Abstract

Investment in education and training programs that give people the skills and knowledge

required to work with AI and other emerging technologies is crucial if we are to realize the full

promise of AI. Over the last decade, artificial intelligence (AI) has made great strides. Machine

learning has come a long way recently, especially in the areas of deep learning and reinforcement

learning.[1] This report presents the evolution of research on AI in business research over time.

Formal studies on public perceptions of its effects, with most available opinions, in this study,

we pursued a secondary approach to gather data and scrutinized 20 pertinent articles sourced

from available reservoirs.

Keywords:

AI (Artificial Intelligence), Data Collection, Data Mining, Market-Basket Analysis, Machine

Learning, Big Data Analysis, IoT (Internet of Things), Deep Learning, Predictive Analytics,

Machine Learning Algorithms, Cognitive Computing, Human-Like Decision Making.

1. Introduction

1.1. Background

“If there is a chance to improve and upscale the quality and efficiency of systems and operations,

why wouldn’t everyone take the opportunity?” was raised by a Nikhil Kulkami, a recent

graduate. He is very much the next generation that will have much to say in how the workplace

and society will change. This is the scope of where we are planning to head, especially with the

drive to implement more intelligent solutions for the infrastructures available in today’s society;

this drive is being achieved through integrating artificial intelligence (AI), which is the goal of

allowing ‘computer systems to exhibit intelligence in some manner. Artificial intelligence has

1
progressed from the development standpoint in computer science to being fully used to extract,

analyze and predict vast amounts of data in varying industries. The integration of technologies

like AI is helping improve the operation and efficiency of several infrastructures, as well as

reducing their financial costs.[2]

1.2. Objective

The objective of this research is to explore how artificial intelligence (AI) is influencing and

transforming business research methodologies.

2. Literature Review

2.1. Historical Perspective

The background of Artificial intelligence (AI) may lie in ancient cultures of Greek (e.g., the

mythological robot Talos), Chinese (e.g., Yueying Huang’ dogs) and other mythologies (Nahodil

& Vitku, 2013) [3], where automatons were believed to be imbued with real minds, capable of

wisdom and emotion. Yet, the term emerged in a workshop at Dartmouth College (United States)

in 1956 (Nilsson, 2010) [4], which is labelled as the birth of AI.

Since then, research on AI has passed from different fields of knowledge. Social scientists have

been discussing ethical and legal implications of AI (Cath, 2018)[5], computer scientists have

developed advanced deep learning algorithms (LeCun, Bengio, & Hinton, 2015)[6], while

researchers in business management have studied the impacts of AI on customers, firms, and

stakeholders in an increasingly automated and interrelated business world (Huang & Rust, 2018)
[7]
. Similarly, a unique and consensual definition of AI has been hard to get. Lately In recent

times, (Russell & Norvig, 2016)[8], summarize the various definitions of AI systems into four

2
categories along two dimensions: reasoning–behavior dimension and human performance–

rationality dimension.

These are:

1) Systems that think like humans, 3) Systems that think rationally, and

2) Systems that act like humans, 4) Systems that act rationally

Notably, several researchers have attempted to conduct a comprehensive literature review on the

use of AI in business research. For example, Côrte-Real, Ruivo, and Oliveira (2014) [9], perform a

systematic mapping of the diffusion stages of business intelligence and analytics (BI&A)

implementation, proposing a future research in the then rather neglected post-adoption stages.

Moro, Cortez and Rita (2015) [10], conduct a literature analysis between 2002 and 2013 focused in

Business Intelligence (which uses some AI algorithms for predictive analysis) in Banking. Tkáč

and Verner (2016) [11], review two decades of research on the application of artificial neural

network in business and found most of the examined articles discussing expert systems with

applications. Finally, Duan, Edwards, and Dwivedi (2019) [12], analyze relevant articles published

in International Journal of Information Management to identify issues and challenges around AI

for decision making in the era of big data, proposing theoretical development and AI

implementation. These efforts present useful knowledge about the advancements in AI and

business research.

2.2. Current State of AI in Business Research:

AI refers to the ability of a digital computer to perform tasks that are commonly linked to

humans, such as the ability to reason and generalize, discover meaning, or learn from past

experience. In other words, AI is a machine that is capable to learn and imitate or simulate the

3
intelligent behavior of humans. Meanwhile, machine learning is a branch of AI that applies

algorithms to synthesize the underlying relationships of data and information. The machine

learning system can be used to convert automatic speech into a semantic structure expressed in

the form of words. In addition, machine learning improves the efficiency of marketing functions

in every step taken by the customers. To make prediction of customer behavior, supervised

learning is needed to provide a learning basis for future data processing. Supervised learning is a

learning model built to make prediction, given unforeseen input instances (Kotu and

Deshpande,2019). It has algorithm to response to the dataset and make classification model to

generate future data processing. Therefore, with the help of machine learning and supervised

learning, AI can provide valuable data to predict their behaviors in real-time process using

automation. Although the terms “AI”, “automation”, and even “robotics” are interchangeably

used, these terms are different. In particular, AI mostly uses algorithms to learn a process and

involves logical reasoning, learning, and problem solving whereas automation and robotics use

sensors and manual programming (Oswald and Mascarenhas,2018). The main purpose of AI is to

develop software to imitate a human mind just like how humans handle general problem solving,

learning, and decision making in specific ways through an expert system and computer vision.

