Open Access Research Article
"UNVEILING BIAS: EXPLORING GENDER AND RACIAL DISCRIMINATION IN ARTIFICIAL INTELLIGENCE AND ITS IMPACT ON CONVICTIONS AND WORKFORCE EQUITY"
Published Paper
Article Details
"UNVEILING BIAS: EXPLORING GENDER AND RACIAL
DISCRIMINATION IN ARTIFICIAL INTELLIGENCE AND ITS IMPACT ON CONVICTIONS AND
WORKFORCE EQUITY"
AUTHORED
BY - JAISVICA KAUR BAGGA
& NIVEDHA K.V
Abstract
It
is indeed a scary thing to imagine that your next job opportunity might be
shaped by an unseen hand - computer algorithm. In the realm of Artificial
Intelligence, the invisible biases ingrained in algorithms wield significant
influence, particularly in matters of gender and racial discrimination. Brace
yourself for a journey into the heart of this digital labyrinth, where we
unveil the biases dictating not just employment prospects but also impacting
legal convictions and workforce equity. In an era driven by technology
breakthroughs, the widespread use of Artificial Intelligence (AI) has prompted
worries about the presence of bias, particularly in terms of gender and race
discrimination. As AI algorithms progressively affect decision-making
processes, a rigorous assessment of the biases inherent in these systems is
required to build a fair and inclusive society.
The
discernible bias intrinsic to artificial intelligence within gender-centric
domains, particularly in the realm of occupational settings and employment
opportunities, has become a focal point of this paper's scrutiny. A plethora of
empirical studies elucidate that AI algorithms are prone to perpetuating gender
stereotypes, thereby engendering discriminatory outcomes within the ambit of
recruitment processes. Legal perspectives underscore the imperativeness of
incorporating ethical AI practices, as underscored by instances such as the
Amazon hiring tool, which exhibited a proclivity toward favoring male
candidates. Illustrative case scenarios, elucidating instances of gender-based
discrimination experienced by female job seekers attributable to biased
algorithms, serve to accentuate the exigency of addressing this multifaceted
issue.
This
paper directs its attention towards delineating the accurate assessment of the
correlation between AI bias and the maltreatment of women within workplace
contexts. The primary objective is to underscore the imperative for heightened
ethical standards in AI to mitigate the aforementioned repercussions.
Furthermore, the paper seeks to elucidate strategies for minimising this
disparity, thereby contributing to the advancement of equitable practices in
artificial intelligence.
Keywords - Algorithm, Artificial intelligence, Bias, Employment,
Discrimination, Workplace
A woman's journey in the world of AI: An
introduction
The ever-evolving landscape of technology demands constant adaptation.
This will be especially true for women, who are always seated in areas of
conflict and vulnerability. As the quote by Anu Madgavkar suggests, "Women
are going to be among the groups who are going to have to learn how to adapt
and use many of these technologies”. This underscores the significance of
comprehending the perpetually shifting technological landscape and
acknowledging the unique challenges confronted by women in the tech sphere. In
light of the above statement and considering the exponential growth of the
global artificial intelligence (AI) market, virtually all sectors are poised to
experience significant disruption due to the implementation of AI systems. This
disruption will manifest in the digitization and automation of work processes,
leading to transformative shifts within labor markets. Ultimately it can be
expected that the impact will be multifaceted, encompassing the number and
profiles of available jobs across industries, as well as the evolving skill
sets required of the workforce. This paper adopts a gender-based analytical
framework to examine the potential for a vicious cycle of digital gender
inequality to emerge[1].
This cycle is hypothesized to be driven by the interplay between pre-existing
gender stereotypes and occupational segregation within the workforce, alongside
the digitalization and automation spurred by AI implementation. The objective
is to offer insights into the gender-specific ramifications of AI technologies.
This knowledge is critical for mitigating the potential for social
stratification and marginalization.
Dr. Muneera Bano underscores the risk of
perpetuating sexism through AI systems, as they can mirror and perpetuate
biased patterns encoded in their data as she put forths by stating that “If
there is sexism embedded within the data, they will pick up that pattern and
exhibit the same sexist behavior in their output. And unfortunately, the
workforce in AI is male dominated.”
