delivery stores, cyber profiling, call record
analysis, grouping of cancer patients grouped by their gene expression measurements,
grouping shoppers based on their browsing and purchasing histories,

9. What is reinforcement learning? Name some practical applications of Reinforcement

Learning.

Reinforcement learning (RL) is an area of machine learning, concerned with how software
agents ought to take actions in an environment so as to maximize some notion of cumulative
reward.
Applications
• Resources management in computer clusters: Designing algorithms to automatically
allocate and schedule computer resources to waiting jobs, with the objective to minimize
the average job slowdown
• Traffic Light Control
• Robotics
• Chemistry : optimizing chemical reactions.
• Personalized Recommendations:
• Bidding and Advertising. Alibaba Group has shown how Multi Agent Reinforcement Learning
can be used in multi-agentbidding solution (DCMAB). Example: Taobao ad platform is a
place for merchants to place a bid in order to display ad to the customers.

10. Do you think that the hype regarding AI is justified?

AI has the potential to improve business operations and decision making. However, as per
the Gartner’s Hype Cycle, most of the AI solutions are still in the stage of Innovation Trigger,
where there is a lot of media interest but commercial viability is unproven. While most
innovative products are still in development and commercial testing stages, the hype around
AI is more because of the perceived potential.
The Gartner’s Hype Cycle categorises products/ innovations into the following stages:
 Innovation Trigger: Potential technology breakthrough, media interest, no usable
products exist and commercial viability unproven
 Peak of Inflated Expectation: Number of success stories, also scores of failures.
Some companies take action, many do not
 Trough of Disillusionment: Interest wanes as experiments and implementations fail,
investments continue only if surviving providers improve their products to satisfy
early adopters
 Slope of Enlightenment : More instances of how it can be used begin to crystallize,
second and third generation products appear. Enterprises fund pilot projects
 Plateau of Productivity : Mainstream adoption starts to take off , criteria to assess
provider viability more clearly defined Technology’s broad market applicability and
relevance clearly paying off

You can refer to the Gartner’s Hype Cycle for 2019 given below:
 Additional Sources:
https://www.gartner.com/smarterwithgartner/5-trends-appear-on-the-gartner-hype-cycle-
for-emerging-technologies-2019/

11. Are machines capable of replacing humans?

The biggest fear of mankind is that machines will soon replace humans. While certain areas
of repetitive tasks have seen replacement of men by machines ( eg. Chatbots, RPA, driverless
less systems), as of today, machine donot have appear capable of replacing humans. At
present, AI is in the stage of Artificial Narrow Intelligence. While we appear to be moving
towards Artificial General Intelligence, it is still much further away.
12. What is RPA and how is it different from AI?
Traditional RPA is automation of tasks that were otherwise carried out by humans. It does
not involve learning or intelligence, unlike AI. RPA’s were designed to take the robot out of
humans and create a digital workforce. However, over the years, RPA’s have evolved to
include cognitive abilities, thus bringing Intelligent RPAs to the forefront.While still in the
initial phases, AI is already making strides with RPA tools. This is leading to
the emergence of Cognitive Robotic Process Automation(CRPA) software
bots.
13. Name some tools for RPA.
Blue Prism, UiPath, Automation Anywhere, Pega systems, Kyron, Jacada, WinAutomation,
NICE Systems.

14. Name some RPA applications in your functional area

Some of the Popular RPA Applications include:
RPA in Banking, Finance & Accounting - Supplier onboarding, Handling , Price comparisons,
Validating Contract terms, Supplier pricing comparisons, Accounts receivable and payable
reconciliation, Customer onboarding and master data maintenance, Account Opening and
Closing, Data extraction for accounting, Aggregating and analyzing financial and operational
performance, KYC, Financial Review Preparation, Account Reconciliation, Standard Journal
entries, Regulatory Compliance and Reporting, Operational Finance and Accounting,
Onboarding New Customers etc

