Making artificial intelligence accessible to humans

Inventing the future at Google

Ask the Job Doc.
Ask the Job Doc. –

Imagine if you could map out your e-mail exchanges. The network of communication would be tangled and complex as messages flow in and out of your inbox — difficult to capture in that way. But creating a visual e-mail portrait was one of the early projects undertaken by Fernanda Viegas, a computational designer and machine learning researcher who now works at Google Inc. in Cambridge. Brazilian-born Viegas’ art and design background serves her well in showing the “Big Picture” — using information visualization to make complex data accessible and useful. As leader of Google’s Big Picture research team, which is part of Google Brain, Viegas’ says here mission involves “democratizing and humanizing AI (artificial intelligence).” When she started working on information visualization more than a decade ago, it was “an exotic academic niche.” But now, she and her colleagues at Google apply AI to everything from medical records to the world’s largest virtual snowball fight. “Data is ubiquitous in our lives, and it’s critical for everyone, from the engineer to lay user, to be able to understand and reason about complex information,” says Viegas, who also is co-leader of Google’s PAIR (People + AI Research) initiative. She spoke to the Globe about how humans and artificial intelligence can work together.


“I’m passionate about bringing sophisticated technologies out of the lab and into people’s lives. I’ve done this with data visualization and now I’m hoping to do this with AI. Technology should be available to everyone and not just used by the elite. That’s why we call it ‘human-centered machine learning.’ My team has launched deeplearn.js, a ML (machine learning) library that runs on browsers – no need to download or install any software. It speaks the language of the web – Javascript— opening up the technology to millions of people who wouldn’t have used it before: from kids to artists, farmers, journalists, etc. It’s incredibly exciting to see a powerful technology like machine learning put to interesting uses by non-experts outside of university and industry labs.

“We’re working on AI systems that doctors can trust, looking at how AI may empower creative professionals, and exploring how recommendation systems such as YouTube can offer a more relevant experience for users. I can envision applications like tutors that can help teachers personalize lessons for different students, given their different learning styles.

“The link between my graphic design background and my high-tech work was at the Media Lab at MIT, where I did my graduate work. That’s where where I became steeped in the computer science community. I became a research scientist in visual communication and was fascinated by the power of the web to foster a social style of data analysis. If people can upload their own data, and create interactive visualizations, then they have the tools to contextualize complex abstract patterns.


“A simple explanation, really of what I do is that I transform data – usually numbers – into pictures. An older example is my wind map, a real-time, living portrait of wind currents over the US. While it was an artistic exploration for me a while ago, bird watchers, bicyclists, and others have since used it for practical applications. I love it that I get to invent the future. We’re building things that don’t exist yet, we are pushing the boundaries of how technology can better serve people.

“I try to express richness of detail, a sense of wonder and visual beauty in my work today, and I think it was inspired by growing up in Rio, surrounded by a gorgeous natural scenery. My bedroom window opened up to the rainforest. I’d spent countless hours watching monkeys move through the trees, listening to the noises, breathing in the fresh air. The richness and texture of the forest and sea around me were profoundly inspirational. For me, composing with data points can also ultimately be a portrait in expression and loveliness. A picture really can be worth a thousand bits.”