________________________________________________________________________________

This Research Paper is written by Dakshita Sharma from Amity Law School, Noida.

________________________________________________________________________________

Chapter 1: INTRODUCTION

1.1. OVERVIEW

Throughout human history, innovation has been at the center of progress. In particular, scientific innovation has significantly contributed in shaping the civilizations of today. While new ideas and innovations significantly aid humans in taking huge leaps in terms of progress, they do come with certain caveats. There is no doubt that new technologies are aimed at making life easier for humans. However, while most individuals use new technologies for reasons of efficiency, convenience, and productivity, there may also be severe mischievous applications of the same. History serves as the best witness for the need to regulate new technologies and innovations. Nuclear technology, for instance, when first discovered, had several applications in the generation of sustainable power. However, no sooner had it been discovered, it was used to develop nuclear weapons. This is why the global superpowers felt the need to regulate the use of the same by formulating international treaties such as the Non-Proliferation Treaty. Similarly, there are countless other examples of brilliant minds thinking of seamless solutions to human problems by innovating new technologies, which seldom ends in said technologies requiring legal regulation due to mischievous applications of the same.

The case of Artificial Intelligence is no different. Contrary to popular belief, Artificial Intelligence traces its roots back to the mid-20^th century when Alan Turing, the famous British mathematician, spoke about the scope of developing intelligent machines, as well as the methodology of testing their intelligence, in his 1950 paper, “Computing Machinery and Intelligence.”

Thus began the nearly 75-year journey of the development of Artificially Intelligent machines and programs, which will be discussed later in this paper. Presently, AI occupies a pivotal, transitional, and transformational position among humans, wherein on the one hand it appears to seamlessly perform the more “laborious” tasks in the form of chat boxes, image enhancers, plagiarism checkers, and more, and on the other hand it poses threats of misuse of the same in the form of voice modulators, deepfake technology, and more.

Thus arises the need for a carefully curated approach towards attempting at pre-emptively regulating Artificial Intelligence right away as it is still in the rudimentary stages of development if one looks at the bigger and picture which describes just how much this revolutionary technology is capable of.

The dangers of AI or the possibility that it could be misused to indulge in cyber-crimes, is often misconstrued as resembling the picture that science fiction has painted in this regard. This paper will be dealing with the former, exploring the more practical threats of this revolutionary technology, rather than focusing on fiction.

For the purposes of providing a bird’s-eye view of this paper, a brief overview of what is to come in this paper can be traced as follows:

The chapter on “Introduction” (which contains sub-sections on Overview, Research Problem, Objectives, Thesis Statement, Research Questions, Significance of Research, Research Methodology, and Tentative Chapterisation), shall be followed by the chapter on “Artificial Intelligence” (which contains sub-sections on A brief history of AI, Development of AI, Present capabilities and scope of AI, Pros and Cons of AI, and Legal challenges with respect to AI), which shall be followed by the chapter on “Cyber Law and AI” (which contains sub-sections on Is the existing legal framework enough to regulate AI?, Recent laws which aim to regulate AI, Proposed Digital India Act 2023 and AI, and Way Forward), which in turn shall be followed by the chapter on “Understanding the jurisprudence with respect to AI with the help of case laws”

1.2. RESEARCH PROBLEM

The digital revolution has brought about significant changes in the realm of the ever-growing and ever-evolving cyber space, giving rise to new and intricate issues surrounding cyber security. This ever-expanding information superhighway, now, more than ever, faces the need to have a supportive and strong ecosystem in terms of Cyber Security. This need arises from the fact that new innovations in technology are accompanied with the invariable need to constantly revisit and revise the existing legal framework which regulates the Cyber Space, owing to the fact that existing laws seldom prove to be loophole-ridden and somewhat obsolete in trying to regulate new technologies. Moreover, the legal frameworks surrounding Artificial Intelligence in the cyber domain remain unclear and prove to be overwhelmingly insufficient on the question of being able to tackle newer and fresher challenges such as AI.

Thus, this dissertation, aims to tackle the research problem at hand:

The inefficiency, ineffectiveness, ambiguity, and the loophole-ridden nature of existing legal mechanisms governing the cyber space, caused by a severe coherence and lack of clarity on the same, present significant challenges for developers of AI programs, users of said programs, and all other parties concerned. Significant considerations and reservations revolve around the inexplicable inconsistency and uncertainty surrounding the implementation and enforcement of conventional information technology laws to the regulation of the mystery that Artificial Intelligence is. The shortcomings of lawmakers at addressing the long-overdue problem of being able to effectively revisit and revise the existing legal framework in an attempt to safeguard the cyber interests of individuals. The complexities of jurisdiction, technological advancements, and changing societal norms make it even more challenging for legal practitioners, policymakers, and stakeholders to navigate the ever-evolving legal landscape of Cyber Law. To tackle these challenges, it is crucial to have a thorough grasp of legal principles, ethical considerations, and policy implications in the digital age. This emphasises the pressing nature and importance of the research problem at hand.

