4.

RESULTS

4.1 Impact of Artificial Intelligence

Diagram 5: Relationship between Artificial Intelligence

Artificial Intelligence and Machine Learning have become synonymous with information
technology technologies in the twenty-first century. However, not knowing these words
and conflating them can act as a hindrance to mastering these concepts. Machine
learning is a subset of artificial intelligence – consider it a subset of artificial intelligence.
Artificial intelligence is a broader term that refers to the process of developing
algorithms that think like humans. The learning component of these algorithms can be
encoded by the person who also encodes the rationale that the algorithm may obey.

Machine learning is a subset of artificial intelligence in which the learning component of

the AI is not directly encoded. Rather than that, the human instructs the AI about
learning from data and drawing inferences or "predictions" based on new data. As a
result, the computer "learns" on its own using historical evidence.

AI is a collection of tasks that computers are capable of doing. For complex tasks,
machines learn these tasks using artificial learning and deep learning. These methods
work in tandem to turn data into observations in data science. Given that companies
typically benefit from unique perspectives, it is reasonable to expect that we will
continue to see an increase in the use of these autonomous computer processes in the
future.

Machine learning and the different methods used to do it share many traits at the heart
of specific ethical concerns. Two of these traits seem to pose the most legal problems:
many modern machine-learning methods are ambiguous, which means that even
professionals equipped with the appropriate equipment cannot know whether and how
inputs are turned into outputs, for example, how precisely a personal profile results in a
rating for a mortgage application or parole decision. Second, these systems require
extensive data sets for testing and evaluation.

4.2 Pros and Cons

Figure 9: AI Pros and Cons

The public is interested in learning how it can disrupt their market and reinvent their
businesses. However, the media has occasionally painted an unrealistic image of AI's
capabilities in recent years. Artificial intelligence is now disrupting online browsing,
advertisement, banking, logistics, and media, among other industries. Though AI can
change a wide variety of sectors, this transformation will not occur immediately.
Surprisingly, considering the breadth of AI's influence, its applications remain minimal.
Almost all of AI's recent success has been made via a single type, in which certain input
data is used to produce a simple answer quickly.

The ability to input data and produce production can change many industries.
Supervised learning is the scientific term for developing this program. This program is a
long way from the sentient robots promised by science fiction. Additionally, human
intelligence does far more than process inputs. These systems have advanced
exponentially in recent years, and the strongest are now developed using a technique
called deep learning or deep neural networks, which was partially influenced by the
brain. However, these devices are very far from science fiction. Numerous researchers
are investigating alternative models of artificial intelligence, some of which have shown
promise in regional contexts; there could well be a development that enables higher
levels of intelligence to be achieved, but there is currently no concrete road to this
target.

The Achilles' heel of today's supervised learning software is that it needs a massive
amount of data. Users must demonstrate several samples of inputs and outputs to the
machine. If an average human can perform a mental task in less than a second of
thought, it will likely be automated using artificial intelligence either now or soon.

AI needs a careful selection of inputs and provision of appropriate data to assist the AI
in determining the relationship between the data provided. Making strategic choices has
now revolutionized several businesses. It is on the verge of revolutionizing even more.

Once executives have a firm grasp of what AI can and cannot achieve, the next move is
to incorporate it into their strategies. It requires an appreciation of how meaning is
generated and what is challenging to replicate. The artificial intelligence ecosystem is
unusually accessible, with most leading researchers posting and exchanging ideas and
even open-source code.
4.3. SWOT Analysis

Figure 10: AI SWOT Analysis

Strengths

Increase efficiency in the workplace.

Instead of wasting hours of the workforce on repetitive, menial jobs, workers can
program artificial intelligence to perform them. While we have previously used
computers on manufacturing lines, AI enables them to serve many activities more
effectively than ever before.
It is advantageous for both businesses. Businesses save resources by using computers
to perform routine operations (rather than humans). It reduces operational costs and
even fines for noncompliance.

Adopted by many businesses.

Ai is being used in different industries. AI is being used to filter data, and it is also being
used in facial recognition and academic science. There are so many systems that are
now powered by AI and these system gives advantage to different industries.

A higher standard of living.

Artificial intelligence is also found outside of the office. Individuals who own smart
speakers and light bulbs are now using AI in their homes. These systems simplify home
management and will help you save money on your power bill. You can also find AI in
your vehicle if you purchase a reputable brand such as Tesla.

It's strange in some aspects. A few years earlier, artificial intelligence was confined to
science fiction novels, video games, and films. It has become ubiquitous, even though it
has not yet reached its maximum potential.

Weaknesses

Artificial intelligence, on the other hand, remains unnatural.

As you are probably aware, artificial intelligence is a form of technology. It may be a

robot or an algorithm. However, it is not human. And this final point remains both a
positive and a drawback. As a strength, this ensures that people who work in
occupations that need a touch of "humanity" will feel secure, knowing that their jobs are
not yet up for grabs by the technical overlords.