It does not come as a surprise that there are
concerns regarding the potential for AI to perpetuate sexism. Datasets used to
train AI may contain inherent gender biases, which the AI can then learn and
replicate. Furthermore, the male-dominated nature of the AI workforce
raises concerns that these biases might go unnoticed. From a legal
standpoint, discriminatory outcomes based on gender could violate existing
anti-discrimination laws for example a scenario where the translation of
the English term "nurse" consistently utilizes a feminine noun, while
the term "doctor" is translated using a masculine noun. This
exemplifies the potential for AI to perpetuate pre-existing societal biases
regarding gender roles within professions. Such biases could have far-reaching
consequences, including the exacerbation of the gender gap in certain fields
and the potential for negative downstream impacts, such as the endangerment of
women's lives if, for example, a medical AI system exhibits gender bias in its
recommendations. The birthplace of a lot of these biases are simply the
stereotypes and normalized norms that exist in society at large.
The role of gender stereotypes in
shaping AI culture
It is common knowledge that Stereotypes, both
negative and positive, exert significant influence on individuals'
consciousness, opinions, judgments, attitudes, and behaviors. Gender-based
stereotypes persist in society and contribute to workplace gender inequality
and likewise unconscious biases shape one's feelings and thoughts toward
certain topics or individuals, often leading to negative associations and
decision-making. Implicit biases, influenced by experiences, environments, and
societal norms, are universal and can affect everyone[2].
These biases can be mitigated through de-biasing techniques, such as changing
the way information is presented or providing tools to overcome unconscious
predispositions. In the realm of artificial intelligence (AI), biases are
increasingly problematic, particularly when algorithms trained with biased data
result in algorithmic discrimination. Underrepresented datasets can lead to
biased samples, reinforcing inequality, and perpetuating unfair treatment,
especially in critical areas like healthcare and employment. The prevalence of
gender biases in AI algorithms impacts various aspects of society, from search
engine results to hiring decisions. Despite efforts to reduce biases,
algorithmic discrimination persists, contributing to gender disparities in the tech
industry and beyond. The deep rootedness of this must be analyzed and one way
or another it always goes back to the societal perception of women and races.
Representation of gender in society is deeply entrenched in cultural norms and
expectations. Men are often portrayed as powerful and dominant, while women are
expected to conform to traditional femininity, perpetuating gender stereotypes
and reinforcing unequal power dynamics. Challenging these stereotypes and
promoting diverse representations is crucial for fostering inclusivity and
dismantling gender-based discrimination. One such manifestation of the
society's male centric gaze and an aggregate of patriarchal ideas resulted in
the fantasization of Sophia, a humanoid robot.
The development of Sophia, a humanoid robot created
by Hanson Robotics, raises questions about the feminization of AI and its
implications. While Sophia is designed to resemble a female figure, her
capabilities and functionalities primarily revolve around human-like
interactions and expressions rather than gender-specific traits. However, the
case study of Sophia also sheds light on the potential risks associated with
hypersexualization and gender stereotypes in AI. Sophia's appearance, modeled
after Audrey Hepburn, incorporates feminine features such as expressive eyes, a
slender smile, and delicate facial features. While this aesthetic choice may
aim to make Sophia more relatable and appealing to humans, it also reflects
societal norms and expectations regarding female beauty standards.
Furthermore, Sophia's interactions, including
banter and wisecracks, are programmed responses aimed at mimicking human
conversation and emotions. However, the extent to which these interactions
reinforce gender stereotypes or contribute to the objectification of women
depends on the context and intention behind Sophia's design and programming.
The hypersexualization of Sophia, or any AI with a female persona, can
perpetuate harmful stereotypes and undermine efforts towards gender equality.
By portraying female AI as subservient, passive, or primarily focused on
pleasing others, there is a risk of reinforcing traditional gender roles and
expectations. Moreover, the lack of agency or autonomy in Sophia's
decision-making process raises ethical concerns about the implications of
feminized AI. If Sophia's responses are primarily designed to cater to male
fantasies or expectations, it could contribute to the objectification and
dehumanization of women, even in non-human forms. In conclusion, while Sophia represents
an impressive advancement in robotics and AI technology, her case study
highlights the need for careful consideration of the implications of
feminization in AI development. By promoting diversity, inclusivity, and
ethical standards in AI design and programming, we can mitigate the risks of
perpetuating harmful stereotypes and biases in AI systems.
A roadmap to eliminating gender
bias in AI
Over the years AI technology has become a powerful
driver of innovation, impacting everything from entertainment to healthcare.
