Yafah Edelman
This stream will work on gathering and analyzing data in order to shed light on the driving forces behind AI and monitor its impacts.
Stream overview
Projects will be similar to Epoch data and trends outputs.
1. Producing databases tracking important metrics relevant to AI. Possibilities for this include: semiconductor fabs, algorithmic advances, staff at leading AI companies, employment data in fields impacted by AI, as well as gathering information on AI usage using surveys.
2. Data insights analyzing gathered data to reveal important trends, and to communicate about these trends effectively to a broad audience.
Projects might be individually smaller in scope than other MATS projects, allowing scholars to complete multiple potentially related projects over the course of MATS.
Mentors
Yafah Edelman is Head of Data and trends at EpochAI. She researches analytic empirical trends in capabilities, scaling, and datacenter CapEx.
Mentorship style
Scholars will have individual weekly meetings for half an hour with their mentor, as well as a group meeting with their mentor for half an hour. Additionally, scholars will attend Epoch’s weekly Work In Progress meeting.
Representative papers
https://epoch.ai/data-insights
https://epoch.ai/data
Scholars we are looking for
Some useful characteristics (don't need all of these):
- a background in statistics (e.g., majored in stats in undegrad)
- prior forecasting experience
- pandas/matplotlib knowledge
- background in AI and economics
- experience with AI benchmarking (running or analyzing results from them)
- enthusiasm about the work epoch has already done
- scientific writing – e.g., being able to take a statistical result about some data and put it in plain english people without as technical a background can understand
Scholars will likely collaborate with other members of Epoch in addition to their mentor
Project selection
Scholar will pick from a list of projects
Community at MATS
MATS Research phase provides scholars with a community of peers.
.webp)
During the Research phase, scholars work out of a shared office, have shared housing, and are supported by a full-time Community Manager.
Working in a community of independent researchers gives scholars easy access to future collaborators, a deeper understanding of other alignment agendas, and a social network in the alignment community.
Previous MATS cohorts included regular lightning talks, scholar-led study groups on mechanistic interpretability and linear algebra, and hackathons. Other impromptu office events included group-jailbreaking Bing chat and exchanging hundreds of anonymous compliment notes. Scholars organized social activities outside of work, including road trips to Yosemite, visits to San Francisco, and joining ACX meetups.