CAIIB ABFM Module D Unit 4 : Artificial Intelligence

14/10/2023

Table of Contents

CAIIB Paper 3 ABFM Module D Unit 4 : Artificial Intelligence (New Syllabus)

IIBF has released the New Syllabus Exam Pattern for CAIIB Exam 2023. Following the format of the current exam, CAIIB 2023 will have now four papers. The CAIIB Paper 3 (ADVANCED CONCEPTS OF FINANCIAL MANAGEMENT) includes an important topic called “Artificial Intelligence”. Every candidate who are appearing for the CAIIB Certification Examination 2023 must understand each unit included in the syllabus.

In this article, we are going to cover all the necessary details of CAIIB Paper 3 (ABFM) Module D (EMERGING BUSINESS SOLUTIONS) Unit 4 : Artificial Intelligence, Aspirants must go through this article to better understand the topic, Artificial Intelligence and practice using our Online Mock Test Series to strengthen their knowledge of Artificial Intelligence. Unit 4 : Artificial Intelligence

Introduction

The term “artificial intelligence” (AI) refers to the replication of human intelligence in computers that have been trained to think like people and emulate the activities that humans engage in.
The concept of artificial intelligence can be conceptualised as a computer-controlled robot designed to look and behave just like a human person.
The application of artificial intelligence is widespread throughout a variety of industries, including the banking processes, medical healthcare, institutes, surveillances, and the act of social media.
The most exciting aspect of artificial intelligence is the prospect of new research leading to the creation of computer programmes that think with minds that are as fully functional as those of humans.

The technique of making machines that are capable of performing tasks that previously required intelligence and were either carried out by humans or by other machines when necessary.
The creation of an artificial brain and its subsequent transfer to a computer in order for it to carry out tasks in a manner analogous to those carried out by a human, constitutes the entirety of artificial intelligence.

History Of Artificial Intelligence

The study of artificial intelligence is still in its development as a discipline.

1950s : Scientists and researchers began to investigate the prospect of computers processing intellectual powers equivalent to those of human beings, the academic discipline of Artificial Intelligence was born as a field of study.
Alan Turing, a mathematician from the United Kingdom, is credited with being the first person to suggest a test to assess whether or not a machine is intelligent Eventually be known as the Turing Test, a machine plays an imitation game in which it attempts to pass itself off as a human being by responding to a series of questions in a manner that is consistent with how a person would respond.
Turing held the belief that a machine could be judged to have the same level of intelligence as a human being provided it could convince a person that they were having a conversation with another human being when in reality they were not.
John McCarthy, a professor at the Massachusetts Institute of Technology, is credited with being the one who first coined the term “artificial intelligence” in 1956.
The symposium, which AI researchers later came to refer to as the Dartmouth Conference, was essential in establishing AI as a separate field of study.
The conference also identified the primary objectives of artificial intelligence, which are to comprehend and simulate the cognitive processes of people and to create robots that behave in a manner that is analogous to this.

Applicability Of Artificial Intelligence

There are many different industries that have found applications for artificial intelligence, such as medical diagnosis, stock trading, robot control, law, remote sensing, scientific discoveries, and even toy manufacturing.
Nevertheless, many uses of AI are not viewed as utilising AI.
According to Nick Bostrom’s research, a significant amount of artificial intelligence has made its way into general applications, without being classified as AI.
Many thousands of AI applications are deeply embedded in the infrastructure of every industry.
The algorithms of the artificial intelligence are designed to make the decision by using the real time data, combining all the information by using the sensors, remote inputs, digital data, and from different sources.
Research in artificial intelligence has the potential to make an important and useful contribution to the education of people.
At the very least in many instances, an intellectual difficulty can be handled by first breaking it down into pieces and then coming up with a solution for each of those individual components.
Educators and cognitive scientists have come up with the concept of intelligent computer assisted instruction (CAI), in which a computer would be programmed to act as a “tutor” that would observe a student’s efforts as they worked to solve a problem.

Contribution Of Google

Google has made immense contributions to the field of Artificial Intelligence over a period of time.

Some of the important contributions of Google to Artificial Intelligence are as under:

Search Engine Algorithm -Google RankBrain

The adoption of Google’s RankBrain, a search engine algorithm that is based on machine learning and its use was officially confirmed on October 26, 2015.
It assists Google in processing search results and providing users with search results that are more relevant to their queries.
RankBrain was mentioned by Google in an interview in 2015, and the company stated that it was the third most significant factor in the ranking algorithm, after links and content.
“RankBrain was used for less than 15% of queries as of 2015,” according to the report.
According to the findings, the results produced by RankBrain are within 10% of those produced by the human search engine engineers working for Google.