The role of AI in increasing the efficiency of companies was explored in several past studies. For

instance, López and Casillas (2013) explored the potential of AI-based systems in the marketing

context whereas Caoet al. (2015) highlighted the potential of AI in bringing automated

negotiation for e-commerce. On the other hand, Vanneschiet al. (2018) developed a model to

predict the probability of default when it comes to payment in e-commerce. Meanwhile,

Omoteso (2012) stressed on the development process of AI systems in auditing. Besides that, the

adoption of AI to assess internal control systems and monitor the effectiveness of audit

4
committees was also reported (Lo and Campos,2018). Notably, AI is important for businesses to

analyze the behaviors of customers for the development of specific marketing strategies. Studies

have demonstrated the positive in-fluence of applying IoT solutions on the process of developing

long and successful relationships through engagement insights. For instance, Rada ceanu (2007)

examined the potential use of AI to reproduce specific actions that must be validated by the

human factor in terms of productivity, quality, and competitiveness. In education, Bajaja and

Sharmab (2018) proposed a framework of tool on multiple learning models and AI techniques to

determine the most suitable of learning styles for a particular environment. For medical

purposes, AI is believed to minimize human error and subsequently, improve the reliability of

imaging interpretation (Fazalaet al.,2018). Meanwhile, Fujii and Managi (2018) observed a shift

from biological- and knowledge-based models to specific mathematical models and AI

technologies, particularly in the United States and Japan. When it comes to customer targeting,

personalization is very important given the significance of niche markets. Segmentation is a

process of dividing the market into specific parts with similar behaviors (Cahill, 1997). Through

segmentation, companies can gain competitive advantage as they can optimize their resources on

the target customers. In the past, the availability of quality data was limited and dominated by

demographic information from the field reports. Without AI, segmentation can be rather costly

given the need to collect a large customer database from many different areas. Nowadays, AI has

made segmentation easier and cheaper, as it is designed like the human brain to recognize and

solve problems. Moreover, in this digital era of AI technology, marketers are able understand

their customers on a deeper level. Therefore, marketing has begun to treat different customers

differently at a lower cost for its data storage with more effective technology solutions, general

advancements in know-how, and the ability to reach customers through digital channels.

5
Recently, using the power of AI, such as predictive analytics, marketers can predict the

behavioral patterns of customers with the combination of their demographic information to

identify the appropriate target customers. Based on the pre-diction output, marketers would be

able to understand the customers ‘behaviors, motivations, and expectations in order to deliver

relevant messages. This would help them to create an optimized and targeted campaign for the

target customers. Moreover, AI can provide data at a more granular level and predict the

behavior of customers based on their past behavior. Data of customers who engage with specific

brands benefit marketers. For example, the transactions of a female customer who browses and

purchases bags of specific brands contain information on her age, gender, marital status, and

income. AI using machine learning with a larger amount of data can deliver better

personalization on customer behavior, which differs from the kind of data the marketers used to

acquire. The machine would model each customer and then predict suitable pro-ducts and brands

for the target customers based on its learning of various behaviors. Therefore, marketers can

identify their target customers for segmentation with ease by referring to the outputs of machine

learning. In the past, demographic information was previously used as an indicator of human

behavior instead. Today, the customers’ needs, motivations, and online behaviors (particularly

their digital activities) are more appropriate indicators for AI systems to conclude these

customers ‘interests. With the help of predictive audience segmentation technology, marketers

can have access to valuable information of customers.[22]

It is now clear that the COVID-19 pandemic has caused a rapid acceleration in demographic,

political, economic, and technological transformation. Under these trying conditions,

organizations today have had to develop their adaptability to deal with shifting market dynamics

and customer behavior. In order to remain competitive, established companies regularly adopt

6
new technology. One of them is artificial intelligence, which is set to revolutionize analytics in

the near future. The word "artificial intelligence" is used to designate a broad variety of cutting-

edge analytic tools, software programs, and logic-based approaches that simulate human intellect

in areas such as learning and problem solving. However, artificial intelligence technologies

provide various opportunities for organizations to alter their operations across a wide range of

industries as part of the digital transformation. Decisions on loans, credits, and sales projections

are just a few areas where AI is being put to use. Benefits from AI include the automation of

manual activities and the facilitation of improved procedures via human and AI collaboration. [3]

3. Impact of AI on Data Collection:

Data Mining

Large corporations’ decision support systems often contain millions or even hundreds of millions

of records of data. These complex data volumes are too large to be understood by managers.

Consider, for example, Capital One, a consumer lending company with nearly 50 million

customers and decades of data (the company was founded in 1988), it’s easy to see how record

counts quickly grow beyond the comfort zone for most humans.10 Two points about data

volume are important to keep in mind. First, relevant data are often in

independent and unrelated files. Second, the number of distinct pieces of information each data

record contains is often large. When the number of distinct pieces of information contained in

each data record and data volume grows too large, end users don’t have the capacity to make

sense of it all. Data mining helps clarify the underlying meaning of the data. The term data

mining refers to the use of powerful computers to dig through volumes of data

to discover patterns about an organization’s customers and products. As seen in the Research

Snapshot “Mining Data from Blogs,” this can even apply to Internet content from blogs. It is a

7
broad

term that applies to many different forms of analysis. For example, neural networks are a form of

artificial intelligence in which a computer is programmed to mimic the way that human brains

process information. One computer expert put it this way: A neural network learns pretty much

the way a human being does. Suppose you say “big” and show a

child an elephant, and then you say “small” and show her a poodle. You repeat this process with

a house

and a giraffe as examples of “big” and then a grain of sand and an ant as examples of “small.”