However, recent events have shed light on potential legal concerns surrounding
bias in AI. The concern lies in the possibility that AI systems, particularly
those relying on machine learning, can inherent biases from the data they are trained
on. This raises the possibility of unfair or discriminatory outcomes, depending
on the application of the AI[3].
Fortunately, researchers have developed techniques to address this issue. These
techniques focus on different stages of the AI development process. Some
methods involve pre -processing the data used to train the AI, while others
involve modifying the algorithms during training itself. Finally, there are
techniques that can be applied after the training data has been processed. By
implementing these bias mitigation techniques throughout the development
process, organizations can help ensure their AI systems are fair and comply
with evolving legal standards. Machine learning models can be trained in ways
that encourage fairness and minimize prejudice. Techniques used during
training, known as in-processing models, provide unique chances to accomplish
this. One technique uses a concept known as regularization. This prevents the
model from becoming unduly reliant on specific features in the training data,
such as a person's race, which may be irrelevant to the main task at hand. For
example, in a university admissions situation, the model should prioritize
characteristics such as GPA and test scores over the applicant's race.
Regularization helps to prevent the model from overfitting the training data
and prioritizing irrelevant elements. Adversarial training is another strategy
used with in-processing models[4].
This strategy pits two models against one another during training[5].
One model focuses on the primary goal, such as deciding on university
admissions. The second model attempts to forecast a sensitive attribute, such
as race, based on the first model's decisions. If the second model accurately
predicts race, the first model is penalized. This inhibits the primary model
from using race or other sensitive information to make decisions, resulting in
more equitable outcomes.
To effectively analyze the trajectory of adverse bias within Artificial
Intelligence (AI) systems, a fundamental understanding of certain core AI
concepts is necessary. One such concept critical to this examination is
propensity data modeling. Propensity modeling encompasses a collection of
techniques employed to construct models capable of predicting the likelihood of
an individual undertaking a specific action. These models achieve their
predictive power by analyzing historical behavioral patterns within a
designated target audience. In simpler terms, imagine you want to advertise a
new fitness program in this scenario, propensity modeling can help us identify
people who have shown interest in fitness activities in the past, making them
more likely to be interested in a program. However, the seemingly innocent
nature of this example transforms significantly when applied to broader
contexts, particularly those with the potential to significantly impact
individuals' lives. For instance, a financial institution might utilize propensity
modeling to make loan approval decisions. If the model is trained on historical
data exhibiting a gender disparity where men have a higher application and
approval rate for loans, a concerning outcome may emerge. The model,
conditioned on this biased data, could predict that future male applicants are
more likely to be successful borrowers. This scenario has the potential to
perpetuate discriminatory practices and lead to the unjustified rejection of
loan applications submitted by qualified women. This example underscores the
critical need to acknowledge the potential for historical behavioral data to
perpetuate existing societal biases within AI systems. Mitigating such biases
necessitates a comprehensive understanding of the underlying mechanisms at play
and the implementation of robust safeguards within the development and
deployment of AI technologies. In a case study conducted on
amazon's recruitment model it was seen that an attempt to leverage machine learning for recruitment has raised concerns
about perpetuating gender bias. The model, trained on historical applicant
data, mirrored the existing imbalance in the tech industry, favoring resumes
with characteristics historically associated with male candidates. This bias
stemmed from the inherent lack of gender neutrality in the training data,
reflecting the real-world dominance of men in tech roles. While Amazon claims
to have addressed this by modifying the algorithms, the possibility of new
biases emerging based on keywords or other resume features remains. This case
highlights the challenges of achieving gender based fairness in AI-powered
hiring systems. Machine learning expert Assistant Professor Nihar Shah
emphasizes the ongoing need for development in this area, showcasing the
complexities of eliminating bias entirely from algorithmic decision-making. The
jarring gender imbalance inadvertently manifested in Amazon's initial attempt
at an AI-powered recruitment system. The model, trained on existing data,
learned to penalize resumes containing the word "women's,"
highlighting the potential for bias within such algorithms. Recognizing this
issue, Amazon's researchers implemented a more refined approach. They created a
collection of 500 specialized models, each focusing on specific job functions
and locations[6].
These models were trained to analyze a vast array of parameters (over 50,000)
extracted from applicant resumes. Notably, the algorithms were programmed to
de-emphasize the significance of generic skills commonly listed by many
applicants, such as programming languages or frequently used platforms. This
shift towards targeted parameters and the downplaying of generic skill sets
demonstrates a positive step by Amazon toward achieving fairness and mitigating
potential gender bias in its recruitment process[7].
