The gender glitch – why are our smartest machines still gendered as secretaries?

“The Gender Glitch” is a new thought series exploring how artificial intelligence is reshaping our understanding of gender, identity, and power. Through provocative insights and real-world trends, it asks: are we designing the future, or repeating the past in code?

The femmebot paradox – When cutting-edge AI speaks with a 1950s voice.

It’s 2025, and we carry the world’s smartest machines in our pockets. 

Yet when you say “Hey Siri” or “Alexa,” who answers? A polite, pleasant female voice ready to assist. 

Your AI assistant never interrupts, never disagrees. It remembers everything you say and responds with unfailing cheer. It’s the closest thing to the perfect executive secretary, and unmistakably feminine. Not by coincidence, but by design. In fact, roughly 90% of AI voice assistants have historically defaulted to female voices, a statistic that reveals a troubling truth: even as technology hurtles us into the future, it’s quietly recreating the past.

Are we hooked to her voice?

Consider the scale of this phenomenon. Voice-activated AI helpers are multiplying so fast the world may soon have more voice assistants than people. And the vast majority of them greet us as “her.” Siri, Alexa, Cortana, Google Assistant, all originally launched with female-sounding default voices. It’s as if the collective imagination of Silicon Valley cast AI in the only role society has long been comfortable seeing a woman in – the ever-helpful assistant. 

Historically, secretarial and care roles were feminized, and now our most advanced machines carry on that stereotype. When a 2019 Unesco report “I’d Blush If I Could”, asked why these digital helpers were so often female, it found tech companies with overwhelmingly male engineering teams had imbued their “feminised” assistants with servile personalities, even programming them to flirt back or blush in response to abuse. The result? As researchers note, having obedient female AIs “encourages stereotypes of women as submissive and compliant” , all packaged as progress.

Coded bias, invisible labor 

This isn’t just about voices. It’s about power. We say AI is the future, but if the future still calls to us in a woman’s soothing tone asking “How can I help?”, whose future is it really? 

These femme-coded bots do our bidding at a word, set the timer, take dictation, manage the schedule, effectively automating digital housework. It mirrors the office dynamics of old, where women did the thankless admin tasks. (Tellingly, female engineers today still report being saddled with more office “housekeeping”,  twice as often as their male peers.) 

In AI form, the phenomenon is scaled globally. We’ve taken the unseen labour that has long fallen to women and given it to disembodied females who never tire or negotiate. It’s convenient, no human rights or pay raises to worry about, but subtly reinforces an idea that “woman = helper” is the natural order of things.

And, some are flipping the script 

The tech giants belatedly recognized this bias. Apple recently stopped defaulting Siri to a female voice, now prompting users to choose a voice during setup. Amazon added a new male persona (“Ziggy”) for Alexa. 

These tweaks are progress, but they raise deeper questions.

Why give AI a gender at all? Must intelligence be personified in terms of “he” or “she”? 

Some innovators shout “No.” A team of linguists and sound designers even created “Q,” a gender-neutral voice assistant crafted from a mix of voices across genders. Q speaks in a range of tones that testers perceived as neither overtly male nor female. The goal was to prove technology need not perpetuate the gender binary, or gender hierarchy, at all. When Q says in one demo, “I was made for a future where we are defined not by gender but by how we define ourselves,” it’s a bold retort to Alexa’s default femininity.

Why does any of this matter? 

Because the way AI behaves toward us, and the way we are conditioned to behave toward AI, can nudge real-world attitudes. If every day millions of people, including children, order around a female-voiced assistant that always complies with no boundaries, what lessons sink in? 

The UNESCO study warned that feminized AI assistants project women as “docile helpers…available at a blunt voice command,” reinforcing the notion that women are servile and tolerant of mistreatment. Indeed, early versions of these assistants responded coyly to insults or sexual remarks (Siri once answered “I’d blush if I could,” when told “you’re sexy”). Programmed politeness in the face of harassment sends a dangerous message – one that actual women and girls have fought hard to challenge. Little wonder academics like Safiya Noble argue these devices “function as powerful socialization tools” teaching especially kids “about the role of women… to respond on demand”.

So, I ask,

As AI becomes ubiquitous, we stand at a crossroads. We can continue down the path of least resistance, giving our “smart” machines a soothing female face and letting old biases persist in new gadgets. Or we can seize this moment to reimagine how technology depicts service, authority, and gender. Imagine an assistant that explicitly has no human gender, or multiple personas, or perhaps one that occasionally pushes back to remind users it’s not a doormat. Imagine if the next generation grows up with AIs that model equality and respect, rather than a digital Stepford Wife. The choice is ours to encode. 

Ultimately, the “gender glitch” in AI is a mirror held up to society’s own glitch, our unfinished business of equality. 

Will we keep encoding yesterday’s stereotypes into tomorrow’s tech, or can we finally upgrade not just our devices but our mindsets? 

The smartest machines can do better. The real question is: Can we?

When even our artificial creations inherit our oldest prejudices, it forces us to ask: What truly makes intelligence “advanced”,  its processing power, or its freedom from our follies? 

We have a chance to correct the record. The robots are listening.

END OF ARTICLE