Chapter 2: ARTIFICIAL INTELLIGENCE

2.1. A BRIEF HISTORY OF ARTIFICIAL INTELLIGENCE:
In order to understand the deeper, more complex issues surrounding the regulatory Aspects of Artificial Intelligence, we must first make a pit-stop to learn about the origins of the same, and even before diving into the origins of AI, we must define it. The term “Artificial Intelligence” pertains to the emulation of human intellectual processes by machines and computer systems. Machines’ abilities to perform human-like cognitive functions such as thinking, learning, perceiving, problem-solving, and decision-making, is the essence of AI. Highly functional and specialized applications, including expert systems, natural language processing, recognition of speech, and machine vision, all fall under the already wide and ever-growing ambit of Artificial Intelligence. The basis of AI is processing information and vast amounts of analyzation, with the objective of correlation such that important patterns may be identified, which in turn, may be used for the purposes of future predictions. While the vast majority of it’s potential remains a mystery to even the best experts on the subject, the present significance of AI lies in it’s huge potential of being capable of providing organizations with an unprecedented amount of data and related insights into their operations which were previously unknown. While experts are of the opinion that AI will eventually leap forward and outperform humans in most tasks, the technology presently remains underdeveloped and somewhat untapped in terms of taking the said leap just yet. Still, it is very much capable of outperforming human capabilities in certain task executions.

As discussed earlier, the late Alan Turing, a world-renowned British mathematician, is widely regarded as the father of theoretical computers, as well as Artificial Intelligence, owing to his invaluable contribution towards the same through his 1950 paper, titled “Computing Machinery and Intelligence.” However, the term “Artificial Intelligence” was first used in the  1956 Dartmouth Conference, organized by American computer scientist John McCarthy. It was at this conference that the term “Artificial Intelligence” was first adopted. This is widely regarded as the gateway to the rest of the world discovering the far-fetched ideas of the ability of machines to solve problems using human-like cognitive functions.

Even though the research and development surrounding AI received an unprecedented amount of traction in the build-up to the 1970’s and 80’s, governments the world over started to recede their funding into AI-centric projects because when one starts looking at empowering machines to start behaving and acting like human beings, the questions of ethics often sneak up.

Still, owing to the significant interest of scientists and the general public in AI, there were several landmark advancements made in the realm of AI. Shakey, the world’s first all-purpose mobile robot was created in year 1969. This was a landmark moment in terms of the development of AI as this mobile robot could accomplish tasks with a purpose rather than a set of instructions.

There was renewed interest in the field of AI in the build-up to the 21^st century and fresh initiatives in this field were met with fresh funding and fresh ideas. For instance, in 1997, “Deep Blue’, a highly capable and highly powerful supercomputer, was created by Tech-giant IBM. The capabilities of this supercomputer proved to be a testament to the vast potential of AI as this supercomputer went down in the history books for defeating the then World Chess Champion, Gary Kasparov.

Several other milestones gradually shaped the history of AI as we know it today. While there are countless other examples of the progression of AI-powered machines, perhaps the best ones are listed as follows: In 2002, the world’s first commercially viable robotic vacuum cleaner was invented. The period from 2005-2019 saw AI take unprecedented leaps in terms of its progression, in the form of voice recognition, robotic processes automation (RPA), dancing robots, smart houses, and more. From state-of-the-art voice recognition systems such as those pioneered and engineered by Tech-giants like Google and Apple, to self-driving vehicles such as those pioneered and engineered by Elon Musk’s Tesla, AI really has come a long way since it’s inception. Another famous example and yet another feather in the already feather-rich hat of Artificial Intelligence is the LinearFold AI algorithm, which was made available to scientists who were developing the vaccine for SARS-CoV-2 virus during the COVID-19 pandemic. This program was credited with the ability to predict the virus’s RNA sequence in under twenty-seven seconds, 120 times faster than erstwhile approaches.

2.1.1. Types of Artificial Intelligence:

Artificial Intelligence may be broadly classified into the following four types:

2.1.1.1. Reactive Machines: This is perhaps the simplest form of AI, wherein the computers, in this case known as purely reactive computers, are not equipped to retain memories or patterns or prior experiences for use in the present. These machines solely concentrate on the problem which has been presented to them in real-time and respond in real-time in the best possible manner as per their capabilities. A reactive machine is one which is capable of employing its intelligence to not only perceive, but also respond and react to the problem in front of it. While Reactive Machines as a branch of AI are impressive, perhaps their biggest drawback is the fact that since they are not equipped to store memory, thus, they fail to rely upon past experiences to influence real-time decision-making, which hampers their ability to compete with other AI systems and machines which do have the ability to store prior information as a reference to provide future solutions.

2.1.1.2. Limited Memory: The term “Limited Memory” exhibits an AI system’s capability to retain prior information and predictions such that it can use this information with reference to updating future predictions. While Limited Memory AI machines are a step ahead of Reactive Machines, they are still limited in their functionality as they can exclusively utilize only that portion of data which they have saved for a brief period of time, that is, temporarily stored data. The very foundation and architecture of Machine learning, at this point, proves to transition into something much more complicated when there is a question of limited memory involved. When Limited Memory machines are gathering information and weighing potential decisions, AI with limited memory proves to be more complex and offers more possibilities than reactive machines.