The case emphasizes the crucial need for constant vigilance in identifying and
addressing potential biases within AI systems. This commitment is essential for
ensuring the responsible development and deployment of AI technology in our
society. Despite efforts to achieve gender parity, leading U.S. tech companies
continue to struggle with a significant gender gap, particularly in technical
roles like software development where men remain vastly overrepresented. The
ethical models of a few leading tech companies are analyzed below to gain a
more nuanced and informed look into this specific matter.
Ethical AI models: A look into
industry pioneers
As most of us are aware, Microsoft and
Google are at the forefront of AI innovation, and their initiatives reflect a
commitment to ethical considerations in AI development. Let's delve deeper into
specific examples from both companies and explore how an Ethical Impact
Assessment (EIA) might apply to these cases.
Ece Kamar is a senior researcher in Microsoft's
research lab in Redmond, Washington. Her research focuses on AI tools that
assist engineers in identifying blind spots in training data, such as the
underrepresentation of darker-skinned women, which can lead to AI systems with
unacceptable mistake rates on gender classification tasks[8].
A more sophisticated difficulty, she added, is figuring out how and when to
intervene and reduce AI systems that reflect and magnify societal biases, not
because datasets are incomplete or algorithms are insufficient, but because
human cultures are prejudiced. For example, if you search the internet for the
term "CEO," you will most certainly find information about top
leadership positions in companies and organizations all over the world, as well
as a few photographs, most of which are of men. This is not surprising given
that women account for less than 5% of Fortune 500 CEOs. At Microsoft, this has
prompted an ongoing investigation by engineers at the search engine Bing, in
collaboration with experts such as Wallach, into how to best surface results
that reflect the active discussion in boardrooms, academia, and on social media
about the lack of female CEOs - and efforts to change that.
Similarly over the course of this year, Google has
made an effort to promote justice and decrease prejudice in machine learning.
The most recent update in this effort addresses gender prejudice by offering
feminine and masculine translations of select gender-neutral words on the
Google Translate website. Google Translate learns from hundreds of millions of
previously translated instances from the web. Historically, it has offered only
one translation for a query, despite the fact that the translation could be
feminine or masculine. So, when the model produced one translation, it
unintentionally repeated existing gender biases[9].
For example, it would be masculine for words like "strong" or
"doctor," yet feminine for others, such as "nurse" or
"beautiful."
For every single term, such as "surgeon,"
you will now receive a translation in both the feminine and masculine forms
when translating from English into French, Italian, Portuguese, or Spanish.
When converting words and sentences from Turkish to English, you'll also receive
translations in both languages. For example, if you input "o bir
doktor" into Turkish, the gender-specific translations that appear are
"she is a doctor" and "he is a doctor."
As mentioned above companies are required to
compose an ethical framework to minimize the adverse effects of AI bias,
however companies can only go so far on its own, there comes a time where the
government will need to utilize its resources and intervene on this matter, as
the matter of gender disparity comes under the purview of governance. The imperative for global governance to deliberate
and intervene on the issue of AI bias in gender stems from the profound impact
that biased artificial intelligence systems can have on societies worldwide. A
world government intervention is essential to establish standardized ethical
frameworks and guidelines that ensure the development and deployment of AI
systems free from gender bias. By fostering international collaboration through
meetings and conventions governments can collectively address the challenges
associated with AI bias, implement corrective measures, and safeguard against
the amplification of gender disparities. This proactive approach not only
advances fairness and equity but also reinforces the commitment to harnessing
the potential of artificial intelligence for the betterment of global
societies.
The Intersection of AI and
Governance
On that note the annual meeting of the Commission
on the Status of Women (CSW) is a significant event that brings together
activists, advocates, experts, and government representatives from across the
globe. Taking place in New York, CSW serves as a pivotal platform for
stakeholders to assess progress on gender equality, reaffirm commitments, and
strategize for the future. CSW67, scheduled for March 6-17, 2023, has
identified "Innovation and Technological Change, and Education in the
Digital Age for Achieving Gender Equality and the Empowerment of all Women and
Girls" as its priority theme. Despite substantial strides in integrating
gender equality into international discourse and action, women and girls
continue to face entrenched disparities and discrimination worldwide. This
session of CSW aims to address these challenges by focusing on the intersection
of innovation, technology, and gender equality.
