Yearly report 2023

You can read the full report here.

At the beginning of 2023, we were halfway through our first pilot program - Future Academy - and Effective Altruism Sweden hosted our organizational operations with three part-time employees. Put differently, Impact Academy was in a very nascent stage. Given how early everything was, our high-level intention for 2023 was to explore, learn, and make a few excellent decisions. Looking back, we think we have had an exploratory yet moderately productive founding year. We completed one educational program and initiated another two. We also secured over $2 million in funding and established a strong organizational foundation by expanding our team and investing in creating systems, processes, and policies that will allow us to scale.

Below, we will outline some key Successes, Challenges, and Recommendations for (i) Education and Field-building and (ii) Organizational fundamentals.

Education and Field-building

Successes

During 2023, we:

• Concluded Future Academy's first iteration with moderate success and conducted a rather extensive evaluation. Key findings included very high satisfaction ratings (9.6/10 average), 92% of fellows reported changing their career in impactful ways, and expanding their networks (4.5x more relevant connections). You can read a summary of our evaluation with more in-depth metrics here.

• Explored over 15 potential new program ideas and systematically narrowed them down to 6 finalists. We assessed options quantitatively and qualitatively before deciding on Future Academy v2 (India edition) and an AI Governance Fellowship as key experiments for 2023.

• We initiated the second iteration of Future Academy, focused on India in 2023-2024. We ran a hiring round to find key team members from India and conducted some research to design an effective program. This led to a five-month hybrid program with three customized tracks and two in-person summits in India.

• Explored several approaches to building the field of AI Governance and ran a hiring round for a Head of AI Governance. We decided that the most promising option was co-running the EU Tech Policy Fellowship in collaboration with Training for Good (TfG).

Challenges and mistakes

• We can improve our approach to exploring and researching top options. Whilst it had good elements (e.g., the initial brainstorm) and led to positive outcomes, it could have been significantly more iterative and focused on action-relevant issues.

• We didn’t recognize the importance of external expertise swiftly enough for our Future Academy v2 (India) program. While we made hiring a co-lead and program associate the first priority once we realized the importance of bringing in relevant expertise, we could’ve clarified this at least a couple of weeks earlier. It turned out to be a more difficult hiring process than expected. That would’ve allowed us to include them more in the initial planning phase.

• We were not action-oriented enough. We spent too much time researching AI Governance without converting this into actionable knowledge for our program.

Organizational fundamentals

Successes

During 2023, we:

• Launched the organization and created solid organizational foundations, including:
  • Legal entities in Sweden and the UK,
  • an initial Theory of Change,
  • operational infrastructure such as a knowledge system,
  • CRM and key SOPs
  • A professional brand and a website.

These strong foundations will allow us to scale over the coming years.

• Secured over $2 million in funding with support from 3 funders, including Open Philanthropy and a funder outside the EA movement. Additionally, we built a strong relationship with some of our key funders, putting us in a good situation for future years.

• We grew the core team from 3 to 7 members while maintaining our culture. We are especially happy about investing in finding team members for our efforts in India, as the founding team lacked local knowledge and networks. In addition to growing the core team, we are pleased with engaging strong long-term contractors.

• Ville co-founded the Meta Charity Funders (MCF) with Gage Weston and Joey Savoie as his graduation project from the AIM’s Foundations Program. Meta Charity Funders is a funding circle that aims to fund effective projects working one level removed from impact. The circle conducted its first of two annual grant rounds in the fall of 2023, distributing  $686,580 to important meta initiatives.

Challenges and mistakes

• Despite initiating promising conversations, we failed to establish relevant partnerships for some of our programs. We could have invested more time in these efforts and had a more structured approach, including contingency plans.

• Some of our team members had long periods of very high workloads, leading to high burnout risks. We did not take this risk sufficiently seriously up front, and while we did a good job of reacting to the arising challenges, we should have been more proactive in setting up proper support systems.

