The role of gender, androgyny, and AI in humanoid robotics

It is this evolution of demand that is now bringing to light the complex interplay between societal norms, gender stereotypes, and robotic design, and is raising significant design questions; especially regarding gender representation and the integration of Artificial Intelligence (AI).

Gendered design reflecting societal biases

Historically, humanoid robots have mirrored societal stereotypes in their design. For example, female-appearing robots like Nadine, developed in Singapore, often embody nurturing characteristics, while male-appearing counterparts such as ASIMO or Atlas are designed with more mechanical or robust features, reflecting an outdated perception of gender roles.

However, this trend is not just limited to Western cultures, it is also prevalent in Japan and Middle Eastern societies where robots often adhere to the respective cultural norms and biases regarding gender roles​​.

According to Ying Liu, Head of AI R&D at TGO and Visiting Researcher at Queen Mary University of London, we need to examine the data to ensure an unbiased output: “AI developers generally do not control the algorithms to go a specific direction regarding gender stereotypical biases. If the algorithm forms a stereotype or bias, it is because they exist in the data we fed into the ML structure, and it is an important factor which affects the outcome. Of course, it is not ethical to have these as a factor in the algorithm. So, it is generally separated from the model structure afterwards.”

Can true androgyny in robotics be achieved?

In response to these gendered biases, there's a growing trend towards the creation of androgynous humanoid robots.

Androgynous robots are designed to neither distinctly represent male nor female characteristics, and so they are perceived to offer a more neutral perspective on gender roles. This is an approach that is gaining traction as it avoids reinforcing traditional gender stereotypes and appeals to a broader, more diverse audience.

However, the challenge is that many people will still assume a robot’s gender even when they have been designed to be neutral – indicators such as form, shape, and behaviour can all affect a person's presumption of what gender they believe a humanoid robot to be.

Incorporating AI in design

The integration of AI in humanoid robots further necessitates careful consideration at the design stage. AI systems, if not designed with gender neutrality in mind, could inadvertently reinforce gender stereotypes through their interactions and decision-making processes.

For instance, AI-driven robots need to be programmed with algorithms that are free from gender-biased data, ensuring that they interact and respond in ways that do not perpetuate existing societal biases.

According to Liu: “The current strategy is to decompose a ML structure, study what it is doing, figure out the components which forms gender stereotypes manually and take the component out of the ML system. There are a few difficulties that come with that strategy; a) It reduces the accuracy of the ML system; b) It relies on the ML engineer's knowledge to discover and find the components which forms the stereotype; c) Sometimes we just do not know it exists in the system. It relies on feedback from users to discover it and report back. Incorporating it into regulations or standards will motivate companies to allocate the necessary resources.”

Design strategies and their impact

Various design strategies are being explored to address the gender representation in robots coupled with AI considerations:

Mimicking human gender characteristics: robots like NASA's Valkyrie, designed with female superhero elements, aim to challenge traditional gender norms, but even this design risks perpetuating stereotypes. Integrating AI in such designs should focus on ensuring that their interactions are gender-neutral and inclusive.

Customisable gender features: options like x.ai's scheduling bot, which allows users to choose gender cues, promote diversity, and challenge gender binaries. AI in these contexts should adapt to user preferences while maintaining neutrality in its responses and functionalities.

Genderless robots: designing robots without gender cues, such as Matlda and NAO, focuses on functionality rather than gender, avoiding the reinforcement of stereotypes. AI in these robots should be programmed to support this neutrality, ensuring that their interactions are universally relatable and unbiased.

Challenges and opportunities

The way forward in addressing the design challenges of balancing societal norms and AI’s potential for reinforcing gender stereotypes is not an easy task, and it is one that calls for a multitude of diverse perspectives at every stage of conception and design.

Further, the growing consumer demand for gender neutrality mirrors broader societal shifts towards inclusivity; yet overcoming deep-rooted biases in robotics and AI algorithms remain a significant hurdle. Therefore, research and development in this area are crucial to enable understanding of how robot gender perceptions impact human attitudes towards gender equality.

A crossroads in robotics design

The choices made by designers and consumers today will not only shape the future of the design of robotics and AI, they will also have a significant bearing on societal perceptions of gender. This evolution in robotic design reflects a societal movement towards challenging traditional norms and embracing diversity and inclusivity.