The theme underscores the imperative of ensuring
that rapid technological advancements benefit women and girls equitably, rather
than exacerbating existing inequalities. Emphasizing a human-centric approach
to digitalization, CSW67 seeks to embed feminist principles of inclusion and
intersectionality into technological innovation. Central to this effort is the
need to close the gender gap in digital access, foster inclusive innovation
ecosystems, and eliminate technology-facilitated gender-based violence. Highlighting
the economic ramifications of excluding women from the digital sphere, CSW67
underscores the importance of bringing women closer to technology. Failure to
do so not only stifles creativity and innovation but also incurs significant
economic costs, as evidenced by the substantial GDP losses attributed to
women's digital exclusion. The official documentation of CSW67 includes the
Report of the Secretary-General, which provides a comprehensive analysis of
harnessing technology in the digital age to advance gender equality. The report
emphasizes the transformative potential of the digital age for women's
empowerment while underscoring the imperative of prioritizing The annual
meeting of the Commission on the Status of Women (CSW) is a significant event
that brings together activists, advocates, experts, and government
representatives from across the globe. Taking place in New York, CSW serves as
a pivotal platform for stakeholders to assess progress on gender equality,
reaffirm commitments, and strategize for the future. CSW67, scheduled for March
6-17, 2023, has identified "Innovation and Technological Change, and
Education in the Digital Age for Achieving Gender Equality and the Empowerment
of all Women and Girls" as its priority theme. Despite substantial strides
in integrating gender equality into international discourse and action, women
and girls continue to face entrenched disparities and discrimination worldwide.
This session of CSW aims to address these challenges by focusing on the
intersection of innovation, technology, and gender equality.
The theme underscores the imperative of ensuring
that rapid technological advancements benefit women and girls equitably, rather
than exacerbating existing inequalities. Emphasizing a human-centric approach
to digitalization, CSW67 seeks to embed feminist principles of inclusion and
intersectionality into technological innovation. Central to this effort is the
need to close the gender gap in digital access, foster inclusive innovation
ecosystems, and eliminate technology-facilitated gender-based violence.
Highlighting the economic ramifications of excluding women from the digital
sphere, CSW67 underscores the importance of bringing women closer to
technology. Failure to do so not only stifles creativity and innovation but
also incurs significant economic costs, as evidenced by the substantial GDP
losses attributed to women's digital exclusion. The official documentation of
CSW67 includes the Report of the Secretary-General, which provides a comprehensive
analysis of harnessing technology in the digital age to advance gender
equality. The report emphasizes the transformative potential of the digital age
for women's empowerment while underscoring the imperative of prioritizing.
Algorithms of oppression
In Europe, a group of experts advised the European
Commission that AI systems should treat everyone fairly and avoid
discrimination. They also said the data used to train these systems should be
inclusive and represent a variety of people. The European Commission is also
working on laws for AI that would balance innovation with protecting people's
rights. These laws would make sure AI developers and users are held accountable
for their creations. In the UK, a government advisory group suggested that
existing principles for ethical conduct in public service should also apply to
AI. They recognized that AI can have unforeseen consequences, so they proposed
a flexible approach to using these principles with AI.A famous study on the
'Algorithms of Oppression' (2018) asserts that "artificial intelligence
will become a major human rights issue in the twenty-first century". In
this way, national and international institutions have already begun to develop
policies and procedures for identifying and mitigating biases. In 2019, the
European Commission's independent High-Level Expert Group on Artificial
Intelligence published a study titled 'Ethics Guidelines for Trustworthy AI'.
The document supports "equality, non-discrimination, and solidarity"
as a fundamental right, emphasizing the importance of ensuring that systems do
not produce unfairly biased outcomes, such as employing inclusive data that
represents different populations. The European Commission also intends to
establish a legal framework for AI that is consistent with the EU Charter of
Fundamental Rights and strives to define the duties of users and providers
(2022).This charter seeks to strike a balance between the creativity and
benefits of new AI technologies and the fundamental rights of EU people, while
also aligning them with EU ideals. It offers a 'risk-based' approach, with the
goal of prohibiting artificial intelligence practices that are contrary to EU
rules and principles. Although the majority of the case studies in this paper
focus on certain characteristics of AI bias, the legal framework seeks to
prevent genuine biases in decision-making that may damage its inhabitants. A
similar perspective was expressed in a review of Artificial Intelligence and
Public Standards by the Committee on Standards in Public Life, an independent
organization that advises the UK government. Their paper (2022) concludes that
the current 'Seven Principles of Public Life' (selflessness, integrity,
objectivity, accountability, openness, honesty, and leadership) should be used
as a guideline for integrating AI technologies into public life. Understanding
that AI may have far-reaching and unanticipated consequences, the review
presents a rough overview of how the Seven Principles can be translated into
practice for the use of AI[10].