AI Hub

AI Hub provides developers and data scientists working on artificial intelligence (AI) systems with access to a collection of components to use in their work.
Making artificial intelligence more accessible to more companies requires making it simpler for them to find, exchange, and reuse the tools and work they already have.
However, until very recently, there was a lack of machine learning expertise among workers, which made it difficult to construct a comprehensive resource.
The AI Hub is a one-stop shop for plug-and-play machine learning (ML) content.
This content includes pipelines, Jupyter notebooks, TensorFlow modules, and more.

Advantage:

1st : make available to all companies in the world high-quality machine learning resources that have been built by Google Cloud AI, Google Research, and other teams located within
2nd: it gives businesses access to a private and protected portal where they may upload and share machine learning resources within their own companies.
Because of this, it is simple for companies to reuse pipelines and deploy them to production in GCP — or on hybrid infrastructures by utilising the Kubeflow Pipeline system — in just a few simple steps.

Kubeflow Pipeline System

Container-centric end-to-end machine learning (ML) processes are what Kubeflow pipelines are all about.
Components, which are self-contained collections of code that are packaged as container images, are what are used to construct pipelines.
In the machine learning (ML) workflow, each component of the pipeline is responsible for a specific stage, such as preprocessing, data transformation, or training a model.
The Kubeflow Pipelines system is responsible for orchestrating the execution of pipelines, which includes the creation and running of component containers in the sequence specified by the workflow graph so as to .
Build and distribute repeatable machine learning workflows with the Kubeflow Pipelines system.
Create machine learning experiments and deploy them into production using the Kubeflow Pipelines system.
Kubeflow Pipelines is a new component of Kubeflow, which is a well-known open source project initiated by Google.
It packages machine learning code in a manner analogous to the construction of an application in order to make it accessible to other users within an organisation.
Kubeflow Pipelines offers a workbench for the composition, deployment, and management of reusable end-to-end machine learning workflows.
This makes it a hybrid solution without lock-in that can be used from the prototyping stage all the way through production.

In addition to that, it makes it possible for users to conduct experiments in a quick and dependable manner, allowing them to explore a variety of ML techniques and determine which ones perform best for the application they are developing

Google Duplex and Hold for Me

An innovative artificial intelligence technology, known as Google Duplex,
At first, its use was limited to reservations at restaurants; but, since then, it has been broadened to include various kinds of activities.
In May 2018, during the Google I/O developer conference, Google CEO Sundar Pichai made the initial announcement of Google Duplex.
He demonstrated how the service could schedule phone appointments using a voice that was controlled by AI without requiring the user to take any action.
The artificial intelligence was not only able to comprehend what was being said on the other end of the line, but it could also provide appropriate responses to the questions that were asked of it and add “ums” and pauses in its speech so that it appeared to be more human-like.

Artificial Intelligence In Banking And Finance

Utilising AI-based systems allows for increased productivity, which in turn leads to cost savings, as well as the ability to make decisions utilising information that is unavailable to human decision-makers.
The employment of an AI algorithm system allows for the detection of fraudulent activity, as well as the easy identification of anomalies.
Following the implementation of AI technology inside the banking industry, the services sector has become one that is more technologically relevant and focused on the consumer.

A few examples of how artificial intelligence is being used in the banking industry are given below:

Customer service/engagement (Chatbot)
Robo Advice
Predictive Analytics with a General Purpose Focus
Cybersecurity
Scoring Credit and Direct Lending for Customers

Hybrid Information System (HIS)

A software system known as a hybrid information system is created by combining various artificial intelligence methodologies and techniques, such as a fuzzy expert system, a neuro-fuzzy system, and a genetic-fuzzy system.
This results in the construction of the hybrid information system.
An efficient learning system, also known as an HIS system, is one that not only combines the beneficial aspects of various learning paradigms and representations, but also over comes the limitations of processing capabilities.
These systems are also utilised for the purpose of finding solutions to issues that arise in a variety of contexts.

