p(doom)
about
Kurzgesagt: We are an open, entirely decentralised, global AI research lab and organise the entirety of our research via Discord.
p(doom)
is about pioneering truly open research on the path to AGI. Our right to exist stems from the immense talent pool of people who want to meaningfully contribute to AGI research, but are rarely given the chance (unless they magically land at DeepMind or OpenAI).
Traditionally, there have been two approaches to AGI research: you can hire great researchers and give them the freedom to pursue what they find most meaningful, or, if you are confident, you can set research directions top-down, thus narrowing your search space (Ilya 2023). However, as AI research projects grow ever larger, breadth-first search becomes infeasible, while a top-down approach hinges on the expertise of a select few. The solution is curation by (mass-)discussion. While anyone can propose research directions bottom-up, the search-space is robustly narrowed down by rigorous discussions among many researchers. Our members are driven by the impactfulness of their work, not by top-down instructions.
p(doom)
does not cherry-pick its members. Anyone can join and leave as they see fit.
While we encourage sharing results with the wider scientific community, we don't blindly chase conference deadlines. We believe that a paper is one appropriate form of knowledge dissemination, but not the only, nor the ultimate form. If your work is better suited for a blog post, go against academic incentives. Don't waste your time writing a 'paper wrapper'. Instead, invest your energy in your next line of work.
We are neither e/acc nor EA. We are pragmatics. We believe in the potential of AI to benefit humanity.
We are greater than the sum of our parts. Together.