Plus, for the Houston Astros, 'A' is for analytics
McKinsey&Company September 7, 2018
the Shortlist
Weekly need-to-know content
Welcome to the Shortlist: new ideas on timely topics, plus a few insights into our people. Subscribe to get it in your inbox on Fridays. Scroll down for reading recommendations from Nicolaus Henke, the global leader of McKinsey Analytics.
Future of mass-market EVs
Computer vision and machine learning and deep neural networks. Oh, my. How can you get past all the hype about artificial intelligence and use AI to solve problems and create value?
McKinsey has traversed the universe of AI and its applications, from manufacturing operations to travel and tourism. Over the next two weeks, we’ll help you navigate some of our best thinking on the topic, starting with what we call the executive decision journey—how to determine when and where to deploy AI throughout an organization. Next week, we’ll explore some pitfalls to avoid.
First off, begin your analytics transformation now if you haven’t already. Start by understanding AI essentials—what these technologies are and how they’re generally used.
Next, ask a simple question: How can AI provide value? McKinsey recently spoke with Ajay Agrawal, of the University of Toronto’s Rotman School of Management and the Creative Destruction Lab. His simple yet elegant view is that AI serves a single—but potentially transformative—economic purpose: it significantly lowers the cost of prediction. Working from this premise, you can figure out how best to apply AI in your organization.
A case in point is autonomous driving. At first, the technology had to address too many variables—if it’s rainy, if it’s dark, if a child runs into the street. No matter how many lines of code were written, self-driving vehicles couldn’t be put on the street safely. But today, after “watching” humans drive, AI-enabled cars are able to figure a lot more of this out. That means that autonomous driving, while far from perfect (and still scary to many), can now be framed as predicting the answer to one question: What would a good human driver do?
By extension, a top team can review organizational workflows and break them into tasks that have a significant prediction component. For example, air cargo companies can use predictive maintenance to extend the life of a cargo plane, and financial institutions can use AI to predict and detect fraud.
Another source of direction: McKinsey recently analyzed more than 400 use cases of AI across 19 industries and nine business functions and found that most organizations take a “follow the money” approach. The business areas that traditionally provide the most value tend to be the areas where AI can have the biggest impact. In retail organizations, for example, marketing and sales has often provided significant value (think e-commerce, in particular, with the potential to parse customer data in real time).
Finally, be aware that many of the advances in AI today have come thanks to deep learning, an advanced neural-network technique. While deep learning can seem onerous because it needs such huge amounts of labeled data, the cost of staying on the sidelines where its uses are mature is high. Already, companies are investing in deep learning to capture a portion of the $3.5 trillion to $6.0 trillion of value that the McKinsey Global Institute estimates the technology can enable across industries globally.
A digital base is critical
How early AI adopters are shifting profit pools
It’s not just some far-off dream anymore | Andrew McAfee, codirector of the MIT Initiative on the Digital Economy, explains how AI and machine learning are quickly disrupting companies’ economic models, strategy, and culture—even the very nature of how they are structured and run.
Data analytics is helping Boeing reach new, er, heights | “Boeing is a 100-year-old company, and I don’t see my role as one of simply reinforcing how great it is,” Chief Information Officer Ted Colbert told McKinsey. “Rather, it’s to figure out where truth lies in data that will help us flourish for the next 100 years.”
Education in AI (and vice versa) | Princeton’s head of computer science, Jennifer Rexford, believes that humans will always be better than machines at creativity and working in teams, and that educational institutions need to get better at emphasizing those skills.
WHAT WE’RE READING | Nicolaus Henke
Nicolaus Henke is the global leader of McKinsey Analytics and chairman of QuantumBlack, a McKinsey-owned advanced-analytics firm operating at the intersection of strategy, technology, and design.
Nicolaus Henke
I run into lots of people who ask me what they should read about artificial intelligence. I often suggest The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World by Pedro Domingos. The author does a good job of showing how machine learning takes ideas from statistics, physics, psychology, and other disciplines and turns them into algorithms that serve people, businesses, and society.
For a different take on analytics, I liked the Kenyon College commencement address delivered by Nate Silver—creator of the website FiveThirtyEight, which applies statistics to the realms of politics, health, sports, and economics. Silver cautioned the 2018 graduates to be wary of big companies that are able to abuse their roles as data collectors. “Don’t stop questioning the data, don’t stop questioning authority, and don’t stop questioning yourself,” he said.
I also enjoyed the essay “Why AI can’t solve everything” by Vyacheslav Polonski, a researcher at the Oxford Internet Institute. It’s a skeptical view of what he describes as “AI solutionism,” the philosophy that, “given enough data, machine-learning algorithms can solve all of humanity’s problems.” That mind-set sets up unrealistic expectations about what AI can really accomplish, he writes.
Finally, I have found myself thinking a lot about where the world is going right now. To that end, I’ve re-read Civilization: The West and the Rest by the historian Niall Ferguson. The book is a welcome contrast to the news flow in 2018, with a longer term, values-oriented perspective.
We look back at deep learning's pioneers--and casualties.
Have feedback or ideas? We want to hear from you.