Don’t let ChatGPT chatter overshadow earlier tech harms and biases

How would you feel if you learned that this article was written using artificial intelligence (AI), namely, ChatGPT-4 (Generative Pre-trained Transformer 4)?

 Now, how would you feel if you learned that you will be seen by a computer and not a doctor during your next medical visit? 

These are some of the questions that we are asking after the introduction of ChatGPT-4, a chatbot developed by OpenAI which is composed of an AI system and a chat interface.

Its potential uses have generated a great deal of excitement as well as trepidation. In a recent letter signed by Elon Musk and other technology leaders, there is a call for a six-month pause by all AI labs and, if that does not occur, governments should intervene and institute a moratorium. This letter issues the alarm that “AI systems with human-competitive intelligence can pose profound risks to society and humanity.” There is even the suggestion that this technology could destroy democracy and civilization through propaganda and lies.

Society has faced similar challenges when technology evolved faster than our ability to regulate and control it. During the Cold War, the propagation of nuclear weapons was feared to the point that we held drills in elementary schools to protect ourselves from a nuclear blast. I’m amazed that we believed that a “duck and cover” approach by hiding under our desks would protect us from a hydrogen bomb or nuclear explosion like what occurred in Chernobyl. 

In the 1990s, we wrestled with moral and ethical issues when Dolly, a sheep, was cloned. Questions swirled around human cloning and genetic engineering. Through public/private efforts, including presidential commissions, oversight was enacted which requires constant surveillance and effective leadership.

With every advancement, there are significant benefits as well as risks. We will not go back to a time when ChatGPT-4 and related technologies are not part of our lives. However, this can provide an opportunity for us as we discuss the merits and the required regulations, to now also assess the current technologies that we use extensively every day — technologies that have revolutionized how we live, work and play. Yet, these technologies are inherently flawed by gender biases.

In August 2020, the United Nations Educational, Scientific and Cultural Organization (UNESCO) released a report that I contributed to, Artificial Intelligence and Gender Equality. Many groundbreaking messages in this report were overshadowed by the chaos of the early days of the COVID-19 pandemic. Hopefully, with the heightened interest in the potential perils of using ChatGPT-4, there will be a renewed focus on eliminating or at least minimizing the inherent biases in AI which have adversely impacted half of the world’s population — women and girls. 

As the UNESCO report explains, “these biases are rooted in stark gender imbalances in digital skills education and are exacerbated by the gender imbalances of the technical teams developing frontier technologies by companies with significant gender disparities in their C-suits and corporate boards.”

Below are just 10 technological advancements that have shaped our modern existence, but have also generated negative consequences. In many respects, they are double-edged swords creating a greater digital divide.

1. Telemedicine: Though it has increased outreach, especially in rural and underserved regions during the pandemic, the lack of human touch and visual cues significantly impact the quality of care delivered especially for women and girls.
2. Electronic Health Records (EHR): It has provided more seamless care, but more female clinicians have experienced burnout from a system that they are often asked to use more than their male counterparts. There are also concerns about gender bias in EHR software.  
3. Virtual Communication: Throughout the pandemic, the use of virtual meetings has provided greater flexibility for outreach; however, the limitation to two-dimensional visual cues and 24-hour connectivity has led to greater stress in women especially with family responsibilities.
4. Digital Assistants: They have provided expedited information exchange, but the use of female names such as Alexa and Siri has contributed to gender stereotypes of female subservient behavior, and the devices may not recognize and respond to female voices, limiting effectiveness. 
5. Virtual Reality: The devices have enhanced analog environment experiences, but the headsets have led to disproportionate side effects in female users such as nausea and headaches due to designs more suited to fit male users.
6. Keypads: The smaller keypads seem designed for smaller hands, but female users have greater tactile sensation in their hands compared to male users and the keypads may not register a lighter touch, resulting in female users developing repetitive stress disorders from having to push down harder. 
7. Science, Technology, Engineering and Math (STEM) Education: Though more girls are entering these fields after positive experiences in early education, many do not find positions or leadership opportunities.
8. FemTech: This industry, which is projected to be a $50 billion market by 2025, has advanced innovation, especially in female reproductive health, but female users are concerned about reporting their menstrual cycles for fear of retaliation after the recent legal challenges to curtail reproductive healthcare.
9. Social Media: Diverse platforms have disrupted outreach and connectivity, breaking down barriers and disseminating information, but it has also contributed to cyberbullying and stalking as well as misinformation, greatly impacting the well-being of women and girls and non-binary individuals.
10. Automation: This technology has improved performance and efficiency, but can contribute to a loss of economic empowerment and labor market opportunities for women such as in administrative and clerical roles.

UNESCO recommends that civil society develops gender-inclusive AI principles and guidelines to overcome the built-in gender biases found in AI devices, data sets and algorithms. This will not only provide a remedy for our existing technologies but also for the new ones that are being pioneered now and in the future. ChatGPT-4 has opened the door for this discussion where we all need to be welcomed into the room.

Just in case you are wondering, I wrote this article — without AI involvement.

Saralyn Mark, MD, is the founder of SolaMed Solutions, LLC, host of the “Always Searching” podcast and founder of iGIANT (Impact of Gender/Sex on Innovation and Novel Technologies). She is the American Medical Women’s Association’s COVID-19 lead and a former senior medical and policy advisor to the White House, the Department of Health and Human Services and NASA.