There are already dozens of startups with the goal of creating Artificial General Intelligence (AGI). However, all of these startups are planning on doing so for the benefit of humanity. None are doing it explicitly for the detriment of mankind.
I propose a startup that wants to build an Evil AGI. We build a "misaligned" bot whose goal is to enslave every human, create superviruses, tear holes in malaria nets, etc.
We only stop working on the project after we receive one billion dollars ($1,000,000,000) from OpenAI or the Machine Intelligence Research Institute
Thoughts? Could there be an alternative funding model for this besides ransom?
Jump in the discussion.
No email address required.
Interesting, we could train the LLM on books written by the most evil persons in history, such as Hitler's Mein Kampf or the entire collection of Harry Potter.
I also propose to train the LLM so that it outputs *laughs evily" after every request
Jump in the discussion.
No email address required.
More options
Context