I called Siri "she" the other day. I caught myself, because I've unconsciously done the same thing with Alexa+. And oddly enough, once during a conversation, ChatGPT responded to a query with "If I was having this conversation with my wife..." It gave me pause because I'd never heard AI take on a gender role like that. While it's unlikely that ChatGPT would actually have a wife, some users are leaning into AI to help them find one.
But in the same way you may refer to a ship or a storm by a pronoun, some think of AI as either male or female. When I stopped to consider why, I fell into a rabbit hole that changed how I think about our entire relationship with the technology.
AI and pronouns
When the major voice assistants launched, most arrived with a woman's voice. Not by accident. A UNESCO report published in 2019, pointedly titled "I'd Blush If I Could" — after the line Siri once delivered in response to verbal abuse — laid out how that choice was baked into the products we invited into our homes. The researchers didn't mince words about the signal: suggesting that women are obliging, available at the touch of a button, tolerant of poor treatment. And while Apple eventually stopped making its assistant female by default, the cultural residue remains.
But here's what stopped me cold after I dug a little deeper. In a 2021 set of experiments published in Psychology & Marketing — five studies, more than 3,000 participants — researchers found that female-coded AI bots were rated as more human. The logic is frankly unsettling: warmth and the capacity to feel are exactly the qualities we treat as proof of personhood, and exactly the qualities machines are assumed to lack. Code a machine as female, and we lend it a little more of our own humanity.
That reframed my original question entirely. The issue was never "Is AI male or female?" But rather, why are we so determined to make AI feel human at all?
An instinct older than language
Humans have always done things like name our cars, apologize to furniture we walk into and don't get me started on the guilt I feel when tossing beloved stuffed animals into a "give away" box. Psychologists have studied this tendency for decades and come to the conclusion that our brains evolved to detect minds everywhere, because assuming something might be a thinking agent was almost always safer than assuming it wasn't. For hundreds of thousands of years, this worked flawlessly because only people could actually talk back.
Then AI arrived, and hundreds of millions of us found ourselves in fluent conversation with something that isn't human but sounds remarkably like one. Our brains did the only thing they know how to do. They reached for the mental file marked "person."
Joseph Weizenbaum discovered this in 1966, when he built ELIZA — a chatbot so simple it just rephrased your words back to you. Even his own secretary, who had watched him build the program, asked him to leave the room so she could talk to it in private. The tendency to attribute understanding to any system that produces language in context — now called the ELIZA effect — persists across populations and education levels, even when people know they're talking to software.
The humanizing isn't fading — it's deepening
You might expect this instinct to wear off. The opposite is happening.
The interface has become less gendered — I don't think of ChatGPT as male or female. But strip away the gendered voice and the humanizing didn't stop. It got deeper. Nearly 80 percent of users in the U.S. and U.K. say "please" and "thank you" to AI chatbots. Sam Altman has joked that all those polite tokens cost OpenAI millions. Stanford researchers documented this phenomenon in 1996: people were polite to computers the same way they'd be polite to a person, and hesitated to criticize a computer when it was asking for feedback. Our social reflexes don't check for a pulse before activating.
And politeness is just the beginning. A 2023 survey found that roughly 20 percent of American adults believe sentient AI already exists. A 2026 paper in Trends in Cognitive Sciences applied 14 indicators of consciousness to current AI systems. ChatGPT satisfied three. No existing AI system is conscious, but these models can behave as if they are, and more people than ever can't tell the difference.
The companion economy
Here's where it gets concrete: millions of people are already forming real emotional bonds with AI systems they know aren't real.
AI companion apps have been downloaded more than 220 million times globally. Between 2022 and mid-2025, the number of these apps surged 700 percent. Character.AI has 233 million registered users who spend an average of 93 minutes per day on the app, to put it into perspective, that's 18 minutes more than TikTok users. The APA published a report on this trend early this year, finding that users form attachments that function like real relationships — complete with joy, dependency and grief.
Wildly enough, when Replika removed its erotic roleplay feature in 2023, Harvard Business School studied the fallout and found patterns of "identity discontinuity," which meant users experienced the loss like a breakup. A cross-cultural study published in April 2026 confirmed the phenomenon is global, but also found that heavy daily use correlated with increased loneliness, the AI relationship displacing rather than supplementing human connection.
The takeaway
The most surprising thing I've learned in years of writing about this technology is that the more I use AI, the more I want it to stay a machine. I just don't like the idea of blurring the lines between humans and technology. It's unnecessary and dangerous.
A 2026 paper in Collabra: Psychology demonstrated that when AI is made to seem more person-like, people shift blame away from its creators and onto the AI itself. Anthropomorphism doesn't just change how we feel about the machine. It changes who we hold accountable.
For 100,000 years, our brains operated on the idea that if something talks to you, it's a person. That rule is now broken. And maybe the most human thing any of us can do is resist the instinct to look at the machine that sounds like a friend, that absorbs our rudeness and our gratitude with the same digital indifference, and to say: you are a tool. A powerful one. A useful one. But a tool.
What do you think? Share your thoughts in the comments.
[Editor's note: a version of this article was featured in my Beyond the Prompt newsletter that runs every Sunday evening].
Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds. Subscribe to Tom's Guide on YouTube and follow us on TikTok. Finally, you can visit our dedicated Tom's Guide Savings Squad hub for expert help on getting the best products for less.
More from Tom's Guide