The following examples highlight the importance of HIS in the field of finance:

Portfolio Management

The management of a portfolio is an involved and complicated task that contributes significantly to the decision-making process.
HIS has seen widespread use in portfolio selection, and it has been playing a vital role in the operations of a great number of organisations and financial institutes.
The term “artificial intelligence” describes one of the most fundamental aspects of the modern world, and financial institutions have started incorporating related technology into their services and products in order to maintain their relevance

Stock Market Prediction:

A wide range of computer methods, which are necessary due to the highly unpredictable nature of the stock market.
Because hybrid systems are able to combine the skills of many systems with the special traits that each system possesses, they are utilised to a far greater extent in the field of financial prediction than they are in any of the other AI disciplines.

The Future Scope Of Artificial Intelligence

In Artificial Intelligence, the computer performs the following functions:

The processing of the natural language in order to make it possible for it to communicate effectively in English, natural language
For the purpose of storing the auditory inputs, it requires the Knowledge Representation.
Once the inputs have been saved, the next step in automated reasoning is to use the knowledge that has been saved to answer and question or draw any graphics.
Machine Learning is required in order to adopt all of the functions in order to take advantage of newly processed and stimulated ideas and patterns.

Neural Networks

The term “artificial neural networks” refers to a category of exceptionally effective methods that have seen a surge in popularity over the past few years.
The reason for this is that when utilised in supervised data mining applications, they are capable of producing extremely accurate predictions.
These networks are examples of highly adaptable algorithms that can be used to solve a wide variety of modelling challenges, including supervised and unsupervised issues.
When there is a categorical dependent variable, neural networks can be used instead of logistic regression and decision trees, or they can be used in conjunction with both of those methods.

Because of their high degree of adaptability and the fact that they are capable of working with continuous dependent variables, neural networks are suitable for use in situations that include regression.
Neural networks have the potential to evolve into models that are significantly more sophisticated, more flexible, and potentially more accurate when utilised in applications where other methods such as regression, logit, and decision trees may be used.
Flaws: One of the models’ flaws is that it might be challenging to understand what they are trying to convey.
When there are numerous input variables and those variables have non-linear correlations with the target variable, neural nets are especially successful.
Neural nets are especially effective when there are many input variables.

Control Theory And Cybernetics

Ktesibios of Alexandria is credited with building the earliest self-controlling machine, which was a water clock that had a regulator that kept a steady flow rate.
This was about the year 250 B.C.
The definition of what an artefact is capable of doing was shifted as a result of this creation.
In the past, only living organisms had the ability to adjust their actions in reaction to shifts in their surrounding environment.

Other types of self-regulating feedback control systems include the steam engine governor [developed by James Watt (1736–1819)]
Thermostat [Cornelis Drebbel (1572–1633)] who also developed the submarine.
The 19th century saw the development of the mathematical theory behind reliable feedback systems.
Norbert Wiener is widely regarded as the seminal figure in the development of what is now known as control theory (1894–1964).
Wiener, along with his colleagues Arturo Rosenblueth and Julian Bigelow, questioned the behaviourist dogma, much in the same way as Craik did (who also employed control systems as psychological models) (Rosenblueth et al., 1943).
They believed that purposeful behaviour originated from a regulatory system that was attempting to reduce “error,” which they defined as the

The Connection Between Thought and Language

Verbal Behaviour – 1st released by B. F. Skinner in the year 1957.
Written by the foremost authority in the field, this description of the behaviourist method of language acquisition was exhaustive and specific in its coverage of its subject matter.
However, in a strange turn of events, a review of the book became just as widely known as the book itself, and it nearly entirely extinguished people’s interest in behaviourism.
Linguist Noam Chomsky, who had recently released a book on his own theory and was the author of the review, had written the article.
The book was titled Syntactic Structures.
Chomsky pointed out that the behaviourist theory did not handle the concept of creativity in language; it did not explain how a child could understand and make up words that he or she had never heard before.

Goals

The overarching challenge of emulating (or fabricating) intelligence has been subdivided into a number of specific challenges.
These are specific characteristics or skills that researchers anticipate an intelligent system to possess.

The most emphasis has been paid to the characteristics that are detailed below:

Reasoning, problem-solving
Knowledge representation

Rational Agents

The study of rational agents is what’s meant to be encompassed by the term “artificial intelligence.”
A rational agent could be anything that makes decisions, including a person, company, machine, or even a piece of software. After taking into account the agent’s previous and current perceptions (that is, the agent’s perceptual inputs at a particular instant), it performs an action that will result in the best possible outcome.
An AI system consists of an agent and the environment in which it operates. The agents behave in accordance with their surroundings. It’s possible that the environment has other agents in it.
Agent Environment in AI
Planning
Learning