2.1.1.3.  Theory of Mind: This form of AI is an extremely refined and cutting-edge class of innovation and technology, albeit with the caveat that it presently only exists in concept. This kind of machine has not yet been built, with researchers attempting to advance its capabilities. A very refined sense of profoundness as to being able to grasp how people and things in its surroundings can change feelings and behavior, is a necessary prerequisite of this kind of AI. Although, as things stand, presently, the Theory of Mind AI is closer to science fiction than it is to reality. This form of AI has been touted as being able to connect socially with humans and comprehend human emotions, people, and opinions. Even though it seems far-fetched, the scope, capabilities, and sheer drive of the scientific community, form the collective driving force behind the efforts to develop this technology as the next step in the AI revolution.

2.1.1.4. Self-Awareness: This is where things start to get a little horrifying as this form of AI, if fully developed, will result in all devices in the Internet of Things being extremely intelligent, and in full possession of consciousness and feelings, in addition to self-awareness. Self-awareness AI systems will be able to detect human emotions, alongside being able to comprehend and interpret their internal characteristics, circumstances, and moods. If predictions are to be believed, these devices are poised to be more intelligent than the human brain. However, we are still a long way off being able to develop Self-Aware AI machines. This is because self-awareness in AI is all but dependent on two factors. Firstly, it is dependent on whether human researchers will be able to comprehend the premise the consciousness. Secondly, it depends on whether such a comprehension of consciousness can be replicated to the extent of building it into machines.

2.2. EVOLUTION OF AI:

Looking back at how the revolutionary technology of artificial intelligence came to be, most people are surprised to learn that in the first half of the 20^th century, it was actually science fiction cinema which may be credited with the initial popularization of AI. As per a Harvard study, it was actually the infamous portrayal of the “heartless” Tin Man, a character from the motion picture, Wizard of Oz, which kickstarted the widespread popularization of AI in the world. This, as per the study, continued with a humanoid robot who impersonated Maria in Metropolis. As a result of the general public having been sensitized and acclimatized to the concept of AI, by the 1950’s, a brand-new generation of scientists and people belonging to academic and scholarly backgrounds, came to possess the concept of AI culturally assimilated within their minds. Here, we again reference the late Alan Turing, who, through his paper, Computing Machinery and Intelligence, attempted to suggest that since humans can use the available information at their disposal, as well logic and reason to solve problems and make decisions, then perhaps the same can be expected of machines. In his paper, Computing Machinery and Intelligence, Turing discussed the way to build intelligent machines. He also went one step further to suggest the methodology of testing the intelligence of such machines.

Turing, however, was unable to test out his theories because the computers of that day and age lacked the basic prerequisite for intelligence – a fundamental absence of the ability to store commands. This meant that the computers could be told what to do by feeding into them a set of instructions. However, they were unable to remember what they did, as the memory was lacking. Moreover, a proof of concept was needed to fund the research and development initiatives in this field, which was hard to come by.

Proof of concept, however, was showcased not long after that by Allen Newell, Cliff Shaw, and Herbert Simon’s, “Logic Theorist.” This was a program which was designed to mimic the extraordinary problem-solving skills of a human. This project is widely considered to be the first-ever AI program and was then presented at the Dartmouth Summer Research Project on AI, which was hosted by John McCarthy in 1956. The Dartmouth conference proved to be a landmark moment in the history of the evolution of AI as it acted as a catalyst for the next 20 years of AI research.

The period between 1957 and 1974 witnessed an extremely flourishing and thriving environment for the development and progression of AI technology. One of the major reasons behind this was that computers could now store more information. In addition, computers also became faster, cheaper, and more accessible. Another reason was that the Machine learning algorithms vastly improved. As a result, people improved at having a thorough knowledge of the specific algorithms that they could apply to their respective problems. Examples of progress in the field of AI technology during this time period was demonstrated by a machine called “General Problem Solver” by Newell and Simon and another one called “ELIZA” by Joseph Weizenbaum. However, despite these promising strides being made in the realm of AI, it was still not enough as there was a basic lack of computational power to achieve anything substantial, that is to say that the computers of that day and age simply could not process information fast enough, nor could they store enough information.

The next phase of evolution occurred in the 1980’s wherein significant emphasis was placed upon deep learning techniques. These techniques empowered computers to learn with the help of experience. Moreover, expert systems were also introduced in the 80’s, which worked towards mimicking the decision-making process of a human.

AI really began to thrive from the late-1990’s onwards. As mentioned earlier, “Deep Blue”, an AI program developed by IBM, managed to beat the then reigning World Chess Champion, Gary Kasparov in 1997. Moreover, in the same year, a certain speech recognition software by the name of “Dragon Systems” was implemented on Microsoft’s Windows. Another huge leap forward came in the shape of “Kismet,” a robot that could recognize and display emotions.

