Health Giant Files Wrong Copyright Claim
Also AI fails basic math and Xerox buys a printer.

The Vengeful V-Card: When HR Discovers Copyright Law

United Healthcare, a company that usually spends its energy managing deductibles and explaining confusing Explanation of Benefits documents, has now decided to wade into the choppy waters of intellectual property. The company is using the Digital Millennium Copyright Act, better known as the DMCA, to try and scrub a few unflattering images from the public internet today. This is what happens when a powerful legal tool is handed to a PR team that needs a public image crisis to just stop.

The problem is that the images in question, of an individual named Luigi Mangione who they would rather forget, are quite clearly not owned by United Healthcare, making the claim approximately as valid as asking the janitorial staff to rewrite the company bylaws. It is the bureaucratic equivalent of grabbing the wrong stapler from the supply closet and then trying to get a restraining order against the employee who owned it. A comment thread notes the absurdity; one user, `davefitt`, points out that a DMCA takedown is only a few clicks and a quick signature away, and the legal review process is clearly as stringent as the average Slack emoji reaction. The only thing worse than a company getting hacked is a company trying to manage public perception using the wrong end of the law.

The Intern Tries to Do the Budget in LLaMA 3.3

The question of whether the new cohort of Large Language Models, like the recently-released LLaMA 3.3, can handle anything beyond very high-level concept discussions is still very much a live issue. A mathematician’s recent analysis suggests that while AI can talk a great game, it still cannot handle formal mathematics proofs. This means the AI is a great pitch deck designer, but a terrible engineer.

The consensus from the expert, Professor Kevin Buzzard, seems to be that the system is exceptional at generating plausible-looking output, which is unfortunately a problem when the job requires correct-looking output. The AI can write a brilliant summary of a complex proof, but when asked to verify the actual steps, it fails to connect the dots. In corporate terms, the AI aced the cultural interview, but when put in front of the spreadsheet, it just started making up numbers that sounded good.

Critical Healthcare Information Trapped in Adobe Reader

It turns out that highly-critical, life-saving cancer treatment guidelines are still, in many cases, published only as giant PDF documents. This is a classic example of critical information getting stuck in a technological cul-de-sac. A blog post highlights the absurdity of having to Ctrl+F a 500-page document when a medical professional needs to make a decision about a patient’s next steps.

The guidelines are complex and change constantly; two variables that make the PDF format a terrible archival solution. However, because the process has always been done that way, the medical industry has achieved peak bureaucratic entropy. It is easier to maintain the existing, suboptimal system than to refactor the entire workflow to use a database. It is the equivalent of the IT department only accepting trouble tickets via fax machine.

Briefs

• M&A News: Xerox is acquiring Lexmark. This is like two old copier titans merging their respective fleets of slightly-broken office equipment. Now they can both manufacture printers that refuse to print in color, together.
• Hardware Nostalgia: A video on the world's largest tube TV is trending. The most popular piece of entertainment today is a look back at the heavy, hot, high-latency technology of yesterday, which sounds like most enterprise software I use daily.
• The Great Tea Bag Conspiracy: A new study confirms that commercial tea bags release microplastics into human cells upon steeping. We cannot even enjoy a soothing cup of herbal infusion without also ingesting a tiny bit of corporate waste. At least it is a consistent brand experience.