Pretty soon she will figure it out and tell you that a truck is “big” and a needle is “small.” Neural

networks can similarly generalize by looking at examples.11 Market-basket analysis is a form of

data mining that analyzes anonymous point-of-sale transaction databases to identify coinciding

purchases or relationships between products purchased and

other retail shopping information.12 Consider this example about patterns in customer purchases:

Osco Drugs mined its databases provided by checkout scanners and found that when men go to

its drugstores to buy diapers in the evening between 6:00 p.m. and 8:00 p.m., they sometimes

walk

out with a six-pack of beer as well. Knowing this behavioral pattern, supermarket managers may

consider laying out their stores so that these items are closer together.13 Customer discovery is a

data-mining application that similarly involves mining data to look for

patterns that can increase the value of customers. For example, Macy’s commissioned data-

mining

techniques looking for patterns of relationships among the huge volumes of previous sales

records.

8
In 2011, Macy’s sent out millions of catalogs. Not every customer got the same catalog though

and

in fact, tens of thousands of version of the catalog were carefully tailored to specific

customers.14

Female customers 30 years of age will see more handbags, accessories, shoes, and women’s

clothing

than will a middle-aged man. Further, if her individual records show purchases of baby clothing,

she’ll probably see lots of things for toddlers. The end result is that the customer sees more

purchase possibilities that have a high probability of addressing some current need or desire.

When a company knows the identity of the customer who makes repeated purchases from the

same organization, an analysis can be made of sequences of purchases. The use of data mining to

detect sequence patterns is a popular application among direct marketers, such as catalog

retailers.

A catalog merchant has information for each customer, revealing the sets of products that the

customer buys in every purchase order. A sequence detection function can then be used to

discover the

set of purchases that frequently precedes the purchase of, say, a microwave oven. As another

example,

a sequence of insurance claims could lead to the identification of frequently occurring medical

procedures performed on patients, which in turn could be used to detect cases of medical fraud.

Data mining requires sophisticated computer resources, and it is expensive. That’s why

companies like DataMind, IBM, Oracle, Information Builders, and Acxiom Corporation offer

9
data-mining services. Customers send the databases they want analyzed and let the data-mining

company do the “number crunching.”. [27]

WE are living in exciting times where machine learningis having a profound influence on a wide

range ofapplications from text understanding, image and speechrecognition, to health care and

genomics. As a strikingexample, deep learning techniques are known to performon par with

ophthalmologists on identifying diabetic eyediseases in images [1]. Much of the recent success is

dueto better computation infrastructure and large amounts oftraining data.Among the many

challenges in machine learning, datacollection is becoming one of the critical bottlenecks. Itis

known that the majority of the time for running machine learning end-to-end is spent on

preparing the data,which includes collecting, cleaning, analyzing, visualizing,and feature

engineering. While all of these steps are timeconsuming, data collection has recently become a

challengedue to the following reasons.First, as machine learning is used in new applications,it is

usually the case that there is not enough trainingdata. Traditional applications like machine

translation orobject detection enjoy massive amounts of training data thathave been accumulated

for decades. On the other hand,more recent applications have little or no training data. Asan

illustration, smart factories are increasingly becomingautomated where product quality control is

performed withmachine learning. Whenever there is a new product or anew defect to detect,

there is little or no training data tostart with. The na¨ıve approach of manual labeling may notbe

feasible because it is expensive and requires domain• Y. Roh, G. Heo, and S. E. Whang are with

the School of ElectricalEngineering, KAIST, Daejeon, Korea.E-mail: {yuji.roh, geon.heo,

swhang}@kaist.ac.kr• Corresponding author: S. E. Whangexpertise. This problem applies to any

novel application thatbenefits from machine learning.Moreover, as deep learning [2] becomes

popular, thereis even more need for training data. In traditional machinelearning, feature

10
engineering is one of the most challengingsteps where the user needs to understand the

applicationand provide features used for training models. Deep learning, on the other hand, can

automatically generate features,which saves us of feature engineering, which is a significantpart

of data preparation. However, in return, deep learningmay require larger amounts of training data

to performwell [3].As a result, there is a pressing need of accurate andscalable data collection

techniques in the era of Big data,which motivates us to conduct a comprehensive surveyof the

data collection literature from a data managementpoint of view. There are largely three methods

for datacollection. First, if the goal is to share and search newdatasets, then data acquisition

techniques can be used todiscover, augment, or generate datasets. Second, once thedatasets are

available, various data labeling techniques canbe used to label the individual examples. Finally,

instead oflabeling new datasets, it may be better to improve existingdata or train on top of trained

models. These three methodsare not necessarily distinct and can be used together. Forexample,

one could search and label more datasets whileimproving existing ones.[17]

Data collection practices strongly affect the quality of crowdsourced data. This led to extensive

explorations in evaluating raters’ performance and identifying under-performing pools of raters

(Ipeirotis, Provost, and Wang2010; Bozzon et al. 2013; Sober ́on et al. 2013), improving the

overall clarity of the annotation task design (Kittur, Chi, and Suh 2008; Gadiraju, Yang, and

