“Systems thinking is critical in developing solutions to sustainability
challenges”
Course: Research Methodology
Submitted to: Syed Arshad Hussain
Submitted by: Aliza Tariq
Ali Zubair
Nadir Rafique
Shahzaib Marsia
Gender Bias & Artificial intelligence
Introduction:
Artificial intelligence (AI) is the Science and Engineering domain concerned
with the theory and practice of developing systems that exhibit the
characteristics we associate with intelligence in human behavior, such as
perception, natural language processing, problem solving and planning,
learning and adaptation, and acting on the environment. Its main scientific
goal is understanding the principles that enable intelligent behavior in
humans, animals, and artificial agents. This scientific goal directly supports
several engineering goals, such as, developing intelligent agents, formalizing
knowledge and mechanizing reasoning in all areas of human endeavor,
making working with computers as easy as working with people, and
developing human–machine systems that exploit the complementariness of
human and automated reasoning. Here is a figure of what artificial
intelligence agent or an A.I agent requires to interreact within a given
environment.
Most of the current AI agents, however, do not have all the components from
Figure 1, or some of the components have very limited functionality. For example,
a user may speak with an automated agent (representing her Internet service
provider) that will guide her in troubleshooting her Internet connection. The agent
may have advanced speech, natural language, and reasoning capabilities, but no
visual or learning capabilities. A natural language interface to a data base may only
have natural language processing capabilities, while a face recognition system may
only have learning and visual perception capabilities.
Artificial intelligence researchers investigate powerful techniques in their quest for
realizing intelligent behavior. But these techniques are pervasive and are no longer
considered AI when they reach mainstream use. Examples include time-sharing,
symbolic programming languages (e.g., Lisp, Prolog, Scheme), symbolic
mathematics systems (e.g., Mathematica), graphical user interfaces, computer
games, object-oriented programming, the personal computer, email, hypertext, and
even the software agents. While this tends to diminish the merits of AI, the field is
continuously producing new results and, due to its current level of maturity and the
increased availability of cheap computational power, it is a key technology in
many of today’s novel applications.
It was also realized that building an intelligent agent is very difficult because the
cognitive functions to be automated are not understood well enough. This has led
AI researchers to focus on individual cognitive processes, such as learning, and on
studying elementary problems in depths, such as concept learning. The
consequence was the split of artificial intelligence into many different areas,
including knowledge representation, search, game playing, theorem proving,
planning, probabilistic reasoning, learning, natural language processing, vision,
robotics, neural networks, genetic algorithms, and others.
As technology is progressing at a rapid pace these A. I’s have already made into
the lives of people in the form of voice automated search engines such as Siri and
Alexa etc. Furthermore, some companies have decided to use chatterbots in
conducting preliminary interviews through chatbots to eliminate biases that often
occur in the hiring processes. Chatbots come with a set of advantages such as :
they can interview multiple candidates at once at whatever time they are
available,
they are quick and they can be used to facilitate the next steps of the process.
Although they are not a replacement for in-depth face-to-face conversation,
chatbots are one way to streamline the preliminary stages of the interview
process. For example, Talky Jobs qualifies potential candidates and flags
them for more in-depth interviews.
The bot never asks for race, gender or any other information not strictly
related to the position. This ensures that candidates make it to the next stage
of the process based solely on their credentials.
Chatbots can also be used in recruitment. MeetFrank is an anonymous chatbot that
connects talent with companies. Candidates only give their position, skills and
information about the type of work they are looking for. MeetFrank then shows
their credentials to companies who decide whether they want to pursue them based
on skills alone.
AI cannot see color: enhanced video analysis
To reduce unconscious bias and quantify soft skills, some companies are turning to
AIs like Knockri to screen and shortlist applicants. Candidates are filmed
answering questions given to them by their potential employer. Afterward, the
footage is scanned and analyzed by the artificial intelligence to determine the
candidate’s enthusiasm, empathy, honesty and communication skills. The AI has
been specifically programmed to not see race. This helps create a more equitable
hiring process while still determining which candidates are the best fit.
There is evidence to suggest that this could be the future of the interview. As
candidates can record themselves answering questions, the interview can take place
entirely at their earliest convenience. Knockri claims to lead to more diverse hires.
Some companies say it has helped them make more quality hires at a faster rate.
The AI also has the unintended consequence of creating socioeconomic diversity
as well. Larger companies often send recruiters to only the most prestigious
universities because of a finite amount of recruiting resources. Owing to the
process being streamlined with AI, large companies can now afford to cast a wider
net to include less expensive schools.