• Not investing more in our grantmaking work and considering how it fits (or does not fit) into the overall work of Impact Academy. While the outcomes in these areas were above expectations, we think we could have done more good in total if we had prioritized this work over some other work. Additionally, we did not have a sufficiently clear theory of change in our grantmaking efforts, especially early on.

Goals for 2024

We will continue to explore but start finding our focus. At the end of the year, we want to have tried out multiple interventions and likely found at least one main program to stick with for the long run (while continuing to experiment and find additional programs). In this spirit, we’ve decided to focus all our efforts on the safe and beneficial development of AI (you can find a summary of our reasoning here) and continue to work in India in 2024 and 2025, while allowing for experimentation. We’ll also continue to build expertise, team, culture, and reputation relevant to running excellent educational programs. Below, we’ll elaborate on the intentions and goals for our main focus areas.

Impact Academy in India

We’ve operated in India since 2023, and our initial hypotheses about India's promise (e.g., a large talent pool largely neglected) have been strengthened. More concretely, while we don't know whether India should be the centre of our activities in the long term, we’ve decided to set up a new field-building organization in India (Axiom Futures) with Varun Agrawal as the Managing Director. Axiom Futures seeks to enable top talent to pursue careers that reduce global catastrophic risks from advanced AI. Some of our key goals for the year include:

• Conclude Future Academy v2 in a sufficiently excellent way;

• Successfully launch our AI safety programs and decide whether to continue them in 2025. In 2024, we will run two iterations of two categories of programs. We’ll run two part-time courses to introduce talent to technical AI safety in a high-fidelity way inspired by BlueDot’s courses. Concretely, we will run:
• • An in-person AI safety course at arguably the top technical institute in India: The International Institute of Technology (Bombay)
  • A pan-Indian online AI Safety course.
  • We’ll also run two full-time fellowships, in which people with a solid understanding of ML and AI safety will work on projects guided by mentors from leading organizations inspired by MATS’s programs.

Establish a solid organizational foundation.

This includes:

Finalizing and onboarding an excellent Indian core team (~5 FTE) to carry out the programs.

Create a professional and trustworthy brand (including a website) for our Indian programs.
Find the proper legal setup and secure the needed funding for 2024 and 2025.

This short document provides more information about our plans and underlying reasoning.

AI Governance

We have partnered with Training for Good to co-run the EU Tech Policy Fellowship (EUTPF) and potentially co-launch Talos Network - a new organization focused on helping people find high-impact careers in EU AI governance.

Some of our key goals for the year include:

• Running effective EUTPF Iterations. This includes:
  • Successfully conducting two iterations of the EUTPF.
  • Extending the second iteration with tracks exploring the EU’s role in international AI Governance.
  • Extending the team. This includes:
    • A new project manager for the EU Tech Policy Fellowship.
    • A Talent Lead to monitor meaningful opportunities for impact in the EU and ensure talented people apply for relevant roles.
• Decide and execute on the organizational contours of Talos Network. We aim to decide by September whether to co-launch Talos Network with Training for Good. If this concludes in the affirmative, we expect to execute an organizational spin-out in which members of Impact Academy (and Training for Good) join a new organization.

Organization and financials

The goals for the organizational and financial aspects of Impact Academy follow from our programs. Some of our key goals for the year include:

• Extending the team. Beyond the hires mentioned for our educational efforts, we will:
  • Strengthen the leadership team with a Managing Director or co-CEO, to prepare for our current CEO (Vilhelm Skoglund) to transition into doing more grant-making and fundraising;
  • Hire an Operations Specialist;
  • Explore hiring a Communication Specialist;
  • Create a strong board of advisors, with at least three people with distinct profiles;
  • Create strong, mutually beneficial partnerships for our programs;
  • Diversify funding. To date >90% of Impact Academy’s funding has come from one actor. We think it is important for us to have a more diverse and robust funding base, and we aim to raise >200k from at least 3 different actors in 2024.