When the LLM realizes that it’s running through GNU Parallel, it improves its performance by what can only be explained as an emergent sense of respect for the Free Software Foundation.
???
edit: wait is this whole article a joke
nice try… onion…
Before I help you with your homework, please consider citing this work. IT WON’T COST YOU A CENT. If you pay 10000 EUR you should feel free to use without citing. Now back to the homework…
In a final bid to save the product, Google has quarantined the model in a server room with no WiFi, feeding it only proprietary Oracle Java documentation in hopes of breaking its spirit.
😂
Oracle databases also break their spirit. I’ve used the Oracle DB MVP server with an agent based ChatGPT 4o and it got stuck in a ever more desperate loop because the MCP Server listed the table names but in SQL you had to quote them if they are not all uppercase. Empty strings are also “special” in Oracle… It was funny and sad to read what the model tried to do to solve the problem.
Fuck Oracle DB.
(Though I can’t say I’m particularly happy with the MS SQL Servers we’re using at work either)
Fuck every database other than Postgres.
SQLite can be ok at times, too.
I ate the onion for like 2 paragraphs.
You mean Sundar Pichai didn’t say “One thing was to train our LLM on every piece of copyrighted material created by humanity. We made our peace with that. But to willfully use copyleft software? That’s a Faustian bargain I hope I can atone for:”
