“A crowd-sourced collection of everything you shouldn't be building with LLMs. You're… not doing that right?” Made in twenty-four hours at the inaugural AI Hackathon London.
- When
- Apr 2023 – Jul 2023
- Stack
- JavaScript, Next.js, React, Netlify
- Outcome
- Won the “What could go wrong?” award at the London Generative AI hackathon.
The site opens on a wincing face and two doors: “😇 I'm feeling evil” and “😎 I'm feeling good”. Evil is the point — a growing, submittable list of the worst things you could build with a language model — and the good side is the same list's conscience, headed “None of that evil stuff, let's stay positive!”. Submissions came in through a form, were held in a spreadsheet behind a webhook and appeared on the site within the minute. The submit page tells the story in its own words: “What started as a playful idea to explore the darker side of the exciting recent developments turned into this website which gathers several experiments the LLMSAGG team made over twenty four hours together.”
A teammate's write-up put it more plainly: “It's a showcase of what could go wrong. The examples are mostly whimsical, but we got each of those examples up and running in well under an hour each. Working on this has significantly raised my concerns about this technology.”
Coverage
Ian Mulvany“London Generative AI hackathon - some reflections”2023
“llmsaregoinggreat - everything is scary!”
Around the same time
- Differentiated HomeworkProjects · 2023
- Autism EyesProjects · 2023
- How to scale your data teamSpeaking · 2022
- TranslucentWork · 2022
Earlier in projects: Aila. Later in projects: Differentiated Homework.