Bozzon 2017; Wu and Quinn 2017; Han et al. 2019), encouraging raters to reflect on their

answers (Kutlu et al. 2020), or experimenting with several annotation designs to identify the

most suitable oneto capture the appropriate answers (Inel et al. 2018; Lau,Clark, and Lappin

2014; Roitero et al. 2018). More precisely, at the level of the task design, many studies

experimented with annotation scales. For example, Roitero et al. (2018) showed that fine-grained

annotation scales are more suitable and natural than coarse-grained annotation scales to capture

11
web documents’ relevance. Intrinsic motivation and incentives have also been shown to be

beneficial in improving the quality of crowdsourced data (Ho et al. 2015; Kit-tur et al. 2013).[16]

3.2 Big Data Analysis:

A. How AI is used in big data.Online businesses are gathering data that is just astangible as

physical data. It involves consumer informationthat includes their likes and dislikes, shopping

patterns, aswell as individual preferences. Social networking accountsand online profiles, "liked"

and shared content, customerreviews, social activity, loyalty/rewards applications andservices,

and tagged preferences all add up to potentiallyinformative data to the Big Data pool. The

availability ofdata from multiple sources allows the AI to build a wealthof knowledge which will

eventually assist inaccuratepredictions about the consumer based not only on what theybought

but also on how much time they spent spend in aparticular section of a site or store [2]. A host of

otherpieces of data synthesized by AI can add to, ultimatelydiscovering what the consumer

wants. Many researcherslike tools that work well with data analytics, due to AI'scapability to

synthesize data making AI and Big Data appearinseparable. AI learns from all of the data it is

being fed andusing those inputs to create new guidelines for futurebusiness analytics. There are

issues when reporting data thatloses consistency because it is bad data.Creating efficiencies from

data is a major issue for AI aswell as translating science and research into commercialbusiness

value. An Artificial Intelligence (AI) system needsto obtain information from data to be able to

perform itspurpose [2]. Unfortunately, companies are hampered in theirability to combine data

from different sources to reach aclear conclusion about their consumers. As AI becomesmore

advanced it will solve the issues related to datacollection and analysis faster. Before running an

algorithmor program, there must be an agreed-upon approach for(mining) the data, structure, and

12
representation of the datapassed through a machine learning approach or deeplearning approach.

[18]

Deep learning is widely used to glean knowledge from massive amounts of data. There is a wide

range of applications including natural language understanding, healthcare, self-driving cars, and

more. Deep learning has become so prevalent thanks to its excellent performance with the

availability of big data and powerful computing infrastructure. According to the IDC, the amount

of data worldwide is projected to grow exponentially to 175 zettabytes (ZB) by 2025.[14]

4. AI in Data Analysis:

4.1 Machine Learning Algorithms:

According to Arthur Samuel Machine learning is defined as the field of study that gives

computers the ability to learn without being explicitly programmed. Arthur Samuel was famous

for his checkers playing program. Initially when he developed the checkers playing program,

Arthur was better than the program. But over time the checkers playing program learned what

were the good board positions and what were bad board positions are by playing many games

against itself. A more formal definition was given by Tom Mitchell as a computer program is

said to learn from experience (E) with respect to some task (T) and some performance measure

(P), if its performance on T, as measured by P, improves with experience E then the program is

called a machine learning program. In the checkers playing example the experience E, was the

experience of having the program playing games against itself. The task T was the task of

playing checkers. And the performance measure P, was the probability that it won the next game

of checkers against some new opponent. In all fields of engineering, there are larger and larger

data sets that are being understood using learning algorithms. [20]

13
In the Internet of Things (IoT), data is collected by various devices such as sensors[1], radio

frequency identification technology, global positioning system, etc. And various networks are

used to realize the connection between things and things, or things and people[2]. Recently, the

IoT technologies have made their entrances into many fields such as smart city [3],[4],[5]

healthcare [6], intelligent transportation [7], forest protection [8], Recommendation System [9],

and environmental monitoring [10]. In the Earth observation field, hyperspectral remote sensing

technology image a region by hyperspectral sensors. These images are named Hyperspectral

images (HSIs), which can be a more detailed portrayal of the field objects. Thus, they

significantly enhance the capabilities of recognition and classification. HSIs are widely used for

many fields such as Geological exploration, geological mapping, fine agriculture, atmospheric

environment, environmental monitoring. [19]

4.2 Predictive Analytics:

Machine learning represents the latest technique in statistical analysis, pattern recognition, and

predictive analytics (Theodoridis 2015). It uses algorithms to find patterns in data and to make

predictions (Clauson 2018). Generally, machine learning falls into three categories:

“supervised,” in which analysts supply algorithms with enormous amounts of labeled data (i.e.,

examples) from which to learn; “unsupervised,” in which algorithms learn on their own by

finding patterns in unlabeled data; and “semi-supervised” which combines both techniques

(Paskin 2018). Advanced machine learning involving “deep” neural networks (multi-layered

machine learning patterned after the human brain) represents the current state of the art in

predictive analytics and AI (Digiday; Sysomos 2017; Paskin 2018). The latest AI

accomplishment—whether fully autonomous cars, a bot that defeats the world’s best players in

the most complex games, one that predicts cancer, or one that personalizes incentives and

14
rewards—very likely leverages leading-edge machine learning (Brynjolfsson and McAfee 2017).

