Artificial Intelligence has Overstepped its Bounds every single citizen, with a vast network of cameras across the
country for different purposes
Outline
✓ AI tools being used to predict recidivism with racial bias due
1- Introduction to installed criminal justice algorithms by biased humans
Although artificial intelligence (AI) has made humans fairly Case Study: The COMPAS system explaining in its research,
dependent on it, it can never overstep its bounds, being nothing “Black defendants were twice as likely as white defendants to
more than a set of complex algorithms designed by humans to be misclassified as being a higher risk of violent recidivism. And
assist them in economic, political, and social sectors, white violent recidivists were 63 per cent more likely to have
ultimately depending on human decisions and intentions for its been misclassified as a low risk of violent recidivism, compared
outcomes. However, man, being a biased and selfish creature, with black violent recidivists due to miss installed algorithms.”
is utilizing machines for doing unethical tasks that need to be
bridled to harness the maximum dividends of machine ✓ AI systems exhibiting bias and discrimination learning from
intelligence. historical data installed in them
2- Understanding the Term Artificial Intelligence and its Case in Point: In 2018, Amazon shutting down an AI-powered
significance recruiting tool because it was trained to be biased against
female candidates.
✓ The simulation of human intelligence processes by machines
4- Factors Responsible for the Limitations of AI
3- The Evidence Explaining How Artificial Intelligence has not
overstepped its bounds ✓ Lack of Organic Reasoning
✓ AI diagnosing diseases under the supervision of human Case in Point: According to Richard Yonck, a futurist and author,
doctors for interpreting results and deciding on treatment Case “Even if we manage to build artificial general intelligence, its
in Point: IBM’s Watson for Oncology, while providing reasoning can never fully align with human values. Nonhuman
recommendations for cancer treatment, ultimately relying on perspectives are only a benefit if they inform rather than impose
human doctors to make the final decision decision making.”
✓ Autonomous devices facing ethical dilemmas due to the lack ✓ Installation of Limited Algorithms
of human judgment capability in them
Case in Point: Chatgpt, the most advanced AI software,
Case in Point: The “trolley problem” in self-driving car apologizing on every other question having a limited database
development that killed a pedestrian demonstrating the lag at
humans’ end while programming AI to make ethical choices in ✓ Promulgation of Pre-emptive Regulation Frameworks
uncertain situations.
Case in Point: Several organizations, like OpenAI, NIST, and the
✓ AI algorithms in stock trading and financial predictions Australian government, setting regulatory frameworks for data
depending on humans to make strategic decisions privacy, transparency, accountability, and fairness
Case in Point: The “Flash Crash” of 2010, where stock prices 5- Critical Analysis – the implication if AI oversteps its bounds
plummeted in minutes due to high-frequency trading
algorithms requiring the immediate intervention of human 6- Conclusion
regulators to stabilize the market.
Essay
✓ The use of AI in military applications subjecting to human
decisions and ethics Artificial Intelligence (AI) is one of the most transformative
technological innovations of the 21st century. With its
Case Study: Machine gun with AI being used to kill Iran scientist remarkable capacity to analyze massive amounts of data and
Mohsen Fakhrizadeh via a remote controlled by the human learn from experience, AI has the potential to reshape
intelligence of Israel numerous aspects of human life, pushing the boundaries
beyond possibilities. However, even with such colossal
✓ The use of AI for over-surveillance manifesting governments’ advancement, the technological marvel remains inherently
intentions of regulation reliant on human oversight and guidance, making it under
humankind’s bounds. In other words, it is nothing but a
Case in Point: China’s facial recognition system logging nearly testament to the power of human ingenuity. As demonstrated in
the responses made by ChatGPT, one of the most viral AI have been responsible for killing a pedestrian on the road
technologies of today’s world, “I do not have the ability to run despite the fact it was built per traffic rules and regulations and
independently. My responses are generated based on patterns even the vehicle following all the directions installed. However,
and information present in the data I was trained on, but I lack the issue was the folly of trolleys that emerged during self-
consciousness, self-awareness, or the ability to function driving car development, highlighting the challenges associated
autonomously.” So, from diagnosing diseases to dealing with with programming AI to navigate such ethical choices in real-
ethical dilemmas, automation has remained dependent on world, uncertain scenarios. Thus, ongoing human oversight is
humans for input or output. Likewise, when AI makes blunders necessary to deal with such intricacies of autonomous
in performing tasks it is given to, real intelligence of humans has technologies.
