Čauky Mňauky

I used to think “smaller subs can’t be infested with bots, right?” Then, over time, I noticed a slight, but ever-increasing trickle of mistakes a human would never make, often in seemingly well-thought out and empathetic replies. At closer inspection, they turn out to be just very nicely phrased, mundane truisms, or their facade totally falls apart, revealing weird, bland, pointless nonsense.
LLMs seem to be doing very well for English and, I keep reading, Chinese, because there is so much material to train them on. Their quality rapidly decreases with the number of speakers a given language has, though. I’ve learned to quickly spot LLM output (or so I tell myself) in my native language, and it’s always so disappointing.

Years and years before the pandemic, I frequented a popular sort of self-help forum. There was always this one “guy” that had a perfect response for every and any question you threw at him, usually within 30 minutes or so. Always at least one full paragraph. 24/7/365. At one point, I remember seeing that he had given more than a million responses. In hindsight, that’s clearly not human behavior, but at that time, I had no idea what was going on.

I share a similar sentiment, but I’d place the turning point somewhere between 1 and 2 GHz.

Wow! That’s awful. Funny how DDT was seen as an “improvement.”

I keep mine at the angle grinder factory. How about you come and get it, Elon?

Deregulation, ain’t it great.

Reminder that any sort of workplace surveillance is outright illegal in some EU countries. It could even be considered a criminal act.

Nice racket. First you pay the airlines for their tickets, then the ICE with your tax dollars to buy your data from said airlines.