Event Kit
Hacking LLMs
One way to expose and shift awareness around ethical, security and privacy issues in today’s largest LLMs is to expose their workings by hacking them!
The Hacking LLMs with Feminism Repository was created to:
- Spark discussions around today’s LLM Development
- Demystify AI systems and their workings
- Expose and evaluate today’s LLMs with playful hackery
- Build collective understandings of the limitations of today’s approaches
This is part of a broader LAN party experience, which can hopefully create spaces for challenging status quos and sparking more feminist approaches over time.
New ideas and contributions to this kit (as well as new contributions on building truly feminist AI) are very welcome!
Recommended Agenda
Depending on your group size and allotted time, modify the following basic agenda.
- Setup: Take time to assist each other with setup and installation. More instructions in the repository Readme.
- Defining Feminist AI: Before opening notebooks, take some time to write ideas/words/goals of what you’d expect around a Feminist AI. What would a Feminist AI be?
- Notebook Hackery: With these discussions in mind, dive into the notebooks! With a larger group or short time, it might make sense to have each group choose a different notebook.
- Sharing findings: Shareout conversations and discussions from each group. What did you find? How does it compare with what you defined as Feminist AI?
Alternatively, if you have some ideas of notebooks, attacks, hacks or research that can elucidate these goals, you can create new notebooks as part of your LAN party. Please contribute back to the GitHub repository!