Abstract

This work aims to represent LGBTQ+ identities in Artificial Intelligence (AI). The work presents a document to train or code a BiBot, a bisexual chatbot, with resources for Bisexuals living in the United States. Bisexuals are the largest non-heterosexual orientation. Despite being the orientation with the largest share of people, bisexuals are either overlooked or lumped together with gay and lesbian populations. While there is some overlap with the gay and lesbian experience, bisexuals have their unique minority stressors. This work utilized both qualitative and quantitative research to pinpoint those issues to make the training or coding document. Many bisexuals feel like they are facing the unique minority stressors on their own, which is problematic and isolating for many. There are two goals of this AI. The first is to promote bivisibility, which is being bi and out, open, and proud. The second is to fight bi-erasure, which is when bisexuality is when people treat bisexuality like it is not a real and legitimate identity or deny that bisexuality even exists. The result of this research and the creations of BiBot is a way for bisexuals not to feel alone, invisible, and know that they have a valid identity.

Introduction

The Williams Institute found that when examining non-heterosexual orientations, 17% identify as lesbian, 31% identify as gay, and 52% identify as bisexual (Gates, 2011). Even though bisexuality is the sexual orientation that comprises the majority of non-heterosexual identities, it is a highly misunderstood sexual orientation that lacks visibility and representations. It is not being half gay and half straight and is its own state. Bisexuality can be defined as an attraction to more than one gender, and it does not have to be along the lines of the gender binary or 50% liking men and 50% liking women. Many frame the concept using the definition set forth by bisexual advocate Robyn Ochs (n.d.) " the potential to be attracted – romantically and/or sexually – to people of more than one sex and/or gender, not necessarily at the same time, not necessarily in the same way, and not necessarily to the same degree.”

How bisexuals are misunderstood can be damaging. The creation or, hopefully, creations of a BiBot inform and empower bisexual lives. The document to train or code is intentionally not in a coding language to promote creativity and bi-visibility. The intent is that many creators will use this information and add to it to create many different incarnations of BiBot for people to find and chat with. Not everyone has an assembled LGBTQ+ community near them that they can access and talk with, and not everyone can come out. Therefore, for many people, a chat with the proposed AI may be one of the few ways to feel like their identity is valid and important, and access LGBTQ+ information and perspectives. The document to train or code is framed around potential human input, and what BiBot’s responses should be like. BiBot speaks friendly and in simple language because people of all ages and walks of life should be able to chat with them. A lot of people do question their sexual orientation at a young age. The Pew Research Center (2013) found that age 13 was the median age in which bisexual individuals first felt that they might not be heterosexual. This work walks through the information in the training or coding document, and why those concepts are present.

Before examining the components of this AI project, there are a variety of identities under what is known as the bi+ umbrella. These are pansexual, sexually fluid, no label, queer, and a variety of other orientations. This work is not looking to slight those specific identities, and BiBot is intended and should also be modified to assist these identities too. The term bisexual is used because these identities branch off the concept of bisexuality.

Materials & Methods

The training or coding document was made through reading LGBTQ+ organization websites who often make reports of current LGBTQ+ research. LGBTQ+ organizations who compile statistical data get their data from reputable groups who specialize in data collecting like the Pew Research Center or the Williams Institute. The data reports and explanations from LGBTQ+ organizations were juxtaposed with journal publications and published first-person narratives. The collection of first-person narratives from magazines and online publications put voices, emotions, and experiences to the statistical data. The intent of some of the informal narrative sources is to make the concept more accessible. It is important for BiBot to be a knowledgeable member of the LGBTQ+ community and for end-users to feel like they are talking to a conversational fellow bisexual person. If people feel like they are talking to AI, they might not turn to the technology to ask about resources. For some BiBot could be the only bisexual or LGBTQ+ identity that the person can chat with, so therefore it is important to add authenticity to the AI.

