The Importance of Women in AI: Combating Bias Effectively

July 5, 2019
4 min read
Microsoft Italia and iGenius bring the Generative AI into the world of numbers

The difference between virtual assistants and virtual advisors lies in the way they respond to users.

Assistants execute requests (put music on, tell me the time, make a call), whereas advisors answer queries, make assertive recommendations and predictions based on data, and may even “disagree” with users, suggesting that they take a different course of action.

Quite a cool job for an AI, right? Except that most advisor jobs are taken by male AIs: ROSS the Robot lawyer, for example, serving up legal advice at a top American law firm.

Most assistants, by contrast, are female in character, have a distinctly female voice and use language that suggests a lower status than their users.

This gender-biased trend in AI reflects “the lack of women and minorities in tech fields, and artificial intelligence in particular”, according to a May 2019 UNESCO report.

And the gap might get wider: according to UK analyst firm Juniper Research, there will be 8 billion digital voice assistants in use by 2023 — that means there will be about the same amount of people as virtual assistants on the planet.

In your own words

As technology shifts from typing to voice, the conversational interface of our devices takes center stage. Our virtual advisor, crystal, is a woman for all the right reasons.

Breaking the advisor vs assistant gender trend, she’s a smart, knowledgeable colleague that works with users on the same level.

crystal doesn’t assist, but advises on the basis of a relationship built on trust.

“The key to building users’ trust in our AI advisor’s abilities is the accuracy of the data-driven insights she provides them”, says Arianna Stefanoni, Head of Conversational UX at iGenius.

Crystal is an augmented analytics tool that uses machine learning and natural language processing to access complex and siloed data, and process it rapidly. She provides real-time advice and proactive notifications in natural language to help users make effective decisions and guide data-driven business strategies.

“As users progressively receive valuable insights, as well as proactive notifications and predictions from crystal, they start trusting her”, explains Arianna.

“This boils down to the technology behind her, but crystal’s conversational interface must reflect her competence, and her language suggest the same clout as her users.”

Arianna heads the conversational design team at iGenius, and is responsible for designing crystal’s verbal posture, the language with which she interacts with her users.

As a woman in a largely male-dominated industry, she has a big job on her hands. “The foundation of my work with crystal is a set of style guidelines that serves as a compass for me and my team, and it starts with a negation — crystal is not servile.”

“We imagine her as a knowledgeable not an obsequious assistant. She wouldn’t ask how she can help, but work alongside users to guide them to the best possible decisions based on the data sets she connects to.”

“She is assertive, and she has a female voice. It shouldn’t be special but it kind of is.”

No need to apologize

For Arianna, humanizing crystal is key, but that is a big responsibility: the risk of falling prey to bias and stereotypes is high.

“I think humanizing AI makes a huge difference when it comes to user experience, however the fact that crystal is a woman doesn’t mean her speech needs to reflect emotions typically associated with the female gender stereotype”, she explains.

“This is crucial in the context of delivering error messages: my job is to craft lucid, articulate responses and avoid the pitfall of having crystal apologize unnecessarily.”

“When crystal delivers an error message, she shows proactiveness rather than apologizing: she would say ‘I’m working to reconnect to your data as soon as possible’, for example.”

Sometimes, the data available to crystal is limited, leading to her being unable to answer a specific user’s query. “Plainly speaking, this is none of crystal’s fault”, explains Arianna.

“Rather than wasting time having her apologize about it — which would add zero value to the user experience — we make sure crystal helps users continue their experience without unnecessary roadblocks.”

If we’re bound to talk more with our devices, it’s high time women have their real voice heard in AI.

Similar stories