Physician Helen Ouyang argues that doctors shouldn’t reject AI:
A.I. may not replace doctors, but it will change what patients expect from us. Doctors need to adapt…
The reality is that many patients are already consulting A.I. Doctors can keep fearing or condemning those interactions, or they can figure out how to support people using A.I. tools for their health care — cautiously, with clear guardrails. I would never tell patients to ask ChatGPT or Claude for a diagnosis, but perhaps I would suggest they use it to make sense of a new condition or keep up with routine screenings — or translate “diet and exercise” into steps that actually fit into their lives, as I did.
She’s right. And she’s pointing at AI’s real potential. While much of the conversation is about the technology’s potential to replace work, the real story is how it can radically expand options.
Think about what that means in medicine. Patients who would never book an appointment for a minor concern can get a useful answer. Vague advice — “diet and exercise” — gets translated into steps that actually fit someone’s life. Doctors can spend their time seeing more patients or helping the patients who need them most, not fielding questions that don’t require a physician.
This is already how AI is playing out in other industries. Axios’ Jim VandeHei explains how it’s unlocking new possibilities in journalism: “The bigger opportunity isn’t efficiency. It’s new business lines that were economically impossible before AI.”
More people solving more problems that couldn’t be solved before.
Preparing for a recent trip, I downloaded For All Mankind season five — ten episodes of high-resolution video. It took about 20 minutes.
To get that much video 30 years ago, I’d have to schlep to Blockbuster, hope they had the tapes in stock, and pay around $40 in today’s money. Then watch them on a low-res TV before returning them two days later.
Downloading from the internet certainly wasn’t an option. In the late 90s, a single MP3 — three minutes of music — took an hour to download on a good day.
Today, though, my home internet connection runs roughly 10,000x faster than the dial-up of 30 years ago, for about 3x the nominal cost.
No one person or organization planned this improvement. It came about because millions of people — customers, engineers, broadband companies — each pursued their own goals.
That’s what progress looks like when it’s working. There’s no grand announcement. Just millions of people acting on their independent plans — until one day you’re downloading a season of television in the time it takes to pack your suitcase.
It’s an article of faith among many fundraising executives: Keep it short. Nobody reads anymore.
It’s not wrong, exactly. The direct marketers who tried to send fundraising emails as long as their letters — these letters can run eight pages or more — learned the hard way that email is a different medium. That people consume email differently than mail.
(Consider Jeff Brooks’ image of Aunt Ruth settling into her comfy chair to read her mail. Nobody settles into a chair to read email.)
But the people who took “keep it short” as the whole lesson often found their short fundraising emails equally ineffective. Not because they were short, but because they were only short. They’d edited out the emotional argument — the part that actually moves someone to give.
Direct mail wasn’t long because readers were more patient in 1985. It was long because building a case for why someone should care — and care deeply enough to write a check — requires space. The length wasn’t a feature of the medium but a reflection of human nature.
Email changed the form. The reader remained the same.
Teachers who’ve spent years assigning take-home essays are watching students produce polished work they didn’t write. The form — the overnight research paper, the five-paragraph essay — is being disrupted.
Some educators are banning AI entirely, doubling down on old forms and hoping enforcement holds. This misses the point.
The take-home essay was never really about the essay. It was about forcing students to think through the material and demonstrate they’d learned it.
The most interesting educators aren’t the ones fighting AI. They’re the ones asking what writing is actually for and redesigning their lessons around that.
That’s what disruption does. It strips away the form — the multi-page letter, the essay — and leaves intact the problem the form was always trying to solve.
Gmail taught us to be digital hoarders. To save every email, every chat, every search. Because it might come in useful someday.
All that data might feel like a security blanket. But it creates real mental load.
Do you really need that email from 2010 coordinating movie plans? The shipping notification from 2014? Backed up files you’ll never actually look at again — files that could be a liability in a breach?
This digital clutter can have the same effect as physical clutter — overwhelm, the sense you can’t quite move on since you’re worried about what if.
I recently cleaned out years of digital detritus — tens of thousands of emails dating back more than 20 years, gigabytes of data in online storage.
One account now holds 45 total emails — just those that were particularly meaningful. Everything else? No longer a weight.
Turns out nothing I deleted was ever needed. It never was.
Twenty years ago, posting an article to the website I managed took 30 minutes. Not 30 minutes of writing but 30 minutes of work after the writing was done.
Creating the raw HTML page. Updating the HTML on every page that linked to it. Fighting with Internet Explorer 6.
And that was after I’d already paid for Dreamweaver to simplify site management and spent months learning ASP and HTML well enough to not break everything.
It was handcrafted. And it was awful.
Last week I built an entire web app on my commute. My family has been watching films together for years, and I wanted something that would track what we’d seen and suggest what to watch next.
From idea to working app: 30 minutes. The whole train ride was spent on the thing I actually cared about — the feature set and the experience. Not painstakingly centering divs, or debugging a CSS error, or digging out solutions from the snippiness on Stack Overflow.
At one point, I asked my AI collaborator to add a custom date picker. At another, I asked it to create an import-export function for the database. Each change took 20 seconds to type and only slightly longer to execute. Almost magical.
The cost of entry? Twenty dollars a month.
Some people will say it doesn’t count if you didn’t write the code yourself. My kids, discovering a film they love on a Friday night, would disagree.
The tools used to eat your time. Now they give it back. That’s genuine progress.