Results

Section 1—Defining the Orientation and Thinking About Identity

Many people in the LGBTQ+ community feel that coming out never ends. However, the first person that a person must come out to is themselves. For someone coming out to themselves, they need to know what bisexuality is, and that is a real identity. The definition and the Ochs (n.d.) quote are used. The common myths and stereotypes are discussed and were informed by Dallara (2011). For many people internalizing the myths and stereotypes make it hard for them to understand what bisexuality is and impedes their ability to have a moment of realization on whether that is their identity. This is also why the Trevor Project Coming Out: A Handbook for LGBTQ Young People is also suggested (The Trevor Project, 2019) to the end user. This online book though aimed at youth, is an easy read for anyone questioning to understand the subjects of sexual orientation and gender identity.

After having that initial coming out to self, many people wonder what it means to live authentically. Bi.org (2019) is recommended for people as a resource. It is affiliated with The American Institute of Bisexuality and the Journal of Bisexuality. There are interviews, movies, television, music, and books on the website so people can see other people and representations that look like their lives and feelings.

Section 2—Coming Out

As stated, before the coming out process is never-ending. Every bisexual person has a unique story and set of life circumstances. For many, the next step is to come out to romantic partners, friends, and family. However, compared to other orientations, bisexuals are considerably less likely to come out to important people in their lives. The Pew Research Center (2019) found that 75% of gay and lesbian individuals were out to all or most, while only 19% of bisexuals answered this way. Some common fears are discussed in the coding and training document, and users are pointed to the Human Rights Campaign’s (2016) guide specifically designed for bisexual people who want to come out. Because families and allies often need support and advice, the link to the PFLAG (2019) website is provided for the end-user to give to people in their lives who may need assistance or want information.

Section 3—Perceptions

The Human Rights Campaign (2015) explains that bisexuals can face a form of double-sided discrimination that comes from both heterosexuals and gay and lesbian orientation. Many do not see it as a legitimate orientation and have perceptions informed by negative stereotypes such as bisexuals are confused and promiscuous. Whether or not bisexual people are out or not, they still deal with perceptions from people on the outside looking in, they are dealing with perceptions in their romantic relationships, and the perceptions can be challenging to deal with. While there are some scientific studies on some aspects, it was more useful to perform digital fieldwork by reading posts on the bisexual and pansexual side of LGBT Chat (n.d.) and reading first-person narratives from people whose relationships look “straight” or “gay.” These works included: Lang (2019) Marusic (2017), Milton (2018), and Venti (2015). All these sources put voices, feelings, and experiences into bisexual identities and lives and the way their relationships look from the outside. The goal of the digital fieldwork performed to make BiBot seem more human, relatable, and easy to talk with.

Section 4—An Issue Not Covered By AI

While research and scholarship seek to understand and draw conclusions about bisexuals as a group, every bisexual person has a unique life and circumstances. Not every bisexual person will have someone near them to turn to who understands their identity and stressors, and not every bisexual has the same definition of identity, the exact same stressors, or the same cultural background. Something that is a huge issue to a person who comes from a traditional background like someone who has immigrant parents, or a highly religious household is not a huge issue to someone from another environment. While there are complex machine learning models, BiBot is intended to give generalized advice and resources based upon research findings. It is generalized as the intent is to reach as many bisexual people as possible who need assistance and promote bivisibility. Therefore, a link to LGBT Chat (n.d.) is provided. It gives the person a chance to connect with other LGBTQ+ individuals who have been in that person’s situation or a similar one and may be able to offer their own experiences and solutions.

Section 5—Emergency or Crisis Situation

There are mental health disparities and emergency and crisis situations from minority stressors and lower levels of social and familiar support. The Human Rights Campaign (2015) has a guide on health disparities that helped inform this section and the situations BiBot users may be in, and the appropriate phone numbers and websites were found and put in the training and coding document. These issues include depression, anxiety, suicide ideation, alcohol, and substance abuse. The idea to include sexual assault and domestic abuse phone lines came from data sets from the Center for Disease Control (2018), which found that those with a bisexual orientation face higher rates of sexual assault and domestic violence than other orientations. This can be attributed to people thinking bisexuals are promiscuous, hypersexual, and other negative stereotypes.