RPA in Logistics and Supply Chain Management – Order processing and Payment, Email
Automation, Inventory Management Automation, Vendor Selection, Shipment Status
Communication, Supply and Demand Planning, Automating Requests for Proposal and
Quotes, Supply chain managers use robotic process automation to create, update and
manage contract and other data for better supplier management, RPA to research the
supplier and manufacturer markets using consistent criteria, allowing for easier supplier
selection, Invoice processing
 RPA in Marketing – Marketing Operations, Customer Engagement using Chatbots,
Streamlining Digital Ad Placement, Clickstream Analysis, Integrating data from multiple
touch points, Improving customer experience, Sales Analytics, Store Planning

RPA in HR – Candidate Sourcing and Screening, Recruitment Offers, Onboarding and

Offboarding, Compensation and Payroll, Expense and Reimbursement Processing,Data Entry
upload and verification, Reporting, Recruitment Offers

15. What is Feature Extraction?

The process for selecting the variables for a model is called Feature
Extraction. It is also known as Feature Engineering. If data is stored in a
table, a feature is a column of data. Good Feature extraction is essential
for a Machine Learning Model. It basically means identifying the important
attributes or variables that describe the data with sufficient accuracy.
Some methods of Feature Extraction are Factor Analysis, Principal
Component Analysis, Isomap, Partial Least Square.

16. What can Machine Learning be used for?

Identifying errors, Predictive Maintenance, Understanding Customers Preference, Credit
Scoring, Sentiment Analysis, Customer Service, Fraud Prediction, Spam Detection,
Recruitment, Social Media sites for networking .. and many more

17. What is Deep Learning?

Deep learning is a subfield of machine learning. This type of system allows
for processing huge amounts of data to find relationships and patterns
that humans are often unable to detect. The word “deep” refers to the
number of hidden layers in the neural network, which provide much of the
power to learn.

18. How does Deep learning differ from Machine Learning?

Deep Learning and Machine Learning are often confused, they do share
many similarities.
Machine Learning needs labelling for image recognition, it does not just
look at the pixels of the image and find pattern. A process of feature
extraction is required ( ie. Defining beak shape, eyes, etc). This works in
most cases but does not work when features are off the marks. Feature
extraction also has the drawback of ignoring a large amount of the data.
Deep Learning, instead of feature extraction, analyses all the data pixel by
pixel and then finds relationships by using a neural network.

19. Where is Deep Learning Used?

Deep Learning finds extensive application in the field of Medicine, Radiology, Voice and
Image Recognition, By Oil and Gas Industry for locating sources of oil and optimizing
extraction plans, Construction company Bechtel Corp. uses DL for optimizing construction
planning to find the fastest route to build projects using step-by-step simulations of projects,
testing out sequences of installing pipe laying concrete to find the optimal sequence.
Among Financial organization, large investment houses like JPMorgan Chase are using deep
learning based text analytics for insider trading detection and government regulatory
compliance. Deep learning also has a number of use cases in the cybersecurity space. Use
 cases include automating intrusion detection with an exceptional discovery rate. Deep
learning also performs well with malware, as well as malicious URL and code detection.

Some widely publicized Deep Learning applications include:

 Speech recognition used by Amazon Alexa, Google, Apple Siri, or Microsoft Cortana.
 Image recognition used for analyzing documents and pictures residing on large
databases.
 Natural Language Processing (NLP) used for negative sampling, sentiment analysis,
machine translation, or contextual entity linking.
 Automated drug discovery and toxicology used for drug design and development work,
as well as for predictive diagnosis of diseases.
 CRM activities used for automated marketing practices.
 Recommendation engines used in a variety of applications.
 Predictions in gene ontology and gene-function relationships.
 Health predictions based on data collected from wearables and EMRs.
(source: https://www.dataversity.net/deep-learning-use-cases/)

20. What are some of the ethical challenges of AI?

https://www.weforum.org/agenda/2016/10/top-10-ethical-issues-in-artificial-intelligence/
https://www.cognilytica.com/2018/07/24/ethical-concerns-of-ai/