In the future, even the most advanced ML is likely to be eclipsed by emerging and entirely new

forms of AI that can read, react to, and simulate human emotions or even become self-aware

(Hintze 2016). At present, however, and at least for the next several years, ML will almost

certainly represent leading-edge AI (Levy 2016; Manyika 2017). Therefore, this paper limits

exploration of AI to the combination of big data, predictive analytics, and machine learning.

Though AI has seen previous hype cycles going back almost seven decades, this time is almost

certainly different. Organizations today not only generate vast amounts of data, many can afford

the means and tools to capture and store it. Second, computational power has grown almost

inconceivably more powerful and less expensive than it was when the first experiments in AI

were conducted in the 1950s. This increased power permits the analysis of huge amounts of

fastflowing structured and unstructured “big” data. Third, where algorithms used to require clear

and voluminous coding to instruct computers in exactly what to do—an expensive, meticulous,

and time-consuming process—today’s advanced ML algorithms learn on their own, often with

minimum instruction from humans (Brynjolfsson and McAfee 2017; Domingo 2015).

Algorithms that learn from data entered into them, unlike hard-coded programs, can adapt to a

changing world and grow exponentially more capable over time (Levy 2016; Henke 2016;

Manyika 2017; Paskin 2018). [21]

Many companies have recently resorted to data analytics using large datasets in order

tounderstand their target audience better. For instance, taking the case of the global

artificialintelligence (AI) market, American companies have increased their spending on data an-

alytics and business intelligence software, which was expected to reach up to USD 191.60billion

in the United States alone by 2025.1Following the immense growth of consumer-driven digital

15
data and the subsequent need to extract strategic critical information, theservices sector accounts

for the largest share of the predictive analytics software market.Undoubtedly, the demand for

intelligent virtual assistants would continue to rise. In thisdigital era, marketing strategies have

also started to make use of technologies to organizeand process complex datasets of customers to

target a specific market for their products orservices. As these data outputs serve as intelligible

insights for numerous brands, the motto“customer is king”can be usefully extended into digital

marketing, where customers areserved like a king by satisfying their demands personally at any

time. Optimized marketingcampaigns must work with customer sentiment in real time. With that,

communicationscan be adjusted according to the behavioral insights of the customers’emotions

andpreferences.Fundamentally, target marketing focuses on identifying a realistic approach to fit

theproducts and services for the customers. Companies may develop different

strategiesaccording to how they define their markets—the more appropriate their target market,

themore effective their strategies. Segmenting customers is very vital to develop effective

andefficient marketing programs. Marketers divide a large market into different subsets

ofcustomers with common needs, interests, and priorities before strategies to target eachsegment

are designed. Targeting specific customers would require efforts and can be rathercostly given

the need to involve marketing institutions to analyze the markets. However,with the development

of technologies, AI can now be used to define the market intospecific targets. Through data

science and machine learning, marketers can accuratelymake highly nuanced targeted decisions.

The AI-driven marketing machine can utilizesensory inputs to deduce market conditions and

even analyze visual inputs such as facial,object, and gesture recognition. This clearly

demonstrates the importance of machinelearning as an AI tool that can create human behavioral

profile in this digital era.AI has the capabilities to imitate the human brain to provide data by

16
identifying thetarget audience based on the behavioral insights historically. In particular, AI

retrieves,analyses, and presents data in a reliable and efficient way for marketers to acquire a

specificmarketing plan. AI can make a profile of human that covers the capabilities of

image,recognition, and voice via machine learning and semantic search when the

customerssearch and click certain websites to find the required information. In this era of

onlineshopping, such capabilities greatly benefit marketers in identifying the right customers

fortheir products or services.[22]

5. Challenges and Ethical Considerations:

5.1 Bias and Fairness:

Machine algorithms have penetrated every aspect of our daily life. Algorithms make movie

recommendations, suggest products to buy, and are increasingly used in high-stakes decisions in

loan applications [93], dating and hiring [18, 35]. There are clear benefits to algorithmic

decision-making; unlike people, machines do not become tired or bored [40, 98], and can take

into account orders of magnitude more factors than people can. However, like people, algorithms

are vulnerable to biases that render their decisions “unfair” [5, 100]. In the context of decision-

making, fairness is the absence of any prejudice or favoritism toward an individual or a group

based on their inherent or acquired characteristics. Thus, an unfair algorithm is one whose

decisions are skewed toward a particular group of people.[23]

5.2 Data Privacy and Security:

Among the most prominent concerns people have about AI technologies are those regarding

privacy. “Privacy,” however, has proven to be a fairly elusive concept, as indicated by the lack of

any clear consensus about its nature within the philosophical or legal literature. In the

17
philosophical literature, it has been commonly described as a “concept in disarray” (Solove,

2008), and, even more dramatically as evocative of entering an “unknown swamp” (Inness,

1992). Indeed, some philosophers have come to believe that privacy does not represent a single,

coherent concept at all (Thomson, 1975; Solove, 2008). Accordingly, the concerns people have

had about privacy are often vague and ill-formed, which makes it correspondingly difficult to

address these concerns, and explain the ways in which AI technologies do or do not pose threats

to people’s interests. In this article, we will draw attention to some important distinctions that are

frequently overlooked, and spell out their implications for concerns about the threats that AI-

related technology poses for privacy. It is common to assume that privacy is all about being able

to control information about oneself, generally in the form of having a right to prevent others

from obtaining or using information about a person without that person’s consent. Important

criticisms have been raised against such a “control” account of privacy (Thomson, 1975;

