I've fixed 1 and 2, but point 3 is the point of the article. It was in response to a misleading post ... much boosted/repeated and much liked ... claiming that it was illegal for shops to refuse cash.
And in the past when I've tried to alter titles to be "more reasonable" they've been changed back, so I've given up trying to be helpful.
I was talking with someone today and they hadn't heard this story.
Some of my war stores are really close to this, but I'm not allowed to tell them yet, so I'm posting this in case you are one of today's ten thousand[0].
People flooding the 'net with LLM-generated crap are both eating their seed corn and poisoning the well.
Steel manufactured before 1944[0] can be incredibly valuable[1] because it hasn't been tainted by the nuclear tests/bombs fallout, and maybe fairly soon any archive internet material written pre-2010 will be considered equally valuable.
Not to mention pissing in the pool and tragedying the commons!
(By the way, the radioactivity in the atmosphere that was poisoning the steel has pretty much vanished by now, so don't go investing everything in battleship reclamation in 2024.)
I dunno - this seems like a problem ideally situated for adversarial networks to resolve. All you need is a smaller, dumber, validated model that can check and validate the output of the bigger, smarter model, and either use that to directly act as a filter between the user and the larger model, or use it to retrain the larger model until it isn’t dumb and wrong.
Not sure what point this is trying to make, but steel is a man-made alloy, not a naturally occurring substance (except possibly in tiny quantities by chance). The background radiation gets into it through the use of atmospheric air in the production process, so it seems rather a good analogy for background AI output entering the production process of internet "content" and rendering it less suitable for certain purposes.
Language is entirely human made , and it s affected by both communication and the medium. Every time a new mode of communication was invented, language was polluted with other people's language. AI is both a medium and communication which indirectly communicates using the "medial value content" of all other people on the internet. That content was itself already contaminated via too much communication over the past 30 years. So it's not really a fundamental shift , rather an acceleration of a process that s already happening
5 minutes later, someone will post about it on HN, explaining how we could all use this character to shield our private online writings from Big Tech's AI crawlers. 5 minutes after that, someone will change the defaults on nginx and Apache to output that character as the first character of every HTML web page.
Absolutely, except it will 100% go down the Do Not Track way. Nobody will care. I mean, at this very moment AI companies are training their models on data that's been obtained in a less-than-compliant way...
reply