A research mentorship program connecting
experts addressing risks from AI with aspiring researchers in the field.
SPAR is a part-time, remote research program pairing aspiring researchers with professionals addressing risks from AI to work together on impactful research projects. Mentees gain research experience and guidance, and mentors get assistance with their projects.
Mentors can choose to commit anywhere from 2-10 hours/week.
The program runs for 3 months with options for extension, culminating in a Demo Day attended by top AI safety orgs.
We typically look for mentors with research experience matching that of a late PhD student or a MATS scholar. We're excited to receive applications in technical AI safety, policy, or governance, or biosecurity research.
SPAR supports a broad range of research that we think is helpful for helping society mitigate the risks posed by advanced AI, including but not limited to:
Ensuring powerful AI systems follow human intentions
Governing the development of transformative AI
Hardware-enabled governance and preventing the theft of model weights
Reverse-engineering neural networks and their internal representations
Preventing or preparing for AI-enabled bioattacks
Understanding the effect of transformative AI on society and the economy
Gain access to qualified applicants to advance your research proposals. Who you choose to work with is entirely up to you, and there's no commitment to take on any mentees if none are a good fit.
SPAR projects can apply for grants of up to $30k for research expenses like compute and API access.
Many SPAR mentors end up inviting their mentees back to work on additional projects after the program ends, or hiring them to their teams.
Help train the next generation of researchers to do crucial work like yours.
We typically accept applications from researchers with AI safety or biosecurity research experience matching or exceeding that of a late PhD student or a MATS scholar. We're excited to receive applications in technical AI safety, policy and governance, or biosecurity research.
If you are a senior graduate student, postdoc, academic faculty member, or professional at a lab, you can suggest any project within SPAR's focus areas. If you are a junior graduate student or a MATS scholar, your project should be adjacent to your existing research focus.
If you don't fit any of these categories, it must be very clear that you have legible credentials to lead the project you propose. We accept undergraduate applicants to mentor only in exceptional circumstances.
Submit a research project proposal in one of SPAR's focus areas. Mentees will apply to your project, and you can take on as many as you're excited about. You make all the decisions here, from what to ask in the application to which mentees to accept. You can even choose not to move forward with anyone if you don't find a good match.
The SPAR team will promote your project and handle administrative work. We'll also provide guidance for managing the application process, and, if needed, provide you with funding for compute, software, and other expenses.
Once you've chosen your mentees, you'll spend 2-10 hours a week supervising their work as they move your project forward. SPAR provides resources to support your collaboration and ensures mentees make steady progress by requiring regular written reports.
While SPAR officially runs for three months, many mentor-mentee pairs choose to continue their work beyond this point and may collaborate on additional projects in the future.
Your project may be a good fit if:
• It can be scoped for 3 months of work at 5-20 hr/week/mentee, at a team size you're comfortable supervising
• It has a clear theory of change for mitigating catastrophic risks from advanced AI
• It's focused on AI technical safety, AI policy and governance, AI strategy, AI security, or biosecurity
For biosecurity projects, we don't require projects to be focused on AI risks specifically, but we do require their theories of change to potentially be helpful in preventing or mitigating AI-enabled bioattacks (exclusively or in addition to other threats). Given the remote nature of the program, we typically won't accept projects involving laboratory work.
See here for examples of projects SPAR has accepted in the past.