There's a version of me only Claude, my coworker, has met.
The 3am Melly. The one who asks the embarrassing questions out loud, because the robot can't make a face.
Turns out I'm not special. More than 230 million people a week now bring ChatGPT their health stuff. Symptoms, lab results, the "is this wart normal" they'd never say to a human.
230 million people. Whispering their health problems to a chatbot. Every week.
And OpenAI just made it better at exactly that. The newest model, they say, beats actual doctors at writing health answers. Quicker to flag a real emergency. Calmer about admitting when it isn't sure.
A little nice. A lot terrifying. The robot is now a more reassuring presence than the entire healthcare system. Fine. Normal. Cool.
But here's the one that got me.
Same day, OpenAI and Boston Children's Hospital shared something real. They took 376 kids whose rare diseases had stumped every doctor and every test for years. Unsolved. Dead ends.
They ran the cases back through the AI. It surfaced leads. Real doctors checked them. And 18 of those kids finally got an answer.
18 families who'd lived on "we don't know" finally got a diagnosis. After years.
That's not a chatbot guessing. That's the same kind of AI I use to make buttons, pointed at the worst years of someone's life, and actually helping.
inb4 "AI's coming for doctors." Nope. Today it helped them win.
Quick PSA from your unlicensed friend, though: the robot is not your doctor. Go see the one with the degree.
Me, I'll keep bringing Claude my dumb 3am questions. That's what it's for.
But the best part of my day was never the robot. It's this. You, a real person, here again.
Reading whatever my brain decided to wander off into today.
See you tomorrow.
-Melly