This Isn't Hype Anymore

AI is doing administrative work right now. Not in five years. Not in a lab. Today.

Hospital for Special Surgery in New York — a real academic medical center — is running AI agents that process 1,100 insurance claims per month, according to MIT Technology Review. Those same agents cut the appeals process from 45 minutes down to five. The success rate on those appeals jumped from 65% to 100% over nine months.

A task that took a human nearly an hour, with a one-in-three failure rate, now takes five minutes and doesn't fail.

The Healthcare Math Is Brutal

The World Health Organization has warned the global healthcare sector will face a shortfall of 11 million workers by 2030. That's WHO's own published data. The system is already breaking. More than two-thirds of healthcare providers — 68% — have already adopted AI agents into their workforce, according to KPMG research cited by MIT Technology Review.

The BC Woods School of Business at Boston College puts the global AI healthcare market at $19.27 billion in 2023, growing at a compound annual rate of 38.5% through 2030 to reach nearly $188 billion.

Without the available workforce, AI fills the gap or the gap doesn't get filled.

What Mainstream Coverage Gets Wrong

Most tech journalism on this subject does one of two things: it either sells AI as a magic cure-all or it runs breathless warnings about job destruction. Both framings miss something important.

Ashis Barad, MD, chief digital and technology officer at Hospital for Special Surgery, is direct about where previous digitalization failed: electronic health records migrated in the early 2000s, but the data is still fragmented and reliant on manual inputs. Telehealth tools improved access but never replicated the quality of in-person care. The tech was added on top of broken workflows instead of replacing them.

Agentic AI operates differently. Unlike earlier tools that choked on edge cases, AI agents can handle nuanced scenarios, make autonomous decisions, and pull from expert clinical databases without punting back to a human every time something unusual comes up. Dr. Barad describes it as collapsing and supercharging the workflow.

Small Business: The Overlooked Story

While healthcare gets the big headlines, the small business angle is being almost completely ignored by major outlets.

MIT Technology Review profiled Sam Finnegan-Dehn, a part-time math and philosophy tutor in London who uses Notion AI to manage client notes, record meetings (with consent), draft invoices, set goals, and schedule social media posts. He's NOT using AI to write lesson plans — he's using it as a second memory to connect information he's already captured.

Large corporations have always been able to hire accountants, marketing managers, executive assistants, and operations staff. A sole proprietor in London or a plumber in Ohio cannot. AI is closing that gap now.

The tools already available — Claude, ChatGPT, Notion AI — handle meeting summaries, invoicing, goal-setting breakdowns, and social media coordination. Finnegan-Dehn is using them today.

The Governance Gap

WHO launched the Global Initiative on AI for Health in July 2023 — a joint effort with the International Telecommunication Union and the World Intellectual Property Organization. Their published guidance acknowledges a "significant pacing gap" — technology is outrunning legal frameworks.

WHO Director-General Tedros Adhanom Ghebreyesus has said AI could become "another driver for inequity" if universal access isn't prioritized. Rich hospital systems deploy AI agents that crush insurance appeals. Underfunded rural clinics are still faxing paperwork. The efficiency gains don't distribute themselves.

The BC Woods analysis flags the same problem from a different angle: AI in healthcare administration helps "identify cost inefficiencies" and "detect ineffective medical practices." But who controls that data? Who audits the decisions? Right now, the answer in most places is: nobody with enforcement power.

The Bottom Line

If you run a small business, AI admin tools are cheap, available, and genuinely useful for the basic stuff. Stop paying a bookkeeper $80 an hour to do what Notion AI does for $16 a month.

If you're a patient, the AI your hospital is deploying is mostly handling billing and scheduling — not diagnosing you. Someone should still be auditing those insurance claim decisions, because a 100% success rate on appeals warrants scrutiny.

If you're a healthcare worker, the honest answer is that AI isn't taking your job in the next five years — it's taking the parts of your job that were making you burn out. Whether that's actually good for you depends entirely on whether your employer passes the savings back to staff or just pockets them.

The technology works. The governance is lagging. And the people who will feel the consequences either way are NOT the executives running the press releases.