These gradual developments have now finally culminated in the Age of AI revolution, where the technology has multiple implications and is being used everywhere and by everyone.

The finest examples of innovation using AI technology in recent times are listed below:

1. Tesla’s autopilot function: Electric car manufacturer, Tesla, gives buyers the option of putting their vehicle on autopilot mode, just sit back, and relax, while the vehicle drives itself using a radar coupled with AI features.
2. Google’s AI-loaded smartphones: Technology Giant Google has recently stepped up to fight the competition in the smartphone market by giving users an AI-loaded experience in their flagship smartphone series, the Pixel. Google’s phone comes with AI features such as magic eraser, voice commands, and more.
3. Voice Recognition Systems: Voice recognition systems have come a long way since their inception. Presently, personal voice assistants such as Amazon’s Alexa and Apple’s Siri serve as wonderful examples of just how far AI technology has come.
4. Chat boxes: AI-powered tools have recently led to a huge popularization of the technology in the digital realm. One such AI-powered tool is a Chat Box wherein an AI program performs the tasks assigned to it in the form of commands. The finest such example is, of course, that of Open AI’s Chat GPT.
5. Data Analysis: We live in the world of Big Data wherein its all but impossible for humans to curate and analyse mounts and heaps of data. While this task may seem laborious for humans, it is the kind of thing which AI-powered programs really excel in.

 2.3. PRESENT CAPABILITIES AND SCOPE OF AI
Thus far, this dissertation has been able to establish that Artificial Intelligence has immense untapped potential, which, if pursued correctly, could lead to massive progress in the field of Information Technology. There still, however, remains a vast gap between the current capabilities of AI Programs and what the future of AI-powered technology could look like.

In this sub-section, we shall outline the both, the present capabilities of AI, as well as the Scope of the same.

Currently, AI is capable of performing the following tasks:

2.3.1. Deep Learning: Deep Learning refers to the use of artificial neural networks by machines, in order resemble human-like cognitive functions in its understanding and performing various different tasks. Deep learning is widely believed to be the next step of the development of AI wherein huge neural networks, with multiple processing layers, are employed to perform Artificially Intelligent tasks by machines. Deep Learning happens to be a subset of Machine Learning, and is significantly based upon our understanding of how the human brain is structured and how the human brain works. There still, however, remains massive room for improvement in this aspect of AI since the brain is the most mysterious part of the human body, and even humans do not possess a complete understanding of the brain, thereby hampering our ability to replicate something that we do not fully understand.

2.3.2. Machine Learning: Machine Learning refers to the study of programs that can improve their performance on a given task automatically. Machine Learning is one of the founding principles of AI and has been a part of the development process of AI technology since the very beginning. AI-powered machines and programs must have an inside-out knowledge of machine learning in order to function as AI-tools. Moreover, Machine Learning is what gives rise to Deep Learning, the next level in terms of AI-development. Therefore, it is an extremely important capability of AI-powered machines and programs.

2.3.3. Automation: Automation focuses on reducing the need for human intervention in processes with the help of predetermined decision-making criteria, subprocess relationships, and more. With the help of AI technology, the process of Automation can be further streamlined to increase productivity manifold, while also maintaining efficiency of tasks. An excellent example of Automation is Robotic Process Automation (RPA), a type of software that works towards automating repetitive and rules-based data processing tasks that are traditionally performed by humans.

2.3.4. Natural Language Processing: Natural Language Processing or NLP falls into the category of one of the most advanced use of AI that empowers machines to understand, analyze, and communicate in human languages. Some of the most basic examples of NLP are: email filters, predictive text on smartphones, search results on search engines like Google, language translation with the help of voice recognition, text analytics, and more. The next target stage of NLP is being touted as letting gadgets communicate with humans with the help of everyday language with the end-goal of being able to perform tasks.

2.3.5. Speech Recognition: Speech Recognition refers to the capability of AI-powered machines and programs to be able to convert Speech into text and vice-versa in their interactions with humans in real-time. Speech recognition forms one of the key components of AI technology with the best examples of the same being infamous speech recognition software in the form of Amazon’s Alexa, Google’s voice guidance in Google Maps, and Apple’s Siri.

2.3.6. Customer Service: Of late, it has been observed the world over that virtual customer service agents in the form of AI-powered software and programs are seamlessly replacing human agents in the customer service industry. This can, in particular, be observed in the form of smaller-scale AI-powered technology in the form of chatbots, which enables smaller brands to save resources and improve customer satisfaction manifold

2.3.7. Computer Vision: Tasks which are performed using Computer Vision, include various different methodologies of acquiring and analyzing, alongside processing digital images, as well as extracting high-dimensional data from the real world for the purpose of producing symbolic information or numeric information. Thus, Computer Vision is another capability of AI-powered technologies, which is based on the foundations laid down by the likes of machine learning and deep learning. In addition, such intelligent computer systems also indulge in pattern recognition of picture and video data. The sheer ability to offer recommendations is the sole factor which distinguishes Computer Vision from image recognition tasks.