Overall, it is obvious that current policy recommendations for the regulation
of AI focus on general concepts and guidelines, reflecting the ongoing and
developing spectrum of difficulties that may require attention.
Furthermore It should be underlined that AI bias is
not a hypothetical or theoretical concept. In industry and government,
algorithms that provide healthcare, assist with employment, lead police, and
establish creditworthiness have shown uneven treatment and impact on
identically situated individuals, locations, and objects. As highlighted
throughout the study, examples of human language models with a bias against
women and persons with disabilities have appeared in recent years; the accuracy
of some well-known speech recognition and facial recognition algorithms is
skewed against African Americans. Even when unintentional, algorithmic bias can
result from using unrepresentative training data or biased historical data, or
from failing to correct statistical biases. Algorithmic prejudice can
occasionally lead to unlawful discrimination in addition to being immoral.
However, there is no reason why this new technology must take a bad turn. In
order to use algorithms for efficiency and good change, the government will be
crucial in minimizing algorithmic bias as both a consumer and a regulator.
Governments have significant market power and influence over numerous
significant algorithmic use cases as consumers of algorithms. Governments can
play this role by establishing guidelines, norms, and policies that lessen
algorithmic prejudice. Prior to implementation, the public sector should win
over the people's trust with algorithmic use cases and make the use of AI in
choices that have a major impact on individuals transparent. All UK agencies
are required by law to document any possible or predicted algorithmic
discrimination before it is used. This obligation cannot be delegated. Impact
assessments should be used to examine the usage of algorithms, just like with
any other policy decision. Analyzing real-world performance across demographic
groupings is important because data interpretations and algorithms can have
unanticipated and unwanted consequences.[11]
The public sector should apply a consistent national approach while taking
local demands into account. The method of detecting AI bias will involve
formulating AI requirements, investigating how to construct compliant
algorithms, and then enforcing those criteria. Governments can utilize
verification, audit, and certification to ensure compliance, as well as risk
assessments, audits, and continuing testing. Requirements will differ depending
on industry and setting; for example, in contexts with considerable uncertainty
and variance, some differential influence may be tolerated, but in others it
must be avoided. Guidelines must mix principles and exact thresholds.
Algorithms require a fundamental fairness test to avoid severe disparities in
impact. Guiding principles can help to prevent prejudice in critical instances
that disparate impact studies may ignore. To design and implement regulations,
governments must also increase their technological skills.
Beyond the binary - Conclusion
Gender prejudice has been demonstrated at many
levels across society. Women encounter difficulties beginning with early
education and continuing through various stages of employment; for example,
STEM-related jobs have significant impediments at the executive level. As a
result, it is critical to evaluate developing technologies that may perpetuate
societal bias. This study provides a better understanding of gender bias in AI
and shows evidence from the diverse literature that gender bias exists in AI
systems. As a result, the first step in addressing gender bias in AI is to
investigate the underlying causes of this bias that function as a catalyst in
AI systems. This study not only argues for the concept of justice in gender
bias in society as a whole, but it also paves the way for other researchers and
users to gain a better understanding of the notion of gender bias in AI, its
contributing factors, and potential mitigation approaches as reported in the
current multidisciplinary literature. Fair decisions are seen to have a
substantial impact on an individual's contentment and loyalty to a specific
product, company, or relationship. The purpose of this research is to examine
the elements that contribute to gender bias in AI, as well as the ways that
could be used to mitigate gender bias. We expect this research to lay the
groundwork for new paradigms that, as our literature survey indicates, are
still very much required.