to interfere to undo the hazards. Unfortunately, humans, being
biased and selfish creatures, have been utilizing machines for Another area where Artificial Intelligence cannot function
unethical tasks, like invasive surveillance through face- without human intervention for strategic decision-making is in
recognition tools, privacy violations and cyber-attacks, the context of stock trading and financial predictions. The Flash
inculcation of discrimination in artificial brains, and even killing Crash of 2010 serves as a compelling example in this regard. At
their fellow men. Although various ethical and regulatory the event, stock prices experienced a sudden and severe
frameworks have been made to curb the misuse of machines plummet within minutes, leading to market manipulation and
based on the nefarious intentions of humans, ongoing efforts economic turmoil. Since algorithms lacked the judgment and
are needed to harness the maximum dividends of machine foresight to prevent the setback, human regulators had to
intelligence. swiftly step in to stabilize the market. On the other hand, with
the investigation, it was made clear that, ultimately, the human
Before delving into the practicality, the term artificial mind of Navinder Singh Sarao, a London-based point-and-click
intelligence (AI) theoretically refers to the simulation of human trader, played an alleged role in the crash. Thus, AI, while
intelligence processes by machines. In other words, it is about powerful in processing vast data and executing trades, relies on
developing systems that include learning- the acquisition of human expertise to manage catastrophic disruptions.
data and rules for using the data, reasoning- the use of the rules
to reach approximate or definite conclusions and problem- Moreover, the use of AI in military applications, while
solving. In recent years, the field has become essential to showcasing its potential, has remained subject to human
human survival. It has gained significant importance in various decisions and ethical considerations. A glaring example in this
fields, including healthcare, transportation and regard is the targeted killing of Iranian scientist Mohsen
communication, and the economy, due to virtually unlimited Fakhrizadeh in which a machine-gun with AI was employed,
accessibility to computing power and the decreasing costs of remotely controlled by human intelligence, specifically by
data storage. All in all, AI is like a kaleidoscope revealing a Israel. The incident underscores that AI, even in military
dazzling array of patterns and possibilities using human-like contexts, is ultimately directed and overseen by human
intelligence. operators who make decisions and assess ethical implications.
Therefore, AI, along with enhancing military capabilities and
Artificial Intelligence has undoubtedly made significant strides, precision, hinges on the ethical and legal frameworks set by
but it is imperative to acknowledge that it has not overstepped human authorities.
its bounds in any area. One compelling piece of evidence lies in
the occurrence of AI systems diagnosing diseases. The Not only this, the use of AI for extensive surveillance raises
systems, while becoming increasingly sophisticated, are concerns about governments’ intentions and the need for
employed under the supervision of human doctors who regulation. For example, China widely uses a facial recognition
interpret the results and make the ultimate decisions regarding system, which logs data on nearly every single citizen. The
treatment. For instance, IBM’s Watson for Oncology provides country has deployed an extensive network of cameras for
recommendations for cancer treatment based on vast various purposes, from public safety to monitoring social
datasets, and medical knowledge always relies on human behaviour. While AI-driven surveillance can have legitimate
doctors to make the final decision. Thus, the relationship uses, such as crime prevention and public safety, the scale and
between humans and AI being more symbiotic than parasitic is scope of China’s surveillance system underscore the potential
a testament to ensuring that crucial medical decisions remain for abuse and privacy infringements. Likewise, many
in the capable hands of healthcare professionals while independent organizations do this for personal gain. Thus, these
harnessing the power of AI to assist in the diagnostic and are humans, not the machines or systems, who are responsible
treatment process. for the nefarious, unethical surveillance of the masses, invading
their privacy.
Moving further, artificial intelligence always requires human
judgment ability to encounter ethical dilemmas stemming from Likewise, AI tools used to predict recidivism in the criminal
them. In fact, machines are prone to make blunders only when justice system often suffer from racial bias, not because
loopholes are left at human end while programming a particular technology is biased but primarily because they rely on
technology. For example, autonomous self-driving vehicles algorithms that have been shaped by biased human input. A
significant case study in this regard is the COMPAS system, despite its impressive capabilities, is still bound by the
which has gained notoriety for its racial disparities. Research on constraints of its programming and data availability. The
the COMPAS system revealed, “Black defendants were twice as limitations serve as a reminder that AI remains a tool developed
likely as white defendants to be misclassified as being a higher and controlled by humans, and its knowledge and responses
risk of violent recidivism. And white violent recidivists were 63 are based on existing data up to its last training point. AI’s
percent more likely to have been misclassified as a low risk of inability to access real-time, dynamic information and its
violent recidivism, compared with black violent recidivists due potential to provide incorrect or incomplete responses
to mis-installed algorithms.” The instance serves as a stark underscore the importance of human oversight and expertise. It
reminder that the biases inherent in AI systems can perpetuate is an essential factor in ensuring responsible and accurate use
existing racial disparities and further exacerbate issues in the of AI and preventing it from overstepping its bounds, especially
criminal justice system. It underscores the need for greater in contexts that demand precise and up-to-date information.