2.3.8. Pattern Recognition: Another application of AI-powered technologies happens to be Pattern Recognition, otherwise known as Anomaly detection. Such intelligent systems attempt to find inconsistent patterns to decipher and distinguish the normal from the abnormal. Moreover, Anomaly detection can also be employed to comb through mounds and heaps of datasets to detect atypical data. This technology has huge applications in detection of crime patterns, among other things.

2.3.9. Decision-Making: With the help of cutting-edge technologies such as Machine Learning and Deep Learning, AI-Powered tools are able to develop a better understanding of how to approach a task. Hence, if tasked with the responsibilities of making decisions on their own, they are well-equipped to analyze and process the risks and then either recommend a suitable course of action, or perform the most suitable course of action themselves.

2.3.10. Predictions: AI-powered technology also has huge applications in the field of making predictions and educated guesses, owing to the fact that they can easily recognize and understand the patterns of datasets that they are given to analyze and make predictions on the basis of that.

Furthermore, while the scope of AI is an extremely deep topic, this dissertation shall be shedding some light on the same with respect to the case of India as follows:

1. Policing: India has a policy of conventional policing. However, with the advent of new technologies like AI in the digital revolution, AI-driven products offer a whole host of opportunities to embark upon a journey of predictive policing, wherein AI-tools may be utilized to analyze datasets such as CCTV footage, in an attempt to predict the very pattern of time, thereby aiding law-enforcement authorities in the identification and penalization of suspects. To this effect, the Government of India has already begun digitizing criminal records, putting them into a single repository called the CCTNS. Here in the CCTNS, all the data including images of suspects, biometrics, and the criminal history of the convict or suspect, are all available.
2. Agriculture: AI-driven technologies has several applications when it comes to the agricultural sector, wherein AI-tools may be used to asses the amount of water that the crops will be requiring, in addition to AI-powered seed distribution systems, and even AI-driven technologies aimed towards predicting the pattern of rainfall in a year.
3. Analyzing data: AI-driven technology proves to be an invaluable aid in terms of analyzing data and as a result, can improve the efficiency of the systems like power management in cars, mobile devices, weather predictions, video and imaging analysis, among others.

2.4. PROS AND CONS OF AI
Since coming to the forefront of the digital revolution and leading the way since then, we have all heard about this revolutionary next step in terms of technological advancement, making headlines on a consistent basis. AI-driven technologies have been on the front pages for good and bad reasons alike. However, personal opinions aside, AI is here to stay. This sub-section of the dissertation is going to focus on attempting to outline the pros and cons of AI-driven tools and technologies.

2.4.1. First, let’s have a look at the Pros:

2.4.1.1. AI is credited with eliminating human error, and to a large extent, risk: This is a huge advantage of pushing AI-driven tools to be used in order to perform such tasks, which when performed by humans, entails the possibility of human error or an increased risk to humans in the form of injury. An example of the former would be to task AI-driven tools to perform repetitive tasks which would be error-free in comparison to the same tasks being performed by humans. On the other hand, an example of the latter would be to task AI-powered robots to be used in areas with high radiation, thereby reducing the risk of human casualties.

2.4.1.2. Increased productivity in terms of longer productive hours: Humans are generally known to engage in eight-ten-hour workdays. The same principle does not apply to AI since it cannot get tired or exhausted while performing a particular task. This, in turn, ensures that the technology is available to be put to productive and efficient use throughout the day.

2.4.1.3. Being Unbiased in terms of Decision-Making: Humans are prone to a lot of biases when it comes to making decisions. No matter how much a human tries to remain unbiased, human decision-making is more often than not perforated with personal biases. The same decision-making tasks, however, when assigned to AI-powered tools, will lead to the purest form of unbiased decision-making, for example, in approving loans, selecting job applications, etc.

2.4.1.4. Performing jobs that are repetitive in nature: Boredom stems from many of the tasks that are assigned to humans such as data entry, data analysis, report generation, verification of information, and the like. Being of a highly repetitive nature, AI is better suited to perform these jobs in comparison to actual humans, thereby leaving more breathing room for humans to perform jobs that are of a rather creative nature.

2.4.1.5. Reduction of Costs: AI can be credited with being able to save humans a lot of money in terms of being able to replace certain aspects of the existing human labor force. This, in turn, invariably leads to a complete elimination of the need of corporates to search of cheap labor, and instead, assign those tasks to AI-powered machines, thereby freeing up funds to invest in other aspects of their business.

2.4.1.6. Analysis, as well as Acquisition of Data: high volumes of data can be easily processed by AI-powered tools, in comparison to the same being attempted by humans. Moreover, AI-powered tools can also help in breaking down complex data, something which is achievable by humans, but not as smoothly and quickly as it can be done by AI.

2.4.2. Now, let’s take a look at the Cons:

2.4.2.1. The implementation is very costly: It has been observed through multiple studies that the up-front cost of implementing AI solutions for one’s business, the cost can be anywhere between $20,000 to a figure which could be well into the millions. This variation in costs exists because the same AI-tools will not be used by different businesses as each business will tailor the AI to meets its own unique needs. While there exists an argument that in the long-run, this high downpayment will more than make up for what it costs, the downside of the same is that the up-front costs themselves are enough to intimidate most people to steer clear of AI solutions.