However, as a limitation, this implies that AI is constrained. It is a method, but not

always a remedy. AI is capable of communication but not of emotional communication.
As a result, although it can process information, it will be incapable of comprehending or
reacting to the nuances of human emotion.

The opportunity to outwit people.

Developers are constantly expanding the boundaries of artificial intelligence. At the
moment, it is capable of completing a mission, learning, and remembering knowledge.
However, maybe in the future, it can reach a point where it will improve and reinvent
itself without human intervention.

Governments are notoriously unable to adapt.

Legislation, rules, and legislation are falling behind the exponential advancement in
technology. Also, Congress does not grasp how the internet functions in its entirety, so
what chance is there for legal AI use?

Opportunities

Combining artificial intelligence with newer models of technology.

Artificial intelligence is inextricably linked to other emerging technologies such as

artificial learning, deep learning, and the Internet of Things (IoT). It's likely to be
integrated into code, allowing developers to address problems in reverse. It enables
more effective problem-solving, which can help other sectors, such as customer service.

For individuals with disabilities, smart cars accelerate growth.

At the moment, we see the introduction of AI in the car industry. Tesla automobiles use
it to self-drive on highways and park themselves without human help. It is something
straight out of a science fiction book, but it's helpful for people who have injuries that
affect their ability to drive.

Employees are not placed under too much pressure.

It enables us to automate mundane tasks. It is ideal for individuals who despise specific
roles and instead concentrate on the "big picture." Entrepreneurs and startups of
workers who wear many hats and are overworked would adore AI for this.

Threats

People will go jobless.

The general public believes that the implementation of artificial intelligence would result
in work losses. Consider the self-checkout lanes. Several of them and only one or two
workers move in when a customer encounters a problem.

For businesses, a world without human cashiers is a realistic option. one example of AI
automating routine human activities while still eliminating work openings. To address
this, the labor market must evolve. Rather than being supplanted by AI, humans would
need to coexist with it. If this is a sustainable future remains an open question.

Losing Control

People wonder whether AI can become so intellectual that humans will lose control. It
may seem to be a distant fear, but it may be closer than you believe.

4.4 Visual of Conceptual Problem Solution

Diagram 5: Share of AI Impact in Total Impact Derived from Analytics ( McKinsey Global Institute Analysis)

There are numerous limitations of AI, but there are undoubtedly more benefits that
people are now recognizing. As seen in the diagram above, many sectors are benefiting
from the increased use of technology. Artificial intelligence has started to provide
tangible benefits to companies across a range of sectors. It will become more evident
as AI technologies become more available and easier to incorporate in the future.
Despite widespread interest in using AI and machine learning technologies, adoption
and rollout in many enterprises remain poor. Most of this is because companies fail to
factor in such realities about AI programs.

Diagram 6: Job at Risk From Automation (Office for National Statistics)

It is hard to recreate Critical Thinking and Innovation, and by focusing on these aspects,
jobs will not be taken over by Artificial Intelligence. Growing public and private spending
in human resource development ensure that it is more closely matched with market
demand. Improving the job market's situation by closing the demand-supply gap and
fostering the gig economy will secure the jobs of employees.

Diagram 7: AI Safety Problems (Surtees, A., 2020)

The major providers of artificial intelligence must guide rules and standards relating to
confidence and openness in AI applications. These standards must be adhered to
faithfully by all players interested in creating and using artificial intelligence. Both
stakeholders should be mindful of the latent bias in AI algorithms and provide a rigorous
framework for detecting and resolving bias. Another critical element in bridging the
confidence divide is awareness. Users should be educated about the activities of
artificial intelligence, its advantages, and also its shortcomings.

Figure 11: Strategic Approach Solution

The first step in developing an effective AI and machine learning approach identifies the
appropriate challenges to solve and areas to turn. Although the possibilities will seem
limitless, it is critical to begin in the correct position, as this will serve as the basis for
incremental benefit from subsequent AI interventions.

On the other hand, it defies logic to provide an AI solution that cannot be expanded or
scaled in the future. Simultaneously, it is vital to bear in mind the rule of diminishing
returns and avoid losing the strategic edge that an organization can achieve with an
aggressive AI approach that can be applied earlier rather than later.

The majority of data-intensive and process-intensive business functions are ideal for AI
interference. Businesses today have an abundance of data; in practice, most of the
time, the problem is making sense of the massive volumes of data – Big Data – that
they possess. Any of the data they contain may be unusable or biased, whereas others
may be obsolete. Additionally, businesses will not always be looking at the appropriate
data sources or considering potential size when it comes to data reach and accuracy.

Although most businesses agree that AI can be beneficial, they often fail to integrate it
into their current processes and enterprise infrastructure. Utilizing out-of-the-box
artificial intelligence products and solutions can also become complicated as they
cannot be combined into current programs without requiring a complete redesign and
affecting other procedures.

Without the right AI approach, the challenge of seeking a cost/benefit trade-off, and the
pressure to demonstrate rapid ROI, AI systems often fail to take off or achieve
enterprise-level penetration within businesses. AI systems do better when used in
conjunction with a results-driven product strategy, which usually has a faster response
time than other types of projects. As a result, successful AI techniques look to solve the
management of transition and acceptance.