Macnish, 2018), and although we will not enter fully into this dispute, we do argue that it is a

mistake simply to equate a lack of control with a lack of privacy. Understanding why this is a

mistake requires drawing attention to some oftenignored features of the concept of privacy,

including the significant role played by beings capable of semantic understanding (Soifer and

Elliott, 2014; Marmor, 2015; Macnish, 2020).[24]

6. Opportunities and Future Trends:

6.1 Enhanced Decision-Making:

Artificial Intelligence (AI) has revolutionized business decision-making in numerous ways,

significantly impacting various industries and sectors. From automating repetitive tasks to

analyzing vast amounts of data for insights, AI has transformed the way businesses operate,

strategize, and make decisions. Here are some key impacts of AI on business decision-making:

18
1. Enhanced Data-driven Decision Making:

Businesses can make data-driven decisions because to AI's capacity to process and analyze

enormous amounts of data in real-time. Data patterns, trends, and correlations that may not be

immediately apparent to human analysts can be readily found by AI systems. This facilitates the

use of fact-based decision-making by firms, which produces better results.

2. Automation of Repetitive Tasks:

Robotic process automation (RPA), an AI-powered technology, may automate tedious and

repetitive operations, freeing up human resources to concentrate on higher-value activities.

Artificial intelligence (AI), for instance, may automate data entry, report production, and other

mundane jobs, lowering human error and increasing operational efficiency.

3. Improved Customer Experience:

AI has transformed customer experience by enabling businesses to deliver personalized and

relevant interactions. AI-powered chatbots and virtual assistants can engage with customers in

real-time, providing instant support and resolving issues. In order to provide personalized

recommendations and promotions and increase customer happiness, businesses can use AI to

evaluate client data such as purchase history and browsing behavior.

4. Enhanced Risk Management:

AI algorithms can analyze market data, customer behavior, and historical transactions to detect

fraud, assess credit risk, and make investment decisions. This helps businesses to proactively

manage risks and minimize potential losses. Optimized Operations and Supply Chain

19
Management: AI can optimize operations and supply chain management by analyzing data on

factors such as demand, inventory levels, production schedules, and logistics. AI algorithms can

optimize routes, manage inventory, and predict demand, leading to improved efficiency, reduced

costs, and better customer service.

5. Enhanced Product Development:

AI can significantly impact product development by enabling businesses to generate insights and

predictions based on customer feedback, market trends, and other data. AI algorithms can

analyze customer preferences, predict demand, and identify potential product improvements,

helping businesses to develop products that better meet customer needs and preferences.

6. Increased Innovation:

Through the automation of mundane jobs, AI may free up workers' time to concentrate on more

innovative endeavors. When companies have more time and money, they may push for

innovation in all areas of their operations, including R&D, new business models, and customer

experiences. Automating processes like market research, product design, and prototyping is one

way that AI may help with innovation and product development. New product prospects may be

found by using AI-powered technologies to glean insights from user input, perform sentiment

analysis, and locate holes in the market. Through generative design algorithms, AI may also aid

in product design and prototyping, which speeds up the product development cycle and

encourages creativity.

7. Human-like decision making:

20
Cognitive computing, NLP, and other forms of artificial intelligence may mimic human

judgment. Human emotion, mood, and preference analysis is a part of this field since it helps

with decision making in areas like customer service, marketing, and new product creation. Social

media postings and customer reviews are two examples of the unstructured data that AI can

analyze to glean useful

insights and guide business choices.[15]

6.2 Integration with Traditional Methodologies:

Critical Success Factors for the AI Implementation There are several CSFs (Critical Success

Factors) for the implementation about AI project. First one is top management decision for the

AI utilization which is related with business strategy with long term vision. Second one is

organization and human resource. Third one is infra structure. Fourth one is support from end

users. Fifth one is company strategy (Nalwa, 2003). 3.1. Top Management Decision Top

management should recognize the value of AI utilization matching business performance. Also,

top management must aware the implementation strategy of AI system and AI project roll out

plan. When building an AI system, it's best to find the processes that are most likely to have

positive business impact. This greatly increases the likelihood of successful business process

upgrades. The role of top management is to find out AI system implementation priority

discussing with AI project team. 3.2. Organization and Human Resource Regarding AI Project

Without organization and proper human resource, AI implementation is impossible and end up

with failure. Before roll out of AI project, company should set up AI organization and human

resource which are AI specialist team who have enough business domain knowledge and AI

technology back ground. Not relying with outside consultant, company should develop and keep

AI specialist internally with long term vision and perspective view. Also, company should

21
operate change management team of AI project to upgrade and expand the AI project co-working

with business end users (Jansen et al., 2004).[25]

7. Case Studies:

7.1 Successful Implementation:

Automated Transport Self-Driving Automobile Industry Context The Connected car is a great

example of an ecosystem of lots of different types of machine learning, neural nets, image

recognition, actuators, sensors, and robotics all in one platform. It has many systems and

subsystems that are a kind of modularity of components that provide different functions that are

managed with different priorities to control the vehicle. It illustrates how these different types of

machine intelligence can be applied to different problems and opportunities and to combined this

into an overall working system of systems to ultimately achieve assisted driving to full driverless

cars. Systems Used Artificial intelligence has been around for a very long time. The basic

techniques have developed but also there has been a lot of technological innovations. For

example, the ability to have auto-scaling cloud based

computing power means that anybody, large or small, could spin up a vast

number of CPU instances and run a very big machine learning algorithm

on large datasets, then shut it all down again and do it for hardly any cost.

