We read a brief post from and tried to come up with potential ideas to work on for Apart Research’s upcoming research hackathon from May 3rd to May 5th. Here’s some ideas we came up with and discussed:
Kappie & Welsh: Conduct an opinion survey on some demographic of local law students on their use of AI (specifically LLMs) in their academic work, and then write a report based on their findings.
Purplechair: Train a demo Twitter (X) bot to test its ability to stifle political discourse (spam, shutting the conversation down aggressively, etc.)
Purplechair: Replicate pre-existing findings on political bias in LLMs for non-Western democracies (Brazil, South Korea, etc.)
Gumperto: Convince a LLM to come up with/agree with uncontroversial policies with obvious loopholes, preferably policies where the LLM is aware of such a loophole.
Mogu: Test LLMs’ ability to fill in redacted words or sections of a document, specifically classified government documents.
Jord: Replicate findings about an AI’s ability to act as a sleeper agent, and test whether they are sensitive enough to detect important political events such as election season.
We couldn’t fully carry out our activity agenda as usual as the brainstorming session took a little too much time, but we spared some time to do our self-research:
Jord & Purplechair: Filled out their MATS applications
Kappie & Welsh: Read through BlueDotImpact’s AI Safety course on policy.
Mogu: Read through some miscellaneous technical research
Gumperto: Tried to get started on the ARENA 3.0 curriculum