The Demise of LLMs
So the news is talking about how LLMs can be "nazi" and the tech news speaks of how "code (or coding) assistants" (by which they mean slop applied to code) don't really save time and actually introduce more problems than they solve.
LLMs were never really useful. There was never a promise, except totally fictional "valuations", based on false marketing by companies that lost a fortune, then claimed to be "worth" hundreds of billions of dollars. NVidia temporarily benefited from this bubble which they had created. It won't last.
We've just checked BetaNews again. They've dropped all the slop and went back to human authors. That really says a lot.
Some other sites will reach the same conclusion or eventually go offline.
Take for example linuxsecurity.com. It is very obviously just LLM slop, as usual:
From the structure alone, not just the words, it's obviously fake. It says nothing of substance. linuxsecurity.com will die because it's just SEO spam.
We've not yet seen a use case for LLMs; McDonald's just had a massive blunder due to its experiments with LLMs. It's just way too risky and many companies eventually accept they were conned by false promises. Waiting longer for the "next version" or "next thing" will fix nothing.
Microsoft is still trying to frame its business failure as "investment" in "hey hi". It has done this for quite some time already. █