If you read my weekly In the Loop newsletter, you know the AI news cycle skews dark. Models get pulled by the government, layoffs get blamed on algorithms, chatbots hallucinate evidence, and the public has, by every available poll, soured on the whole thing. Some of that pessimism is earned. But it's not the whole story, and I think we lose something when the only AI coverage that travels is the stuff that scares us.
As I was putting together this week’s ITL, there were several stories that showed some of what AI makes possible, so I thought I’d share them with you. There’s no "AI will save us all" messaging here — just a handful of things that happened this month where the technology made a real, human difference.
Let’s start with Boston Children's Hospital, where I worked for a dozen years back in the day. John and Catherine Brownstein have been partnering with OpenAI for a couple years now, doing some truly innovative work. Most recently, they pointed an off-the-shelf model at the genomes of children with rare diseases — cases that had been analyzed by humans, repeatedly, with no answer. The model helped clarify 18 diagnoses. One lead researcher called the roughly 5% yield "a total game changer," especially for the families who had searched fruitlessly for a diagnosis – after years of not knowing, a name means everything.
In Abu Dhabi, a team separated conjoined twins joined at the head after more than 40 hours of operating. What made it possible wasn't just extraordinary surgeons — it was AI. The team used it to rehearse the entire procedure in mixed reality before touching the children, and to grow the skin needed for reconstruction. Only nine such separations have ever been done. This was one of them, and the girls went home.
A man was about to wire his entire $3 million retirement to a scammer. TIAA's AI caught the out-of-pattern withdrawal and flagged it — but here's the part I love: a human fraud team then spent hours convincing him he'd been deceived, eventually reaching his daughter to stop the transfer. As the CEO put it, "AI by itself would not have necessarily protected this person." The machine spotted the anomaly. The people closed the case. That's the partnership working exactly as it should.
Amid all the layoff anxiety, a quieter data point: a study in India showed that women moving into AI roles saw a 145% salary jump. Broader analysis this year points the same direction — the roles that amplify human expertise are growing faster, and paying more, than the ones AI simply makes easier. The gains aren't evenly distributed, and that's a real problem worth naming. But the simple story that "AI only takes" is wrong. For some workers, it's opening doors.
The Royal Botanic Gardens at Kew released a landmark report showing how AI and digitization could be a turning point in the "race against extinction" facing the world's plants and fungi. AI can learn to identify notoriously difficult species — sedges, peat mosses with microscopic features — far faster than people can, and unlock a "genomic goldmine" of fungal data we've barely touched. According to the study, “about 40% of the 70,000 plant species that have been assessed are at risk of extinction, while another 330,000 have yet to be analysed. There are also believed to be another 100,000 plant species still to be named by scientists.”
What strikes me about these five stories is what they have in common: in each one, AI was pointed at something only it could do. Spotting a diagnosis hidden in a genome analyzed by humans a dozen times. Rehearsing an impossible surgery before the first incision. Cataloging the living world before it disappears.
It’s easy to get lost in the doom-and-gloom of headlines meant to scare and divide. But there’s also a lot of good that’s coming from this technology. I, for one, was glad to get that reminder this week.