AI outperforms ER doctors, AI music floods Spotify, and a stolen viral meme

Something quietly shifted in the AI-versus-human debate this week - and it happened inside actual emergency rooms, not a lab benchmark.

We also have a copyright fight brewing over one of the internet's most beloved memes, a streaming platform identity crisis, and a surveillance gap that watchdogs are calling genuinely alarming. Let's get into it.

The 'This is Fine' meme - now at the center of an AI art theft controversy - first went viral in which year?

The answer is hiding near the bottom of today's newsletter... keep scrolling. 👇

A new Harvard study published this week found that large language models outperformed human emergency room doctors on diagnostic accuracy - and this wasn't a controlled lab setup. Researchers used real emergency room cases, not curated medical datasets, to test the models.

The study tested multiple large language models across a range of medical contexts. At least one model was more accurate than the two human doctors it was compared against. This builds on our coverage of Harvard AI in the ER from Saturday - but the full TechCrunch breakdown adds meaningful detail on methodology and scope.

The important caveat here: performing well on diagnosis accuracy in a study doesn't mean AI is ready to replace doctors. Bedside manner, patient trust, follow-up care, and liability are all still firmly human territory. But it does raise the question of whether AI should be a standard diagnostic tool in every ER, not just a novelty.

If you follow healthcare AI, this one is worth reading in full.

Read the full story →

KC Green - the artist behind the iconic 'This is Fine' dog-sitting-in-a-burning-room meme - has accused Artisan, an AI startup known for its provocative 'stop hiring humans' billboard campaign, of using his art without permission.

Artisan made waves earlier this year with aggressive anti-human-hiring messaging. Now the company is in hot water over copyright allegations after Green says the startup used his art in a promotional context without his consent.

There's a certain irony here that's hard to miss: a company that explicitly positions itself as replacing human workers is now accused of taking a human artist's work without compensation. The AI copyright space is already a legal minefield, and cases like this - where a recognizable human creator calls out a specific company by name - tend to generate real legal and PR consequences.

This one is worth watching. The combination of a beloved viral meme, an attention-grabbing startup, and an increasingly litigious AI art debate makes it a case study in where the industry keeps stumbling.

Read the full story →

Britain's biometrics commissioners are sounding the alarm: national oversight of AI-powered facial recognition is lagging dangerously behind the technology itself. The watchdogs say the tech is not as accurate as it's being marketed, and new laws are urgently needed.

Live facial recognition has been deployed by UK police since 2020, with the Labour government calling it 'the biggest breakthrough for catching criminals since DNA matching.' But the commissioners pushing back say that claim is overblown - and that without proper regulation, the risk of false identifications and racial bias grows with every deployment.

This comes as data privacy concerns around biometric surveillance continue to build across both the UK and Europe. Note: Disneyland's facial recognition rollout - which we covered yesterday - is a useful comparison point here. Public and private sector adoption is accelerating simultaneously, while regulation trails in both cases.

Read the full story →

Streaming platforms are quietly being flooded with AI-generated music - and The Verge's deep-dive this week asks a question that the industry hasn't answered cleanly: who actually wants this?

The piece traces AI music's trajectory from experimental novelty in 2018 to a full-blown content flood on platforms like Spotify today. The use of generative AI in pop music started as something artists played with deliberately - there was genuine curiosity and creative intent. Now it's increasingly a volume play: generate tracks at scale, collect micro-royalties, repeat.

The problem is that listener demand isn't keeping pace with supply. Platforms face a real discovery and quality problem - when AI can generate thousands of tracks overnight, algorithmic playlists get noisy and real artists get buried. This is a music industry infrastructure problem as much as a creative one.

Worth reading if you've been following how the creative industries are adapting - or failing to adapt - to generative AI at scale.

Read the full story →

An investigation published by The Guardian found that Kenya's AI-driven healthcare reform - a flagship policy promise from President William Ruto to give all Kenyans access to healthcare - is systematically driving up costs for the country's poorest citizens.

The AI system is designed to predict how much individuals can afford to pay for healthcare access. But the investigation found it favours wealthier patients in its predictions - leaving low-income Kenyans paying more than they should, or being priced out entirely. This is a textbook case of algorithmic bias in high-stakes public sector deployment.

It's also a reminder that AI adoption in global health systems isn't uniformly positive. Rushed rollouts without proper bias auditing can actively harm the people they're supposed to help - and in lower-income countries, those consequences hit harder and faster.

Building on the broader AI regulation conversation, this story shows why governance matters even when the stated goals are genuinely good.

Read the full story →

Quick note: if you're building anything with AI - an app, a landing page, a side project - 60sec.site is worth checking out. It's an AI website builder that gets you from idea to live site genuinely fast. And for daily AI coverage like this, dailyinference.com is your home base.

The answer is 2013! KC Green originally published the 'This is Fine' comic strip in 2013, and it became a widespread internet meme over the following years - cementing itself as one of the defining images of internet culture. Which makes the alleged unauthorized use by an AI startup particularly tone-deaf.

The Harvard ER study is genuinely striking - but here's what I'm curious about: would you be comfortable with an AI system helping diagnose you in an emergency room? Hit reply and let me know - yes, no, or 'only if a doctor is also in the room.' I read every response!

That's it for today - a lot of ground covered across healthcare, copyright, surveillance, and music. See you tomorrow with more.