Plagiarized material about plagiaristic tech, Google embarrassed, an excuse to unsubscribe

Photo by Jonathan Kemper / Unsplash Saying goodbye doesn't have to be hard.

Irony Alert: LLM paper contains plagiarized material.

A Turkish engineer was asked to review a deep learning paper and discovered his own writing in it.

💡

"When Çavuşoğlu first skimmed through the other paper, he came across some sections containing an uncanny resemblance to his own ideas. 'I thought, it's like I wrote that,' he recalled. 'How could it be so similar, did we think about the same thing?' "

Google hastens to roll back AI after embarrassment.

The New York Times continues to be the most reliable source for testing the claims of AI giants. Less than a week after its release, the AI Overview feature failed to make the same essential claims that CEO Sundar Pichai demonstrated in a live feature demo. We've been running mini-testing sessions at Singular XQ. The most important thing for the entrepreneur and business person to know is that single tests are not an adequate demonstration of tools that require stability to perform well for users. A single test can perform well 98% of the time; the truly innovative trick is if it can perform well and consistently over time, demonstrating model stability. Model instability--and related problems you may have heard, like "model drift," "hallucination," and "overfit," is a problem that has plagued LLMs and neural nets since they were first conceptualized decades ago. Users complained loudly about the instability results on AI Overviews so it not-too-subtly disappeared.

There goes 30 years of faithful readership.

We were utterly disappointed in The Atlantic's decision to partner with Open AI. I've long admired Nicholas Thompson's social media videos explaining the top headlines in tech and looked to him for clear-headed analysis. This disappointed and baffled me. I may be imagining it, but his presentation of the news seemed a little less confident than other analyses he presents, even stepping further away from the camera than he usually does. We stand with the New York Times, The New York Daily News, Chicago Tribune, Orlando Sentinel, South Florida Sun Sentinel, San Jose Mercury News, Denver Post, Orange County Register, and St. Paul Pioneer Press in suing Open AI for illegally using their reporters work to train their models. When other outlets--particularly long-form journalism magazines--partner with them, they betray their colleagues in the field making a stand. It's a new form of scabbing. If the New Yorker falls, I may have to go to bed for a week.

Here are the other media deals.

The bright side for me is more noise to cancel out. Here is the complete list that I know of of this writing and I will not quote these outlets or refer to them for anything. Bye, Felicia.