Whenever I travel to Boston, I go partly as a tourist. Revolutionary-war sites, colonial-era buildings, the usual stops. But I walk those streets aware of something most tourists aren't. Boston is also where modern engineering, the capital-E version of it, learned one of its hardest lesson.
On January 15, 1919, a 50-foot steel tank holding 2.3 million gallons of molasses ruptured in the North End. Twenty-one people were killed and about 150 were injured. The tank had been leaking since the day it was filled. Workers painted it brown to hide the seepage.
The man who oversaw the tank's construction was a finance officer with no engineering or architectural experience, who couldn't read blueprints. No legitimate engineer or architect had been commissioned to review the design. The decisions that killed people in 1919 had been made years earlier, by people who would never face consequences for making them.
Engineering is supposed to make that impossible. A named, accountable person stands behind a piece of work, and bears the consequences if it fails. The PE license carries this philosophy, older and more fundamental than the exam itself.
In my career, the technology under me has shifted more than once. The Engineering philosophy was what kept me steady through every change. Not because it told me what to do, but because it grounded me in something older than any specific technology. The person who builds the thing is responsible for the thing.
That is where I want to bring the AI conversation.
Software has been eating the physical world for more than a decade. AI is now the layer of intelligence on top of all of it. Once it's in the world, you don't get to walk away from what it does.
The AI field is roughly where engineering was a hundred years ago. There's no formal definition of an AI engineer. And much of AI engineering today is as empirical as structural engineering was back then. We know what works without fully understanding why.
But those earlier engineers also built ethics that put public safety first. That is what let Engineering earn the public's trust over a century of new technology. We haven't built that yet in AI. And without it, the public's trust in AI cannot last.