2.4.2.2. AI lacks creativity and emotion: These are severe disadvantages of AI-powered technologies because at its core, AI cannot create ideas. It may be programmed to create “novel” ideas, however, when it comes to creating original ideas, AI severely falls short of the natural human capabilities to create original ideas. Moreover, AI also lacks emotion, which could prove to be useful when it comes to making unbiased decisions, it may also prove to be counterproductive when it comes to invoking a little bit of emotion in the decision-making process in order to make the best decision. AI can make the optimal decision with the help of the defining parameters of said decision, however, it cannot replace the inherent human traits of compassion and kindness, which often lead to the best decisions.

2.4.2.3. Degradation: like all machines, AI is also subject to wear and tear, and that is why if someone is looking to implement an AI-powered solution to their problem for a long term, the AI-powered tool in question will always require to be closely monitored so that it can be well-maintained to optimally function in the long term. This includes maintenance on the one hand, and training on the other hand. Training is necessary in order to ensure that the technology does not become obsolete over the years.

2.4.2.4. Unemployment: While there are many benefits of AI such as task-optimization, improved efficiency in performing tasks, increased productivity, etc., one of the biggest downsides of a global implementation of AI-powered technologies, is that it will lead to rampant rates of unemployment. The global Unemployment figures are already quite high, with most of the underdeveloped and developing nations posting worrying figures of unemployment rates. This is predominant in Asian, African, and South American countries, which also makes the emerging youth population of these countries highly susceptible to accept undervalued wages while working as laborers. If AI were to gradually replace large portions of the existing human workforce, then this class of workers who are surviving on already undervalued wages, would further lead to massive layoffs, thereby leading to a severe skyrocket in the prevailing rates of unemployment. This, in turn, will have an extremely negative impact on the socio-economic aspects of human life, as a higher rate of unemployment can have severe ramifications such as increased poverty, increased hunger, increased malnutrition, increased rate of crime in society, etc.

2.4.2.5. Ethical challenges: Another massive hurdle in terms of global implementation of AI-powered technologies, is the ethical barrier for the same. The unprecedented pace at which creation and implementation of AI has invariably given rise to a myriad of questions about the growth of AI and the continued use of this technology. Perhaps the biggest ethical question mark surrounding AI-powered solutions is related to the issue of consumer data privacy. This poses a huge problem when it comes to the question of informed consent of the humans to whom the data belongs in the first place. Another related aspect of this obstacle is the fact that AI happens to be something which is very good at understanding, plotting, and further recognizing patterns. This again leads to a huge problem as this could translate into a bigger problem of AI being able to acquire consumer data even without having direct access to personal information. This, however, is just one single aspect of the larger umbrella of ethical challenges that the large-scale implementation of AI entails. Other problems include the question of unemployment and legal responsibility.
2.5. LEGAL CHALLENGES WITH RESPECT TO AI

The realm of the ever-evolving technological landscape has undergone massive changes since AI has leaped to the forefront of the digital revolution. The much-awaited technology of AI has finally made it to mainstream technological applications, while managing to bring about a paradigm-shifting revolution in the realm of the cyber space, with applications in nearly every major industry, from healthcare to finance. However, for all the good that it offers, at its core, it still remains a largely unregulated technological advancement. The need to regulate AI and to establish all-encompassing legal doctrines in order to minimize a misuse of the same, it is imperative to examine the interplay between this rapidly evolving technological innovation and the existing legal framework of the cyber realm. Legal issues that invariably stem from an increasingly AI-centric global landscape, primarily arise due to the functionality of AI, which makes it possible for this paradigm-shifter of an advancement to perform tasks which were traditionally performed by humans. Examples of AI taking over traditional human tasks are writing, painting, and decision-making, among others. Whether one likes it or not, this shift ultimately raises legal questions, with the most primary batch of legal questions in terms of AI, revolving around questions of intellectual property rights, liability, data privacy, as well as many ethical considerations of the same.

As things stand, it is hard to imagine human society without a substantial, if not a complete integration of AI with nearly all aspects of society, in the foreseeable future. This consequently leads to an emergent need of developing a highly specialized wing of legal doctrines which aim to regulate AI. This, in turn, invariably encompasses a whole host of legal domains, including those of intellectual property law, law of privacy, law of contracts, and law of torts, among other branches of law, which need to be re-evaluated and reshaped to cater to the unique needs of regulating AI. Therefore, a comprehensive and complete understanding of the legal issues surrounding AI, is not just about finding applications of existing legal doctrines to this new innovation, but also about an emergent need to develop fresh legal paradigms that can work towards a holistic accommodation of the nuances of AI

What are the key issues surrounding AI in terms of the legal domain? 

