Design for Augmented Intelligence Bringing a human-centered approach to big data and smart machines
Can we give data a soul? We think we can. We’ll go even further than that: we must. Why? As the driver of artificial intelligence, data is informing a new age of design and production that will rival the industrial revolution.
That should give us a moment of pause, too. If we’ve learned anything from the sci-fi specter of machines gone wrong, it’s that with great computing power comes great responsibility. Data-driven intelligence will soon be embedded in nearly everything we see, touch, and do. Knowing that, it’s our duty to bring our humanity to it — to design it to be relevant, empathetic, and yes, even soulful.
Instead, what we mostly see out there is technologists trying to get a little smarter about design, and designers trying to get a little smarter about technology. That won’t suffice if we are to build a humane intelligent world. In order to get where we need to go, we have to work hand-in-hand — to bring deep practitioners together on interdisciplinary teams. That’s why IDEO is pleased to announce that we have acquired Datascope. Our Chicago studio has been exploring human-centered data design alongside Datascope and a number of clients for the last four years, and we are excited to formally join forces in order to scale those efforts.
Together, we will create an offering we’re calling D4AI: Design for Augmented Intelligence. The choice of “augmented” rather than “artificial” isn’t just fancy wordplay — augmented intelligence extends the capabilities of humans in a way that feels natural. If the intelligence of our devices, systems, and relationships feels artificial, it will never stick.
Augmented intelligence will only be authentically woven into our lives when it feels warm-blooded and responsive — as predictably unpredictable as we are. It’s the surge of emotion when Google Photos recognizes an anniversary and surfaces a photo you took on the last day you were with your grandfather, who died last year. Google Photos assistant feature has moved from being a simple utility to an almost daily extension of memory.
If the intelligence of our devices, systems, and relationships feels artificial, it will never stick.
But for every smart moment, there are countless bungles, too. Our digital assistants can seem utterly clueless — or worse, heartless. If we take the long view, we’re probably in the Cambrian era of intelligent machines. Our approach will be to go in humbly, with our hats off, knowing there’s a lot we don’t know, and much hard work to be done. That said, the promise of artificial intelligence is to make smart things feasible. And D4AI hopes to make them desirable. Humans will always see things machines can’t see, and vice versa.
In starting D4AI, we find ourselves at a moment very similar to when IDEO pioneered Interaction Design nearly 30 years ago. Bill Moggridge, the designer of the first laptop saw how essential it was to design interactions between people and computers that were rooted in human needs. At the time, many people were skeptical of the impersonality of software interfaces. But when treated as a design medium in the context of interaction, interfaces sprang to life in useful and previously unimaginable ways that now fuel digital transformations around the world.
The same can be said about applying design to data-driven interfaces. There is widespread belief that the value of data is unassailable. But that’s simply not true. Data by itself is inert — dumb raw material. Making things smart will mean designing with data in a way that reflects and responds to human behavior. That means making it dynamic. Flexible. Evolutionary. It will have to exist in relationships. And relationships, as we all know, are complicated. Moreover, we now have relationships not just with each other, but also with our phones and networks and bodies and cars.
At IDEO, we think a lot about those relationships. When you’re in your car, for example, it may be communicating with you, your phone, and also its surroundings, as it responds to data collected from sensors trained on the road. In that scenario, who has the agency? Who’s in charge? You? Your phone? The car? A driving service in the cloud?
Data by itself is inert — dumb raw material. Making things smart will mean designing with data in a way that reflects and responds to human behavior.
Designing with data in a way that augments our connections to people and things requires (1) having enough technical knowledge to know how to work with the raw inputs, and (2) having enough of a design mindset to make those inputs relevant.
Afterall, every data set is mediated by people. People identify data as important, record it, and analyze it. People type the billions of search queries that enable search engines to continuously improve. People repeatedly turn the temperature down on a smart thermostat; only then can the thermostat learn to turn it down for us. An engine’s “intelligence,” then, relies on our understanding human behavior and our ability to programmatically harness our data in a way we find useful.
There is an element of human intention and historical bias to be found in data, too, and we should acknowledge that. D4AI’s ability to create intelligent systems that foment meaningful change will rely on combining irreplaceable human insight and the power of machines. Not an easy thing to pull off, we recognize, but an aspiration in which we’re investing.
Here’s a useful example of how data alone can miss the mark without the help of human-centered design: Today, kids are the sensors that tell us there is lead in house paint. It chips off the wall, they eat it, and are poisoned. It’s not until they develop symptoms that alarm bells go off.
The Data Science for Social Good working group thought they had solved the problem for Chicago’s children. A public-health emergency that would have taken 10 years and $77M to fix could be addressed in three months for only $1M with the help of a data model they had developed.
But there was a problem: their model required knowing the whereabouts of every pregnant woman in the city — information that was neither public nor available. Eventually, they found a human-centered solution rooted in another insight, one that pure data could never have predicted: Women who lived in old buildings most likely to be lead-contaminated were also the least likely to seek prenatal care. Based on that insight, the group changed their focus to providing prenatal care in those neighborhoods and simultaneously inspected the buildings where children lived or would soon be born for lead paint — serving two needs at once.
The story is a reminder that predictions are easy, but people are complicated. The right kind of data is not always available or useful or a smart shortcut when you’re trying to address multifaceted human issues.
Predictions are easy, but people are complicated.
In studying stories like this, Datascope’s cofounders would ask themselves What would IDEO do? Likewise, IDEO and its clients face increasingly complex data and design challenges and ask What would Datascope do? Datascope takes as a starting point the human needs inherent in a problem rather than the data that might seem most obviously relevant to that problem. While many other firms in their space might see themselves as data engineers implementing solutions at massive scale or data scientists narrowly focused on researching new statistical models, Datascopers see themselves as designers — designers who work with data and algorithms to create human-centered AI.
Together, we believe that data-driven augmented intelligence is fundamentally changing the world around us. It’s time to make sure we design that intelligence in ways that make us — and by us we mean every human being — better.