Five years ago if you tried to do that you would have to buy an awful lot

of tin and stick it in a data center, this just would not have been feasibility from a practical

perspective. Many more people have now got access to

the kind of computing power you need to actually show progress in some

22
of these AI techniques. Another aspect is the increasing ease with which it

is possible to exploit GPUs and massive parallel processing of those cores,

which were previously just in the realms of computer games and computer

graphics that have now been added to cloud platforms as a computing resource at scale. Solving

the “in-th-car” problem is simply much more difficult than

initially thought. The type of problems that appear in self-driving cars,

are dissimilar to popular examples we have encountered thus far. We may

be able to use deep learning to play GO and Chess with lots of possible

variations of the game but the boundaries and rules are known and will

not be going outside. If you are given an unexpected route in a game are

minor in comparison to a car where there are other road users that may

not obey the law, pedestrians may walk out in front of you and other

possible events. The kind of intelligence that is needed in self-driving

cars is going to be an ensemble approach. This will need lots of different

AI techniques and algorithms, no single one of them is going to be sufficient for an autonomous

vehicle, we are not at a point where you can

simply train a giant neural net to drive a car. It is going to be many different kinds of AI working

together to keep in check each other to make

some of this work. There is massive onboard sensor ability and processing ability, many

vehicles, such as the ones from Tesla need to install a

level of hardware on their as standard that would be capable of the software for autonomous

vehicle, even more than they do now, to potentially

drive from A to B all the way. It is autonomous vehicle ready in that they

23
do not have software that is sufficiently advanced, they can deliver that

software over the air (OTA) and enable those cars to be self-driving later

on, at least that is the theory. Outcomes Artificial intelligence is certainly becoming one of the

most invested in technologies. There are many mergers and acquisitions especially within the

automotive domain. Ford announced in February 2017 that it was investing

1 Billion US dollars in “Argo AI” within the next five years, which is a startup formed in

December of 2016 primarily focused on developing software

for autonomous vehicle technology. Argo AI will become a subsidiary of

Ford and will have its own board of directors and operate as an independent

unit. The Argo AI initiative is considered to be part of Ford’s self-driving car

project, and the plan appears to be the mass production of an autonomous

vehicle ready for car sharing in 2021. There is much change in the automotive industry, and

vehicle manufacturers need to move quickly to adopt connected car technologies, as highlighted

in a recent BearingPoint consulting

thought leadership review of industry impact. Surveillance and Security Counter Measures

Monitoring Social Networks for Fake News Industry context While the percentage of

misinformation is relatively small, we have much

more work ahead on our roadmap. Normally we wouldn’t share specifics

about our work in progress, but given the importance of these issues and the

amount of interest in this topic, I want to outline some of the projects we already have underway:

– Stronger detection. The most important thing we can do is improve our ability to classify

misinformation. This means better technical systems to detect what people will flag as false

before they do it themselves. – Easy reporting. Making it much easier for people to report stories

24
as fake will help us catch more misinformation faster. – Third party verification. There are many

respected fact checking organizations and, while we have reached out to some, we plan to learn

from many more. – Warnings. We are exploring labeling stories that have been flagged as false

by third parties or our community, and showing warnings when people read or share them. –

Related articles quality. We are raising the bar for stories that appear in related articles under

links in News Feed. – Disrupting fake news economics. A lot of misinformation is driven by

financially motivated spam. We’re looking into disrupting the economics with ads policies like

the one we announced earlier this week, and better ad farm detection. – Listening. We will

continue to work with journalists and others in the news industry to get their input, in particular,

to better understand their fact checking systems and learn from them. Systems used Machine

learning text pattern recognition Sources pattern recognition—economic spamming and bot

automation

Human social media likes, shares behavior pattern recognition Human 3rd party verification

services Governance Algorithm policy management—acceptable usage on page and link through

pages Outcomes Facebook is launching an educational tool as part of measures it is taking to

counter fake news. For three days, an ad will appear at the top of users’ news feeds linking to

advice on “how to spot fake news” and report it. The campaign, which will be promoted in 14

countries, is “designed to help people become more discerning readers”, the social media firm

said. But experts questioned whether the measure would have any real impact. “Until Facebook

stops rewarding the architects of fake news with huge traffic, this problem will just get worse,”

Tom Felle, a lecturer in digital journalism at City University told the BBC. From Friday, users

who click on Facebook’s ad will be redirected to its

help centre, where they will see a list of 10 tips for identifying false stories.[26]

25
7.2 Lessons Learned:

Lessons Learned for Business Research Methodologies:

Strategic Alignment with Business Goals:

The case studies highlight the critical importance of aligning AI projects with long-term business

strategies. Business research methodologies should prioritize understanding and integrating AI

initiatives with overarching organizational objectives to ensure relevance and success.

Top-Down Decision-Making:

The involvement of top management in decision-making is a critical success factor. Business

research methodologies should emphasize engaging top-level executives in the decision-making

process regarding AI utilization. This ensures that AI projects are supported by leadership and

aligned with the overall vision of the organization.