Things to consider
Artificial intelligence is not without its drawbacks. Amazon recently shut down its
AI system for being biased against women. The system reviewed job applications
and rated candidates from one to five stars. However, the resume data it was built
on came from those already hired at the company – which, at the time, were mostly
men. The AI subsequently learned to penalize female applicants.
This demonstrates just how easy it is to be unintentionally biased.
The greatest problem with artificial intelligence, therefore, is still the same core
problem: humans can be unreliable and are often bad at spotting subconscious
biases. As humans are the ones selecting the data and building these systems, it is
very possible that AIs could develop unforeseen and unintended flaws or blind
spots. Microsoft is currently developing an AI to check other AIs for potential
biases. This, of course, comes with its own set of challenges. It is difficult to build
an AI to check for blind spots if its creators themselves are not aware of them
already.
There is also a different, though unrelated problem. Although many new AI
systems are being developed for a variety of uses, very few people know how to
properly use and implement them. This will eventually create an AI skills shortage,
where the few people who know how to use these systems will be in very high
demand. Some experts predict AI will create 2.3 million jobs by 2020 – but the
question remains whether there will be anybody to fill that enormous demand.
There are a variety of causes for the AI skills gap. Some remain unconvinced about
artificial intelligence’s abilities and future, whereas others think companies will not
want to implement new technology because of the amount of time, commitment
and money involved. Forbes proposes that most colleges and university programs
simply cannot keep up because of the current rapid rate of innovation.
Artificial Genius is increasingly influencing the opinions and behavior of human
beings in day-to-day life. However, the over-representation of guys in
the format of these technologies ought to quietly undo a long time of advances in
gender equality. Over centuries, human beings developed critical principle to
inform decisions and avoid basing them totally on personal experience.
Gender Biasness:
Gender bias is one of the most common issues that are being faced around the
world. There are being many researches in order to sort out and overcome this
issue. Some places are known to be male dominant societies whereas a few places
are known to be feminist oriented societies.
However, computing device talent learns principally from watching records that it
is presented with. While a machine’s potential to technique large volumes
of records may additionally tackle this in part, if that statistics is weighted
down with stereotypical principles of gender, the ensuing application of
the science will perpetuate this bias. While some current research sought to take
away bias from realized algorithms, they largely ignore decades of lookup on how
gender ideology is embedded in language.
Awareness of this research and incorporating it into methods to computer gaining
knowledge of from text would help prevent the generation of biased algorithms.
Leading thinkers in the emerging area addressing bias
in synthetic intelligence are additionally principally female, suggesting
that those who are potentially affected through bias are more possibly to
see, recognize and attempt to resolve it.
Gender stability in desktop gaining knowledge
of is therefore fundamental to stop algorithms from perpetuating gender ideologies
that disadvantage women. There is a growing consciousness of the effects of bias
in machine learning.
For instance, in a machine used with the aid of judges to set parole,
the assessment of the likelihood of offending used to be located to be biased once
more black defendants. Facial recognition software embedded in
most clever phones additionally works best for these who are white and male.
Scoring systems, fueled by using probably biased algorithms, are more and
more being used to make selections about people’s lives in relation to finance, jobs
and insurance.
However, there is little consideration of the many years of research that exist on
the relationship between gender ideology and language. Incorporating gender
theory, in precise feminist linguistic theory, into
the approach to laptop mastering from textual data may stop getting to know of
gender bias and keep away from the want to alter the algorithms.
In this essay, we concentrate on power mechanisms underpinning the
production and use of ICTs, forms of discrimination and exclusion of
women from the ICT business, and the representation of gender in ICT
research.
Our aim is not solely to indicate underrepresentation of women in the
processes of ICT design and application, but also to emphasize the potential
of new technologies to overcome the patriarchal boundaries which we find
present in the literature on the subject of ICTs vis-a` -vis gender.
According to feminist scholarship traditional ethics ignores gender and the
moral experience of women by focusing on “traditional masculine ways of
ethical reasoning which look to individual, rationalistic, rule-based ethical
models”
It is also crucial to emphasize that it was not our objective in this essay to
verify
theoretical assumptions stemming from research on gender ethics, which
was briefly touched upon above, against the empirical material provided by
applications of the four ICTs under consideration but to scrutinize available
literature on the subject and on this basis arrive at some conclusions which
might be indicative of gender issues which border on ethical issues, largely
unnoticed by ICT designers and producers.
Therefore, the reference to gender ethics serves solely one objective: the
identification of the potential ethical issues that the application of the
discussed ICTs may bring about in the future.
Furthermore, when taking Butler’s (1990) idea of gender performativity into
account, it is essential to consider the ways in which AI machines will
perform gender. It is also intriguing to examine if there are possibilities of
creating an entity which is literally genderless.
