Seven McKinsey colleagues shared ideas that ranged from a deep-in-the-data look at AI to blue-sky thinking about flying cars at the 25th annual SXSW Interactive Festival in Austin, Texas, last week.
They were joined by more than 30 colleagues who flew in from five continents to connect with clients, chat with recruits, and catch glimpses of the future: the latest tech prototypes, products, and start-ups that will be shaping the way we live and work.
Our firm is no stranger to the event. In fact, Collin Cole, an Austin local and McKinsey Design expert, has been actively involved with SXSW over the years as a sponsor and speaker.
“The festival has mirrored the digital and design industries,” he says. “It’s always been a chance to explore and humanize each new wave of tech.”
Here’s how our colleagues made the most of that chance this year.
How healthy is your data?
Collin said he thinks AI will dominate conversations at the festival for the next decade. It certainly did at the panel led by Kasia Chmielinski, a product manager with McKinsey Analytics.
Joined by QuantumBlack legal counsel Alejandra Parra-Orlandoni, Government of Canada data and innovation director Ashley Casovan, and Massachusetts Institute of Technology (MIT) research scientist Vikash Mansinghka, Kasia described new tools and frameworks being developed in both the public and private sectors to reduce bias in data.
“Everyone is obsessed with ‘ethical AI’ and the bias in models,” Kasia explained. “But wait—the bias often comes directly from the data used.”
Kasia would know. Prior to joining McKinsey, Kasia worked with a joint Harvard University and MIT team at the Berkman Klein Center to launch the Data Nutrition Project. “We interviewed 100 scientists about what they do before they create a data model,” Kasia explained. “There were 100 different answers—a free for all.”
So the team developed a prototype that might increase transparency around data sets: the ‘Data Nutrition Label.’ “If you look at the contents of a candy bar, you might not eat it. This tool does the same for data: it tells you what’s inside the data set before your model consumes it,” said Kasia.
Alejandra, who has a background as an engineer and naval officer, draws on her training as a lawyer to help QuantumBlack data scientists, risk experts, and consultants think through the implications of choosing different data sets to train AI models.
For example, microtargeting populations to optimize a healthcare treatment or design a municipal service typically involves data sets that include sensitive information, like gender and age, as well as data that serve as proxies for sensitive categories, like zip codes for race.
Any of these, Alejandra explained, could lead to harmful biases or unfair outcomes when leveraged by an AI model. In addition, diversity in the demographics and experiences of the team selecting the data sets can also influence bias.
“We know that more diverse teams are better at spotting problematic data or anticipating harmful outcomes that could arise by using certain data sets,” Alejandra pointed out. “The degree of public attention on the topic of data bias is fairly new—and very promising. We’re racing forward with developing a structured way of thinking about it and embedding tools and steps to mitigate the risks into our client-service protocols up front.”
A disciplined way to create disruptive products
Dan Chuparkoff always thought he’d be an architect, but after seeing an early version of computer-aided design (CAD), he decided that it’d be easier to change the world with software than with buildings. At McKinsey, he leads teams that create healthcare-analytics solutions. This was his fifth time as a speaker at SXSW.
In a packed workshop, Dan walked an audience of software designers, product managers, and engineers through the concept of artificial neural networks, a branch of machine learning.
Using decks of cards and red/green stickers, each member of the audience acted as a neuron in their seated rows, or “layers” of the network. As cards circulated quickly through each layer, users looked to identify specific characteristics of the cards and passed them to the next layer.
The audience got it nearly perfect on the first try. The reason for the exercise? Many in the room will soon be using advanced-analytics technology like this to create their products, so they should understand how it works.
Dan’s ambitions are lofty. “Don’t just build a good product,” he told the group, “build a disruptive one that upends the status quo!”
But his approach is down to earth. “Do you have a fresh idea?” he asks. “Does it solve a need for massive number of people? Will it improve itself over time?”
Once you answer “yes” to those questions, he’s got more advice: look up government statistics to size a market, map work streams across industries, and use the power of technology to amplify your problem solving—hence the neural-network exercise.
The important thing, he stressed, is just to get moving. “After all,” he told the audience. “Any breakthrough is 1 percent the idea and 99 percent the execution.”
High aspirations for flying cars
Overlooking the clogged streets of Austin, a small group of aerospace and mobility experts—NASA’s Jaiwon Shin, Bell Helicopter Textron’s Michael Thacker, and McKinsey’s Shivika Sahdev—outlined a vision for the future of urban air mobility on a panel led by NPR’s Aarti Shahani.
Think on-demand flying cars: short, quick trips on flying vehicles, quickly and quietly dropping in and out of town. “Who believes this will happen by 2025?” asked the moderator. A third of the room raised their hands. In ten years? The majority agreed it was possible.
“What are the barriers?” pressed Aarti.
One panelist pointed to the complexity in coordinating and managing these low-flying vehicles above crowded urban cities. To scale such travel, it will need to be affordable and quiet, said another, which will require advanced technologies for battery capacity, autonomous flight, and advanced cloud computing. “Infrastructure,” added Shivika. “What physical spaces will be needed for these vehicles? Will they be on existing rooftops or in the outskirts?”
Cities, the group explained, are grappling with access, congestion, and pollution and are open to any new solutions that help them circumvent traffic and lost productivity. For emerging cities, points out Shivika, air mobility could be a chance to leapfrog more expensive ground-based transit for a way of travel that’s cheaper and more sustainable.
All agreed that such a vision will require a new ecosystem of support. “No single player can make this happen,” said Shivika. “Vehicle makers, tech companies, local and federal governments, and others will have to come together to make this a reality.”
Cultivating millennials
“Yeethoven” is what happens when you mash up the music of Beethoven with that of Kanye West. This Lincoln Center program attracted 900 attendees to the venue. Five hundred of them were new. All of them were under 40.
John Casavant, Kathryn Peterson, and Alex Sarian, all of Lincoln Center, along with McKinsey’s Marissa Solomon, detailed how they have used analytics to identify and create more programming, like Yeethoven, to attract the millennial market to the venerable arts venue.
Millennials are experimental, they’ve found, and always open to the next big thing. “They want a wider experience, starting before purchasing the ticket, including a social aspect during the event—food, happy hours, mixing it up—and following up with photos and reporting the next morning, ” says Marissa.
Getting this right has required the organization to collaborate like never before: tech, marketing, programming, and development groups are working together to capture data, develop new events, and use analytics to reach and track new audiences.
One audience member wanted to know how to get started using data in a similar way. “Launch one pilot, get some results, and then launch another,” said Marissa. “Don’t wait for building the perfect infrastructure. That can take decades.”