Why the world’s busiest smartconcierge, Ivy, is a woman

“Can we call her Sandy instead?” The first time we heard that question from someone considering Ivy for their hotel, we froze. And a second later, cheerfully refused to change Ivy’s name.

Naturally, we are attached to Ivy as the name for our smartconcierge. More importantly though, we are proud of the results driven by the consistent persona and the thinking that led to the decision to call Ivy, Ivy.  

Gender and AI: Smartconcierge Ivy is a woman

The first (and more) obvious reason we will never change Ivy’s name to Tim, Sally, Mark or anything else is because we want to offer guests a consistent persona who is available to take care of their every need. Regardless of the hotel they stay in, guests can text Ivy and expect immediate help. This promise of a consistent experience isn’t merely for Go Moment’s benefit. It is in our clients’ interests too. A recent loyalty study shows that while effective brand representatives drive 3.9X higher member satisfaction (!), only 19% of end customers say that brand representatives make them feel special and recognized. That’s an instant win for 81% of guests that Ivy delivers right out of the box.

Let’s say that as a guest, I called the front desk to order a bottle of Nebbiolo. Tim answers the phone and then transfers me over to Mark from room service, who then passes me to Sally at the bar to check if that particular wine is available… that is a fail. I am unlikely to remember anyone’s name later, and already feel like a hot potato being handed off among a dozen people for (what seems to me) a simple request for some wine. Sound exaggerated? Here’s how a long-time hotel front desk worker reported feeling about the question “Don’t you remember me?”:

Let me think about this…average of 500 guest interactions a day…it’s been two years since you stayed with us. So that’s a clean quarter of a million separate interactions since your last stay. Wait…Wait! No. No, I don’t remember you.

Let’s jump to the more important reason why Ivy is called that. Ivy is a woman – that’s how we have thought of her since early design stages. As an aside, I consciously switched from female to woman after several conversations with friends. This is probably an old debate now since almost no one I know uses the word female anymore. Right? Right. If you are still using female, read this Buzzfeed article.

Why did we shape Ivy, an AI assistant, to be a woman? When designing our smartconcierge, we wanted to create a personality that would be attentive, warm and caring. Seems like an easy jump from here to creating an assistant that is a woman? Not quite.

Even back then, our group was aware of stereotypes and the raging debates of why most AI assistants are cast as women. The default setting for Alexa, Google Home, Siri, Cortana is a woman’s voice. Ask your Google Assistant something – a woman’s voice is likely to answer back. Chandra Steele, over at PC Mag, had another example to share:

Consider that IBM’s Watson, an AI of a higher order, speaks with a male voice as it works alongside physicians on cancer treatment and handily wins Jeopardy. When choosing Watson’s voice for Jeopardy, IBM went with one that was self-assured and had it use short definitive phrases. Both are typical of male speech—and people prefer to hear a masculine-sounding voice from a leader, according to research—so Watson got a male voice.

The Real Reason Voice Assistants Are Female (and Why it Matters)

Gender bias in AI is real. And we need to address it. Julie Teigland, Managing Director at EY, talks of why we need to look at this urgently: Why we need to solve the issue of gender bias before AI makes it worse. I can’t agree more.

And yet, Ivy is a woman. While creating Ivy, we went back to evolutionary biology and learned that in general, we find higher pitched voices more comforting and less likely to trigger fight or flight-type reactions. For example, if you said something inappropriate to an AI named “Mark” as a test, and he told you to “cut it out,” it may come off more aggressive than intended in a lower timbre according to evolutionary biology.  

Then there were practical considerations, like text message message character length limits and CUI (Conversational User Interface) design best practices (though the term CUI didn’t exist in any meaningful way when we created Ivy!).

We fell in love with the name Ivy. It’s short, playful, powerful and global – Ivy is a common name across the US, Europe, Australia and large parts of Asia. It was exciting to be able to communicate all the complexity of a system powered by billions of dollars of AI R&D in three letters.

Were we right in designing Ivy as a woman?

I am watching the launch of Q, the first genderless voice (as the creators call it) with interest.

In the hospitality industry, are we ready to embrace this concept? Have any thoughts to share?

Get a grip on complexity with beautiful data visualization

I dig dataviz.

Data visualization isn’t esoteric. Neither is it a world inhabited by math geniuses and data scientists alone. On the contrary, most projects are visually stunning and lots of fun to play with!

What I love most is how the artists (yes, they’re artists; not just creators!) zoom in to the essential within large datasets, unearth insights, and communicate these creatively while destroying the surrounding noise. Done right, this process makes the abstract approachable (sometimes even easy), and the end result is downright magical.

Daniel Kunin’s Seeing Theory project is a brilliant example of thoughtful dataviz. Kunin set out to make “statistics more accessible through interactive visualizations,” and his work absolutely crushes that goal. Seeing Theory has one of the most approachable visualizations of probability and statistics, paired with helpful content. Check out the site: https://seeing-theory.brown.edu

The website was designed using Mike Bostock’s JavaScript library D3.js. If you haven’t heard of Bostock before – and are keen on understanding how an algorithm might randomly traverse a maze – take a look at his Visualizing Algorithms “post,”. Its interactive, informative and simple – or quite simply the gold standard in my book.

Why am I talking of dataviz here? It is clearly one of my passions. More importantly though, I believe that we are not too far from a time when we will ignore datasets if they aren’t visualized beautifully. Or, are we there already?!