transparency, oversight, and ethical considerations in the
development and implementation of AI tools within the legal Another key reason why Artificial Intelligence has not
framework to ensure fairness and equitable outcomes for all overstepped its bounds is the establishment of regulations and
individuals involved. ethical frameworks. Numerous organizations, including
OpenAI, the National Institute of Standards and Technology
AI systems often exhibit bias and discrimination, raising (NIST), and the Australian government, have recognized the
concerns about their ethical implications. However, the fact necessity for guiding principles in AI development and usage.
repeats itself here that machines only work the way they are These regulatory frameworks address key aspects like data
trained. For instance, Amazon’s AI-powered recruiting tool hired privacy, transparency, accountability, and fairness,
employees on the basis of gender discrimination. The system, emphasizing the need to prevent AI from overreaching and
developed to assist in the hiring process, was found to be biased infringing on individual rights or perpetuating bias. Hence, the
against female candidates. The incident serves as a stark proactive approach to AI governance underscores the
reminder of the limitations and pitfalls of AI when it keeps importance of human oversight and ethical considerations in
receiving relevant prompts because it inherits biases from the continued development and deployment of AI technology,
historical data. Thus, it is necessary that unbiased and just thus ensuring that it remains a valuable tool rather than a force
humans must be employed to design such software on the one that oversteps its limitations.
hand, and regular auditing must be done to filter biased
prompts from the history of machines to avoid blunders. All in Nonetheless, if Artificial Intelligence were to overstep its
all, AI needs human oversight and ethical guidance to ensure bounds, it could lead to grave ethical, legal, economic, and
that technology is harnessed responsibly and equitably. safety consequences. Ethically, AI’s misuse could infringe upon
civil rights and values, while legal issues may arise from
After understanding the ultimate human responsibility of all discriminatory algorithms and privacy infringements.
implications of artificial intelligence, the question arises as to Economically, overreaching AI might destabilize financial
why AI cannot overstep its bounds. Broadly, AI’s boundaries markets and exacerbate job displacement. Safety concerns
remain intact for several reasons, one being the absence of include physical threats in autonomous systems, while
organic reasoning. AI, as it stands, lacks the nuanced, context- cybersecurity risks would grow. Public trust and acceptance of
sensitive reasoning that humans possess. The limitation is AI could erode, hindering its positive potential. Therefore,
eloquently expressed by futurist and author Richard Yonck, ensuring AI remains within regulatory and ethical frameworks is
“Even if we manage to build artificial general intelligence, its essential to unleash its benefits while averting these potential
reasoning can never fully align with human values. Nonhuman harms.
perspectives are only a benefit if they inform rather than impose
decision making.” The sentiment underscores the idea that In conclusion, no matter how far AI continues to advance, it has
although AI can process vast amounts of data and make always remained under the intelligence of human minds. Being
predictions, it lacks the depth of human understanding, developed as a result of humans’ inquisitiveness, it cannot
empathy, and ethical reasoning that are critical in various overstep its bounds, assisting mankind both in both
decision-making scenarios. The need for human oversight and constructive and destructive intents. Therefore, rather than
the ability to interpret AI’s output within a broader ethical and fearing machines, more focus must be placed on bridling
social context ensure that AI, even at its most advanced, human intents who program the technology. Moreover, since AI
remains within defined boundaries, leaving ultimate judgment also learn from the prompts it receives – or the historical data –
and ethical considerations to human hands. from time to time, auditing must be done to filter the biased
approach so that a balance must be maintained between its
Moreover, AI has limited algorithms. Even the most advanced AI potential and the risks associated with overstepping its bounds.
software, such as ChatGPT, has its limitations. As an example,
ChatGPT keeps apologizing for every other prompt,
acknowledging its limited database when unable to provide
comprehensive responses to certain questions. Thus, AI,
