What's our Hindenberg?

I usually read multiple books at once, adjusting to whatever mood strikes me. One of them is always nonfiction. Most people know Bill Bryson for A Walk in the Woods or A Short History of Nearly Everything. Right now, I’m reading One Summer: America, 1927. It’s the kind of book that makes you laugh out loud while simultaneously marveling at how far (and not-so-far) we’ve come.

In 1927, aviation was the craze. Bryson paints a picture of America swooning over the possibilities of flight. Americans were told wild things could happen: planes docking on skyscrapers, thrilling citizens as they glided overhead. The enthusiasm was infectious—practicality, however, was another story.

To set the stage, Bryson reminds us that 1927 America was a different place—about a third of the population of today, with half living on farms or in small towns (versus just 15% now). Yet it was no backwater. More than half the homes in the country had telephones, and the U.S. was adding more phones each year than Britain even had. America churned out 85% of the world’s cars. We didn’t have a viable highway system yet, so aviation felt like the future. Skyscrapers were even built with pointed masts to support airship docking.

Let that sink in for a moment. Airship. Docking. On top of skyscrapers.

Never mind that this idea was bonkers. As Bryson hilariously notes, “Imagine the Hindenburg crashing in flames on Times Square” or the streets below being drenched by the vast quantities of ballast water such ships would discharge. (The tiny drip from a window AC unit in the city suddenly seems downright polite.)

Still, this didn’t stop anyone from dreaming. Ideas abounded for city commutes by helicopter and gyroscope. Bryson writes:

“That none of this was in any respect achievable—in terms of engineering, architecture, aeronautics, financing, safety, building codes, or any other consideration—seemed hardly to matter. This was an age that didn’t like practical concerns to get in the way of its musings.”

I was reminded of this while listening to an episode of The Behavioral Design Podcast, where Sam and Aline interviewed Amy Bucher. (Side note: major FOMO for not being in the virtual room with three people I deeply admire or the equally excellent episode with Carey Morewedge.) Amy shared how much respect and admiration she and her data science counterpart at Lirio, Chris, have for each other’s fields.

Toward the end, Amy and Aline—both behavioral scientists who’ve worked extensively with AI teams in healthcare—discussed a phenomenon that feels all too familiar: the tendency to get way out over our skis about what AI and behavioral science can actually achieve.

It struck me because Bryson’s words about 1927 could just as easily describe today:

“Whatever the future held, everyone agreed that it would be technologically advanced, American-led, and thrilling. Curiously, it was the present that people weren’t so certain about.”

Nearly a hundred years later, here we are again. The tech we’re excited about has changed, but the story remains the same. Many people are boldly certain about the future these tools will create—often without much thought about how we get there.

It’s an age-old pattern: dream big, ignore the details. And while the dreams fuel innovation, they also come with a risk: overpromising leads to disillusionment. Today’s AI and B Sci evangelists might do well to borrow a lesson from America’s airship dockers. Before we chart a course to our imagined futures, we’d better make sure we’re clear-eyed about the skies we’re actually flying in. 

For America in 1927, that meant proving large scale passenger air travel was feasible and safe. Ultimately that meant landing planes a bit further from the city center (did you know Chicago’s O’Hare airport code is ORD because it was Orchard Place - a farming community full of orchards - before it was an airfield?) instead of docking on Daniel Burnham’s latest. It meant taking time to build trust that planes could safely take off (and land!) consistently. Remember that part of what made Lindbergh’s feat so exhilarating was how many people before him had tragically failed. 

Healthcare moves at the pace of trust. While many are excited about what AI can do, doctors and patients alike are wary. By focusing first on areas where the risks posed by hallucinations or unknown biases in the data are lower, we build confidence in what might come next. Similarly, by defaulting to overly simplistic behavioral science models that lack an evidence-base (see Elina Halonen’s excellent writing on this for more), both product teams and patients lose confidence in the value of investing in evidence-based behavior change approaches. Product teams need real behavioral science expertise embedded in them so their technique list is more than triggers and nudges. In both cases, it may mean going slower, but also means the solutions don’t blow up in our faces. The best case scenario is the one Amy describes, where the people with data science and behavioral science expertise know enough about each other's expertise to move the product forward separately (at pace) and have enough respect for the expertise, to engage each other before putting patients at risk.