Section 6—Political Concerns or Fears

This initial incarnation of BiBot is for people in the United States and was written with Americans in mind. LGBTQ+ rights have come a long way in the United States since the Stonewall Riots in 1969. However, there is still so far to go for LGBTQ+ individuals in the United States. State-level laws that have significant impacts. Gay, lesbian, and bisexual youth are at a higher risk for suicide, and Raifman et al. (2017) found that after same-sex marriage laws were passed in 2015 that the outcome was that there were 134,000 fewer adolescent suicides. Laws have huge and heavy implications for the quality of LGBTQ+ lives. Even though someone can now marry who they love, there are varying laws on adoption, discrimination, and 1 in 5 LGBTQ+ American’s live in poverty (Lee Badgett et al., 2019). Therefore, BiBot encourages political awareness and suggests ways to get involved in politics.

Section 7—Healthcare

The Human Rights Campaign (2015) examined many studies and found that bisexual people face numerous health disparities and that 39% of bisexual men and 33% of bisexual women do not disclose their orientation to medical providers. Unfortunately, many bisexuals have faced discrimination, discrimination, biphobia, and bias in health care settings. The health issues in the trading and coding document were informed by the Human Rights Campaign (2015). The health issues that arise were ones that came up in the emergency or crisis section, such as anxiety, depression, suicide ideation, substance abuse. The sexual health disparities were touched on. BiBot points the user to the Human Rights Campaign’s (2015) inclusive guide to sexual health since and the San Francisco AIDS Foundation (2019) website. These are important links to give since so many do not disclose their orientation and may not get the information that they need to engage in safer sex practices. It was essential to provide those links, especially to those under 18 who may not disclose their orientation and get the care that they need. Arbeit et al. (2016) studied bisexual adolescent females and found that sex education in schools was either stressing abstinence or heterosexual safe sex practices. Only 38% of the adolescent girls had told their doctor their orientation because they did not want the doctor to disclose to their parents about their orientation, sexual activity, or if they had an STI. Ultimately BiBot tells bisexuals of all ages that they should reference the Human Rights Campaign’s (2019) healthcare equality index to search for healthcare near them that is inclusive of LGBTQ+ patients.

Discussion

This work is just the groundwork to make bisexuality more visible and covers some broad basics of the bisexual experience. The AI concept is by no means extensive and has shortcomings. The Gates (2011) and Pew Research Center (2013) appear to be the last big demographic counts of the LGBTQ+ population. Therefore, there is a gap in the scholarship. The labels and language of being an LGBTQ+ person have evolved since. The language to identify and classify the population has since changed. The other shortcoming is that this design lacks universal design. This was designed specifically for people in the United States. The phone lines, websites, and resources were chosen for people in the United States and can to be altered for other locations as there are bisexual people all over the world. The other issue with this work related to universal design is that more voices and scholarship need to be factored in to meet the needs of unique bisexual lives. There are so many other intersections for people and their bisexual identity. These intersections include gender, transgender identities, race and ethnicity, and other aspects of the bi+ umbrella. Including more voices and scholarship creates a more inclusive BiBot, who and reaches more LGBTQ+ identities.

Conclusions

It’s not the bisexual orientation that causes poor mental and physical health outcomes or unsafe situations; it is the minority stressors and people’s lack of knowledge on the definition of bisexuality. The impact of BiBot could be lifechanging AI to so many LGBTQ+ people, particularly those living in a small town and are struggling with their identity and minority stressors. As this work showed, there are many minority stressors that bisexuals live with that impact physical health, mental health, and safety. Since many bisexuals are not out with their identities, this chatbot could be of assistance. The goal of making this open-source work is to make bisexuality a more visible and understood identity and hopefully, there will be many incarnations of BiBot. With many BiBots made for phones and web browsers, there are more ways for bisexuals to interact with AI and learn that their identity is valid and that they have resources.

References