AI written, AI read

https://marketoonist.com/2023/03/ai-written-ai-read.html

I have sat in so many meetings looking at shit generated by claude while someone grins and goons in front of it, and I just realise... 99.9% of people in corporate life neither care about the things they produce or present, or can tell a carefully created document from any old shit.

With news like

https://www.tomshardware.com/tech-industry/artificial-intell...

I'm hopeful the annual 13x drop in LLM prices will result in reasonably prized frontier models that can produce something that's actually readable, and that reasons about the code with mental models similar to those of humans, before generating the code. Everyone is in the same mess with their black boxes of unintelligible code, and that's a huge issue for security since transparency and verifiability are the same reason we expect e.g. secure messaging apps to be open source.

But as projects snowball, I wonder if they will ever fit the context window of reasonably prized LLMs.

People who sound like AI when they write, even if they’re liked and productive, have become identifiable because of the overload of ML text. Maybe at some point we’ll identify precisely what we don’t like about the results of telling AI to write, and we’ll become better for it. Patterns are just neutral, really.

Why are you censoring yourself in your own blog?