This paper is part of the larger research project called ETICA (Ethical Issues
of Emerging Information and Communication Technologies
Summarizing, we would like to emphasize that the problem of the
underrepresentation of women in the field of ICT is characteristic, to a varying
degree, of all discussed here technologies. Women are considerably absent in ICT
industry, beginning from education and ending at consumption of its final products
(Kelan, 2007).
Therefore, we recommend to engage more women in computer science at all
educational levels, as well as in the process of ICT design, production and
application. However, there must be an awareness that women’s geo-political
locations, cultural settings, various factors of socio-cultural identification such as
race, ethnicity, class, age, disability, religion, and different structures of
subjectivities do have significant implications for women’s participation in the
creation and consumption of ICT products.
Only a sober and creative recognition of those variables can substantially improve
the present gender imbalance vis-a`-vis ICT. Furthermore, in the case of ICT
design and application
Conclusion:
There is an emerging focus on fairness in machine learning generally and it is
essential that women are at the core of who defines the concept of fairness.
Advancing women’s careers in the area of Artificial Intelligence is not only a right
in itself; it is essential to prevent advances in gender equality supported by decades
of feminist thought being undone. We recommend to take gender specificity into
consideration in order to overcome stereotypical constructions of femininity and
masculinity which tend to influence the ways in which technologies and their
consumers are symbolically represented.
The main goal of AI is to develop computational agents that exhibit the
characteristics we associate with intelligence in human behavior. Such an agent has
an internal representation of its external environment which is at their basis of its
reasoning abilities. It is highly desirable that the agent’s knowledge and reasoning
are understandable to humans, and the agent is able to explain its behavior, what
decisions it is making, and why. The agent may reason with data items that are
more or less in contradiction with one another, and may provide some solution
without having all the relevant data. The agent should be able to communicate with
its users, ideally in natural language, and it may continuously learn.
Why are intelligent agents important? Because humans have limitations that agents
may alleviate, such as limited attention span, ability to analyze only a small
number of alternatives at a time, or memory for details that is affected by stress,
fatigue or time constraints. Humans are slow, sloppy, forgetful, implicit, and
subjective. But they have common sense and intuition, and may find creative
solutions in new situations. By contrast, agents are fast, rigorous, precise, explicit,
and objective. But they lack common sense and the ability to deal with novel
situations. Humans and agents may thus engage in mixed-initiative reasoning that
takes advantage of their complementary strengths and reasoning styles. As such,
intelligent agents enable us to do our tasks better, and help us in coping with the
increasing challenges of globalization and the rapid evolution toward the
knowledge economies.
References:
1. Cohen T 2019, "How to leverage artificial intelligence to meet your diversity goals”, Vol.
18 No. 2, pp. 62-65.
2. Millett K 2016, Sexual politics, Columbia University Press.
3. Oleksy W, Just E and Zapedowska‐Kling K 2012, Journal of Information,
Communication and Ethics in Society, vol. 10, no. 2, pp107- 120.
4. Taylor C 2013, Searching for similarity using corpus-assisted discourse
studies, Corpora 8, 1, pp.81--113.
5. Leavy S 2018, 1st International Workshop on Gender Equality in Software Engineering,
Gender Bias in Artificial Intelligence: The Need for Diversity and Gender Theory in
Machine Learning, pp.14-16.
6. Litoselliti L and Sunderland J 2002, Gender identity and discourse analysis, Vol. 2. John
Benjamins Publishing,
7. Romaine S et al 1998, Communicating gender, Psychology Press.
8. Winston PH 1980, Learning and reasoning by analogy. Commun ACM, pp.23:689–703.
9. Kowalski R 1979, Logic for Problem Solving, New York: Elsevier, North-Holland.
10. Cohen LJ 1977, The Probable and the Provable. Oxford: Clarendon Press.
11. Heiko MB 2013, Gentlemen before ladies? A corpus-based study of conjunct order in
personal binomials. Journal of English Linguistics 41, 3, pp. 212–242.
12. Michalski RS 1983, A theory and methodology of inductive learning in: Michalski RS,
Carbonell JG, Mitchell TM, eds. Machine Learning: An Artificial Intelligence Approach,
vol 1. Palo Alto, CA: Tioga Publishing Co, pp. 83–129
13. Lenat DB 1990, Guha RV. Building Large Knowledge-Based Systems: Representation
and Inference in the CYC Project, Reading, MA: Addison-Wesley.
14. Jurafsky D 2008, Martin JH. Speech and Language Processing. Prentice Hall.
15. Quillian MR 1968, Semantic memory. In: In Minsky M, ed. Semantic Information
Processing. Cambridge, MA: MIT Press, pp.227–270