2.5.1. Intellectual Property Disputes: Recently we have witnessed a huge wave of integration of AI-powered tools into the day-to-day lives of individuals who were earlier oblivious to the very existence of this technological marvel. The rapid pace at which AI-powered tools such as chat-boxes, image enhancers, voice generators, and the like, are being interconnected with the very fabric of human existence, means that the average individual is increasingly starting to rely upon AI-powered tools to make life easier. All sections of society, from students and working professionals to entrepreneurs and creative artists, have begun relying upon AI in their respective fields. This could have large scale legal implications, thereby giving rise to fresh legal problems. Perhaps one of the foremost legal challenges in terms of AI, is the issue of Intellectual Property Rights, which questions of ownership and authorship AI-generated works being the principal issues in this regard.

As things stand, AI-generated works are further creating a whole host of fresh hurdles for the legal community to tackle. One of the finest examples in this regard is the generation of creative works by AI-powered tools, for instance, a painting. Now, since the painting has been created by using the capabilities of AI, the existing legal system, as well as the forthcoming legal frameworks, must be well-equipped to answer the questions of copyright in terms of AI-generated works. This is to say, that the legal system must be well-equipped to determine whether the said work can be copyrighted or not. Moreover, if the copyright can be granted for such a creation of an AI-powered tool, then perhaps a deeper question along the same lines would be one of the determining factor of who would hold that copyright – whether the programmer holds the copyright for the creation, or the AI entity itself, or the user who initiated the creation of such a work. Traditionally speaking, the ownership or authorship of copyright in computer-generated works never seemed to be in question since the program was always considered to be a mere tool which supported the creative process, not much different than a pen and paper. However, with the advent of newer and more advanced versions of AI, the computer program is no longer treated as being a mere tool, rather, it is often tasked with many aspects of decision-making while it is involved in the creative process with no human interference whatsoever.

There are primarily two ways in which copyright law can attempt to deal with works in which there is minimal to no human interference. Firstly, copyright law can deny copyright protection for such works that have been generated by a program, or it can simply award the ownership of such works to the creator of the program. Even though there has never been a strict prohibition against the conferment of copyright in works that have been generated by AI, there are many international case laws through which it is clear that the laws of most of the nations are unamenable to non-human copyright. For instance, the US Copyright Office declared that it will “register an original work of authorship, provided that the work was created by a human being.” This stance of the US Copyright Office originates from Feist Publications V Rural Telephone Service Company, Inc. 499 U.S. 340 (1991) through which it was laid down that the laws of copyright only protect “the fruits of intellectual labor” that “are founded in the creative powers of the mind.” Another example of the same can be seen in Acohs Pty Ltd. V Ucorp Pty Ltd., an Australian case in which it was laid down that the intervention of a computer in a generated work cannot be copyrighted since its production cannot be attributed to a human. The best example, however, is the CJEU’s (Court of Justice of the European Union) landmark judgement in Infopaq International A/S V Danske Dagbaldes Forening, in which it was laid down that copyright solely applies to original works, which must necessarily reflect the “author’s own intellectual creation.”

Secondly, the alternative option points towards a solution of handing over the authorship of the generated work in question to the programmer himself. This unique approach of granting copyright to generated works can be evidently noticed in major countries such as Hong Kong, India, Ireland, New Zealand, and the UK. Its not difficult for one to notice that this approach of granting copyrights for generated works, originally stems from Section 9(3) of the UK Copyright, Designs, and Patents Act (CDPA), which reads as follows:

“IN the case of a literary, dramatic, musical, or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken.”

Moreover, the CDPA, through section 178, goes on to define a work generated by a computer as one that “is generated by computer in circumstances such that there is no human author of the work.” It has been since observed that the purpose behind such a provision is basically to account for an exception to human authorship requirements. This is done by recognizing the effort which goes into the creation of a program which is capable of generating works, even though the creative spark is overtaken by the machine in question, and not by a human.

2.5.2. Concerns surrounding consumer data privacy: The inherent effectiveness and prowess of AI-powered technologies stems from the fact that this revolutionary new-age technology primarily relies on large datasets for the purposes of training and operations. This leads to a rise in significant privacy issues. The concerns surrounding consumer data privacy primarily arise when the personal data of the consumers starts being used by AI-powered technologies without the explicit consent of said consumers/users thereby leading to a potential, but substantial breach of privacy laws.

The foremost privacy concerns surrounding AI-powered technologies stem from the potential it possesses for breaches of data, in addition to unauthorized access to personal information. Keeping in mind the fact that a large volume of data is being collected and processed, there happens ton be a high risk of the said data ending up in the wrong hands, be it through hacking, or other forms of security breaches.

Being a constantly evolving entity, the increase in the involvement of personal information when it comes to AI, is inevitable. This, in turn, leads to proliferating the cases of data breaches. For instance, generative AI-powered technologies may me misused with the intention of creating fake profiles or with the intention of manipulating images. The bottom line overwhelmingly points to the horrific fact that personal data in the wrong hands can lead to disastrous, if not monstrous outcomes.