Organizational Readiness:

Successful AI implementation is contingent on organizational preparedness. Business research

methodologies should include assessments of organizational readiness, evaluating factors such as

existing infrastructure, human resources, and the capacity for change management.

Understanding the organizational context is essential for the effective integration of AI.

26
Internal Expertise Development:

The case studies emphasize the importance of developing internal expertise in AI technologies.

Business research methodologies should explore strategies for cultivating AI specialist teams

within the organization. Relying on internal talent rather than external consultants can contribute

to sustained AI capabilities.

Integration of Traditional and AI Methods:

Business research methodologies should recognize the need for integration between traditional

methods and AI. This involves identifying processes with the potential for positive business

impact and ensuring a smooth transition by involving both AI specialists and end users in the

implementation process.

Continuous Evaluation and Adaptation:

The dynamic nature of AI technologies necessitates continuous evaluation and adaptation.

Business research methodologies should include mechanisms for ongoing assessment, keeping

pace with technological innovations, and ensuring that AI projects remain relevant and effective

over time.

Multidisciplinary Approach for Autonomous Vehicles:

27
The case study on self-driving automobiles highlights the complexity of AI applications in

certain domains. Business research methodologies should adopt a multidisciplinary approach,

recognizing that solving challenges in autonomous vehicles requires a combination of various AI

techniques and algorithms.

User-Centric Approaches in Social Media:

In the context of social media and fake news detection, business research methodologies should

incorporate user-centric approaches. This includes educational initiatives for users to enhance

their ability to identify misinformation, contributing to a more responsible and informed online

community.

Ethical Considerations and Governance:

Ethical considerations and governance mechanisms are crucial aspects of AI implementation.

Business research methodologies should incorporate frameworks for ethical AI deployment,

including algorithmic policy management and acceptable usage policies, to ensure responsible

and transparent practices.

Agility and Innovation:

Businesses need to foster an environment of agility and innovation in their research

methodologies. The case studies emphasize the role of technological innovations, such as cloud

computing and GPU utilization. Research methodologies should support the exploration and

integration of emerging technologies into AI projects.

8. Conclusion:

28
The integration of AI systems into business processes offers significant advantages, with the

primary benefits encompassing cost reduction, quality improvement, and quick response times.

As organizations navigate the landscape of AI-driven research projects, it becomes crucial to

reflect on the key findings and consider their implications for future endeavors.

8.1 Summary of Findings:

In summary, the examination of AI applications in business research methodologies has revealed

several crucial insights:

Alignment with Business Goals: Successful AI implementation hinges on aligning projects with

long-term business strategies, ensuring that AI initiatives contribute directly to organizational

objectives.

Top-Down Decision-Making: Involvement and support from top management emerge as critical

success factors. The commitment of organizational leadership is vital for the successful

execution of AI projects.

Organizational Readiness: The readiness of an organization, including infrastructure, human

resources, and change management capabilities, significantly influences the outcome of AI

implementations.

29
Internal Expertise Development: Building and nurturing internal AI expertise is essential. The

cultivation of specialist teams internally contributes to sustainable AI capabilities and reduces

reliance on external consultants.

Integration of Traditional and AI Methods: Successful projects integrate AI with existing

methodologies, identifying processes with the potential for positive business impact and ensuring

a smooth transition.

Continuous Evaluation and Adaptation: The dynamic nature of AI technologies necessitates

continuous evaluation, adaptation, and a commitment to staying abreast of technological

advancements.

Multidisciplinary Approach: Complex domains, such as autonomous vehicles, require a

multidisciplinary approach, recognizing the need for various AI techniques and algorithms to

work in tandem.

User-Centric Approaches: In social media contexts, user-centric approaches, including

educational initiatives, play a crucial role in combating challenges like fake news.

Ethical Considerations and Governance: The incorporation of ethical frameworks and

governance mechanisms is imperative, ensuring responsible and transparent AI deployment.

30
Agility and Innovation: Fostering an environment of agility and innovation is essential.

Businesses should explore emerging technologies, such as cloud computing and GPU utilization,

to enhance the efficacy of AI projects.

8.2 Implications and Recommendations:

For Businesses:

Strategic Alignment: Ensure that AI initiatives align with long-term business strategies, with a

focus on processes that offer tangible business impact.

Top Management Engagement: Foster active involvement and support from top management

throughout the AI project lifecycle.

Organizational Preparedness: Prioritize assessments of organizational readiness, including

infrastructure, human resources, and change management capabilities.

Internal Expertise Cultivation: Invest in the development of internal AI specialist teams to

enhance sustainability and reduce dependence on external consultants.

31
Continuous Learning: Establish mechanisms for continuous evaluation, adaptation, and staying

informed about emerging AI technologies.

For Researchers:

Multidisciplinary Collaboration: Embrace a multidisciplinary approach in research

methodologies, especially in domains like autonomous vehicles.

User-Centric Design: Incorporate user-centric approaches in the development of AI applications,

especially in social media contexts.

Ethical Frameworks: Prioritize ethical considerations and governance mechanisms in AI research

to ensure responsible and transparent practices.

For Policymakers:

Regulatory Frameworks: Develop and refine regulatory frameworks that address ethical

considerations and governance in AI deployments.

Education Initiatives: Support educational initiatives for users to enhance their ability to discern

misinformation, particularly in the context of social media.

32
Innovation Support: Foster an environment that encourages innovation, providing support for

businesses to explore and integrate emerging technologies into AI projects.

33