Considerations for security and enforcement should be incorporated from the outset of
every AI plan. Otherwise, businesses may attempt to retrofit these steps into an existing
solution or reinvent the wheel to arrive at a remedy that works – both of which would be
expensive and unsuccessful.

Figure 12: Bias in Artificial Intelligence

The first step toward regulating these mathematical models must begin with the
modelers themselves. Designers should avoid excessively complicated mathematical
tools that obscure the models' simplicity and explainability when developing these
models. They should closely examine the data used to build these models to prevent
the use of unsafe proxies.

Additionally, they should still remember the model's end goal: simplifying people's lives,
adding benefit to the world, and enhancing our overall quality of life, whether by industry
or academia, rather than relying on Machine Learning metrics such as precision or
mean squared error. Additionally, if the models are being developed for a particular
enterprise, another standard measure of performance would almost certainly have to be
added to the second plane: economic benefit.

As we move toward a more data-driven environment, policymakers will be required to

step in to provide a realistic and open regulatory framework for the use of AI models in
such sectors as banking, insurance, pharmacy, and education. Both of these are critical
components of an individual's life and should be handled with extreme caution.

As practitioners of artificial intelligence, those responsible for developing the

technologies should re-examine their data collection and usage practices. Recent
initiatives provide criteria for reporting models and datasets to eliminate negative
prejudices before their establishment, using the previously listed Model Cards and a
related framework for datasets.

Apart from that, we should strive to develop non-black-box, explainable models,

rigorously audit them, and closely observe their outcomes, taking the time to analyze
some of the results manually.

Finally, we should educate the broader population and general public on how data is
used, what it may be used for, and how it can influence them and inform them
transparently while an AI model is analyzing them.

Engineering Approach

Artificial intelligence has now been ingrained in the fabric of daily life. From an internet
search engine to email spam and malware filtering to detecting fraudulent credit card
payments, AI satisfies an individual's market, entertainment, and technological needs.
There is potential for routine implementation; machine learning and deep learning have
shown success in various fields. There are three types of learning. Unsupervised
learning is used to uncover secret patterns in data. Supervised learning is used for
inference (classification or regression). Reinforcement learning is another method of
learning in which an algorithm determines the optimal course of action based on its
consequences.

Figure 13: Artificial Intelligence in Machine Learning

Artificial intelligence is now upending markets. However, industry and IT leaders often
have a narrow perception of AI. They rely almost entirely on ML. However, AI
innovations are critical enablers of dynamic structures.

Figure 14: Artificial Intelligence System Architecture

Sensors: There is also positive and bad news about data processing, and both are
identical: There is a recent report that around 90 percent of data in the world had been
collected and approximately 80 percent of those are unstructured data, which does not
scale well when processed in traditional relational databases (Martinez D.,2020). Much
of this latest data offers tremendous opportunities in artificial intelligence, but it still
poses significant challenges.

Data Conditioning:

With such a wealth of data at their disposal, data scientists must develop novel methods
to process data while removing all interruptions for the algorithms to work. There are
methods wherein raw data are translated into information, this is needed. The resulting
data information is then fed into a machine learning sub-component for knowledge
extraction.

Human and Machine Collaboration. Machine learning tools are often confused with AI in
their entirety. However, taking a step back, it's clear to see that machine learning is just
one link in a series. Also, after extracting the knowledge, it requires collaboration
between humans and machines. For instance, a machine learning algorithm can identify
potentially cancerous cells from some machines used in the hospitals. However, it will
be up to individual physicians to verify the machine learning tool's evaluation accuracy
and perform further checks. The human-machine collaboration allows the extraction of
information and perspective.

Machine Learning: After passing structured and unstructured data to machine learning

tools, specialists may begin experimenting with various strategies for extracting
information from the conditioned data. Several classes of machine learning methods,
including unsupervised learning, which does not need labeled data; guided learning,
which does require labeled data; and reinforcement learning, which is ideally suited to
situations in which expectations and behavior can be identified without actually requiring
labeled data.

Users: It is the stage at which consumers ingest the generated insights to determine the
appropriate course of action.

Modern Computation: Variety of contemporary computational technology is used in this

process wherein the complexity of data and algorithm is determined
Robust AI: Companies make sure that all algorithms are up-to-date and explainable in
order to remain reliable and secure.

Additionally, there must be a critical examination about the bias of AI. Aside from that,
there must be implementations on training and security to maintain the fair sure of AI
and to avoid competitors from gaining control of all the AI techniques a company is
using. There are significant and influential usage cases that increase user efficiency and
decision-making capacity. However, AI can grow over time to provide more
collaborative implementations, with many human-machine teams collaborating.

5. DISCUSSION

5.1 Interpretation of Results

5.2 VAN Diagram

6. CONCLUSION

5. DISCUSSION
5.1 Interpretation of Results

5.2 VAN Diagram

6. CONCLUSION