Another major concern in relation to the potential misuse of collection of data by AI, is the use of these technologies for the purposes of surveillance as well as monitoring. A good example of the same can be seen in the case of facial recognition technology wherein law enforcement agencies are increasingly relying on the same in order to identify and track suspects. This raises significant questions about the right to privacy and the potential for a severe abuse of these technologies. Possessing the ability to analyze large volumes of data, AI could potentially be used to monitor individuals in a wide variety of ways, which were previously thought to be impossible. This includes, but is not limited to, tracking of movements, monitoring of social media activities, and a deep analysis of biometric data, including facial expressions.

Furthermore, there also exists a concern regarding the potential of AI systems to perpetuate a whole host of existing biases and discrimination. For instance, the data which is used to train a particular AI system may contain certain preference biases, and since the AI is being trained on such a dataset, there is a very large possibility that the said system may learn and perpetuate those biases. This in turn, may have significant consequences, with areas such as employment having perhaps the most severe consequences, wherein AI-powered algorithms could potentially be used to make decisions of hiring, etc.

In order to avoid the aforementioned pitfalls, it is all but necessary for the developers of AI-based systems, to ensure that AI-technologies must be programmed, developed, and deployed responsibly. This means that data should be collected and processed in an increasingly transparent and secure manner such that individuals have little to no control over the same. This also means that AI systems must be designed and tested in an attempt to identify preference and other kinds of biases, and mitigate the same. Lastly, there also needs to be significant efforts in the sphere of monitoring and oversight of these technologies in order to avoid these pitfalls.

2.5.3. Liability in AI-decision making: Being able to determine liability is perhaps the most foremost legal concern in relation to anything and everything which needs to be regulated. As such, the extremely relevant question of bearing legal responsibility for the actions of AI-powered technologies, becomes an increasingly pertinent one. In order to understand the significance of determining the liability in terms of decision-making by AI, let’s take a look at an example. Suppose an AI-powered vehicle is met with an accident. In such a scenario, the legal liability for the same must be determined – where the manufacturer is responsible, or the software developer in question, or the user of the said vehicle. The answer may vary depending upon varying circumstances. This, in turn, leads to the concerns regarding the wide variety of scenarios that legal frameworks around the world need to be well-equipped enough to answer and determine.

The fact that the jurisprudence in terms of cases in this field are fairly new and still evolving, they still often revolve around product liability, as well as negligence claims.

Solutions, in this regard, include the development highly specialized legal frameworks for AI accountability, insurance models for risks arising out of AI, and crystal-clear guidelines in terms of AI deployment in sensitive areas.

2.5.4. Transparency and Requirements of Explainability: The current legal landscape is pushing for legal mandates for AI-powered technologies to be highly transparent in their decision-making process, as well as possessing an element of being able to explain their decision-making processes. This view is gaining massive traction in sectors where the application of AI-systems’ decision-making can have huge consequences, for instance, in the sectors of finance and healthcare.

There is an increasing list of examples where AI-powered systems have either failed or caused harm due to the increasingly opaque nature of their algorithms. This has led to a highly cautionary approach in this regard.

Existing and upcoming legal frameworks, however, may be able to address these concerns by making it mandatory for AI-systems to have ‘explainability by design’, in addition to adhering to the emerging standards and regulations which are focused on delivering AI transparency.

2.5.5. The issue of the legal personality of AI: legal personality pertains to the ability of a subject of law to assume certain obligations, and exercise certain rights, in the context of a given legal system. Where there is talk of legal personality in the context of law, there are two kinds of legal personalities: legal persons and natural persons. If AI were to be given the full recognition of being a legal personality, then in such a case, it could exercise ownership, enter into contracts, become the holder of bank accounts, etc.

Many jurists are of the opinion that the unique legal situation of AI may be compared to that of ‘quasi-persons’ that the law once encountered.

Chapter 3: CONCLUSION

This dissertation has conducted an in-depth study on most, if not all, of the particulars that fall within the larger umbrella under the topic of AI-powered technologies and their regulations thereof. Post the introductory section of the dissertation, we moved towards an exploratory view of the historical aspects of the development of AI and how the technology exists in the present day. Thereafter, it was necessary to analyze the plethora of applications that the AI-powered technologies might have in the present technological scenario, as well as in the future, since such new-age, cutting-edge technologies are constantly on the up and are constantly surprising even the developers themselves as to just how much of an evolutionary leap they depict in the digital realm. Thereafter, we attempted to understand the wide variety of legal challenges and issues that the new-age technologies like crypto, blockchain, and indeed AI-powered technologies, pose for the lawmakers of today and tomorrow. Among these issues, perhaps the most pressing ones can be seen evolving in the realm of Intellectual Property disputes, as well as in the question of assigning an appropriate and suitable legal personality to AI. Once we deep-dived into the intricacies of all of the aforementioned topics, it was important to have an in-depth look at the prevailing legal frameworks which aim to regulate the AI-powered technologies.
Therefore, it may be concluded that the prevailing legal scenario in which we live currently is being rapidly modernized and updated by the competent lawmaking authorities the world over.