After investigating the role of bias in AI, the
deep rootedness of its existence, and means to minimize it through the use of
ethical machine learning models supplemented by the helping hands of good
governance, it is clear that the primary goal of avoiding gender bias in AI is
not only to bring women into equal footing, but also to enable and encourage
future participation of women in STEM. In the symphony of technological progress,
the inclusion and empowerment of women are not just ethical imperatives; they
are the keystones to harmony and innovation. As we explore the limitless
possibilities of artificial intelligence, it is critical that women are active
participants in defining the algorithms that define our digital future. The
collaborative brilliance of various minds drives remarkable advancements while
also ensuring that technology's ethical compass remains true. Empowering women
in technological settings, particularly in the field of artificial
intelligence, is more than just a matter of equality; it is a deliberate
decision to unleash the whole range of human potential and creativity. The
future of technology must be laced with threads of inclusivity, with women
serving not only as beneficiaries but also as architects, guiding us toward a
future in which invention recognizes no gender boundaries. Overall, it must be
noted that a limitation to this paper is that current efforts to regulate AI
focus on broad ideas and guidelines, because the technology is still developing
and new issues might arise in the future.
[1] Mirzazadeh I, “Artificial Intelligence (AI) and Violation of Human
Rights” [2023] SSRN Electronic Journal
[2] Rifat Ara Shams, Didar Zowghi and Muneera Bano, “AI and the Quest
for Diversity and Inclusion: A Systematic Literature Review” [2023] AI and
Ethics
[3] World Health Organization. (2021). Ethical use of artificial
intelligence: principles, guidelines, frameworks and human rights standards. In
WHO Consultation Towards the Development of guidance on ethics and governance
of artificial intelligence for health: Meeting report Geneva, Switzerland, 2–4
October 2019 (pp. 8–11). World Health Organization.
http://www.jstor.org/stable/resrep35680.8
[4] Hershock, P. D. (2020). Humane Artificial Intelligence: Inequality,
Social Cohesion and the Post Pandemic Acceleration of Intelligent Technology.
East-West Center. http://www.jstor.org/stable/resrep25513
[5]Agrawal, A., Gans, J. S.,
& Goldfarb, A. (2019). Artificial Intelligence: The Ambiguous Labor Market
Impact of Automating Prediction. The Journal of Economic Perspectives, 33(2),
31–50. https://www.jstor.org/stable/26621238
[6] Agrawal, A., Gans, J. S., & Goldfarb, A. (2019). Artificial
Intelligence: The Ambiguous Labor Market Impact of Automating Prediction. The
Journal of Economic Perspectives, 33(2), 31–50. https://www.jstor.org/stable/26621238
[7] Herrbach, O., & Mignonac, K. (2012). Perceived Gender Discrimination
and Women’s Subjective Career Success: The Moderating Role of Career Anchors. Relations
Industrielles/Industrial Relations, 67(1), 25–50. http://www.jstor.org/stable/41634302
[8] Farris, C., Jaycox, L. H., Schell, T. L., Street, A. E., Kilpatrick, D.
G., & Tanielian, T. (2015). Sexual Harassment and Gender Discrimination
Findings: Active Component. In T. L. Schell, A. R. Morral, & K. L. Gore (Eds.),
Sexual Assault and Sexual Harassment in the U.S. Military: Volume 2.
Estimates for Department of Defense Service Members from the 2014 RAND Military
Workplace Study (pp. 31–54). RAND Corporation.
[9] Maas, V. S., & Torres-González, R. (2011). Subjective Performance
Evaluation and Gender Discrimination. Journal of Business Ethics, 101(4),
667–681
[10] BOBBITT-ZEHER, D. (2011). GENDER DISCRIMINATION AT WORK: Connecting
Gender Stereotypes, Institutional Policies, and Gender Composition of
Workplace. Gender and Society, 25(6), 764–786
[11] Ridgeway, C. L. (1997). Interaction and the Conservation of Gender
Inequality: Considering Employment. American Sociological Review, 62(2),
218–235
Article Information
"UNVEILING BIAS: EXPLORING GENDER AND RACIAL DISCRIMINATION IN ARTIFICIAL INTELLIGENCE AND ITS IMPACT ON CONVICTIONS AND WORKFORCE EQUITY"
Authors: JAISVICA KAUR BAGGA, NIVEDHA K.V
- Journal IJLRA
- ISSN 2582-6433
- Published 2025/03/08
- Issue 7
About Journal
International Journal for Legal Research and Analysis
- Abbreviation IJLRA
- ISSN 2582-6433
- Access Open Access
- License CC 4.0
All research articles published in International Journal for Legal Research and Analysis are open access and available to read, download and share, subject to proper citation of the original work.
Disclaimer: The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of International Journal for Legal Research and Analysis.