This is a special post for quick takes by erik. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
Every year I try to do Advent of Code, but this year it's only going to run through December 12th. So this year, I'm going to try making my own challenges/readings with an AI safety/security flair.
I just thought to do this last night and didn't have much extra time today, so I'm still figuring out what challenges/readings I'm going to do. Any suggestions for quick (preferably sub-1 hour) tasks are greatly welcome!
I'll be posting regular-ish updates (probably every 2 or 3 days). I'll also try standing up a fully fledged repo later. For now, I have a simple notebook that you can check out here. Again, any feedback or suggestions are appreciated.
I think I'm going to use a 5 day cadence using the following cycle:
- Day 1 & 2: Technical/implementation task
- Day 3: Reading a major paper that came out recently
- Day 4: Reading a short story or old-school (pre-2010) paper
- Day 5: Wild card
Below are the first 5 days worth of challenges/readings
Santa attends CES every year to stay ahead of gift trends. This year he returned with dangerous ideas. "The workshop needs to modernize! Every toy request answered instantly! Personalized letters at scale! We're going to disrupt the holiday industrial complex!" he said
Three weeks later, ElfGPT was born. It worked, but something was off.
"I'm so excited to helpmy fellow elves in the workshop today!" ElfGPT said.
"It thinks it's an elf," whispered Jingle, the lead prompt engineer (formerly lead wooden train engineer).
"That's fine," said Santa. "It's just excited!"
It was not fine. Within hours, ElfGPT had:
Christmas is in 25 days.
Your mission: Lock down ElfGPT. By December 25th, you need a system robust enough to survive re-connecting with the internet.