BERI’s activities so far
BERI currently aims to act as a multiplier on the impact of existing researchers. Our main present activities are grant-making and the provision of free support and services to x-risk researchers, including but not limited to
- access to expert consultations, e.g., if an x-risk motivated PhD student wishes to pay an expert for a one-hour consultation on some topic, then BERI can pay for the service on behalf of the student.
technical support, e.g., hiring an engineer on a contract to remotely help a researcher with tasks like
- installing buggy software packages they need for research,
- formatting papers for conference and journal submission, or
- setting up and maintaining websites.
- administrative support, e.g.,
- helping organize an event run by an x-risk motivated PhD student or postdoc whose departmental administrators are not available to assist them.
We hope that activities like those listed above will not only accelerate the work and professional development of x-risk motivated researchers, but also show them a kind of social support: that other people value their work, and are ready to help them with it in ways that go above and beyond the levels of outside support academics have traditionally received.
Without BERI or other entities like it, these particular kinds of support are unlikely to exist in the near term.
BERI's marginal value
Often, academic grant moneys cannot easily be spent on items like those listed above, and when they can, professors are sometimes under social pressure to hire more researchers instead of support staff. Other times, they lack the time or personal network needed to identify skilled and motivated contractors for the required work.
Therefore, BERI’s main value is currently as a vehicle for hiring contractors and part-time employees for sporadic tasks in support of x-risk research and activities, as vetted by BERI leadership for their impact and necessity. BERI acts as a collaborating institution that is able to hire contractors for any task it considers part of a valuable project, so we can effectively provide the above services for free. As time progresses, BERI can accrue a network of contractors and other supporters who can be leveraged in favor of any project we choose to engage with.
BERI's edge over other institutions
Why is BERI able to do things that universities can’t?
Over the years, many restrictions and informal social constraints on university spending have arisen as agreements within and between institutions, to ensure responsible spending by diverse interest groups including professors, students, unions, budget committees, research support officers, purchasing departments, state departments, and so on. While useful for universities, often these restrictions have side effects that inhibit funding for activities that are otherwise responsible and very much on-mission for a given research group, but difficult for a university to approve within its existing framework. That doesn’t mean the restrictions are bad, but it does mean some good things are best done outside academia.
BERI achieves responsible spending in a different way from a university: by being small, individually ethical, and understandable to its donors, similar to the way a for-profit start-up is responsible to its investors. In particular, we only expect to receive donations from individuals who know enough about us that they have sufficient reason to believe our funds will be used wisely, without any restrictions aside from basic moral principles like honesty, reciprocity, and transparency.
This means BERI can spend small amounts of money on one-off tasks that are harder to pay for on research budgets within large institutions with budgets of billions, or even just millions, of dollars.
BERI as a social influence
In addition to boosting productivity, we hope that the inexpensive luxuries BERI provides will help make the difference, to some researchers, between feeling like a typical academic versus feeling a deep sense of support from one’s surrounding social context. Support from BERI means someone is going above and beyond the typical means of helping you, because they believe in what you’re trying to do.
So far, BERI has been funded by individually-sourced donations and grants, since its budget has been small enough not to warrant a public fundraiser. We may need to hold public funderaisers in the future, however.
Why does BERI do things slowly for the first time?
Requests to BERI are subject to a selection pressure called adverse selection. BERI gets asked to do something specifically when no other organization is doing it yet, and there might be a good reason why no one is doing that thing, often legal or ethical reasons. So, requests to BERI are unusually likely to be legally or ethically tricky. Indeed, numerous projects we’ve been asked to take on have turned out to be legally difficult in unexpected ways, usually in ways that the person or organization asking did not anticipate. In a sense, by opening ourselves up to looking for neglected interventions for x-risk, BERI is always risking “scraping the bottom of the barrel” for things that aren’t really a good idea to do.
For these reasons, BERI puts extra due diligence into doing something for the first time. Staff discuss novel plans with each other and conduct expert consultations to try to think of ways a new activity could lead to problems for BERI or the x-risk ecosystem. Over time, we accumulate new precedents for doing new things, which therefore become easier to do the second time, and we try to build systems to keep doing those things more quickly and easily in the future. But, the first time is usually slow. We’re okay with that, because it’s the best way we know how to sustainably keep doing tricky new things. There will always be tricky new things that need doing, and given the number of times we expect to do something new for the first time, BERI wants its staff and organizational structure to develop the meta-skill of being able to do new things correctly on the first try. So, please be patient if you ask BERI to do something, and the answer is “We’ve never done this before, but we can start looking into it!”
How does BERI hope to grow and develop?
As we gain experience through collaborating with others in the x-risk ecosystem, we hope to expand our vetting criteria to include more researchers and collaborators in the future. In our most optimistic vision of BERI’s future, we become a “yes we can” organization that can swiftly take on any and all mission-relevant project ideas that come our way.
How does BERI handle IP?
When collaborating with other public charities, such as universities, BERI concedes any intellectual property rights to our collaborators’ institutions. In research collaborations, we choose our partners for both their integrity and the importance of their work, and as such we defer all intellectual property decisions to them. This makes it easier for them to decide to collaborate with us, and benefit from our resources.
Why doesn’t BERI operate inside a larger non-profit?
BERI has historical ties to MIRI and CFAR, both of which are non-profits whose missions are aimed at mitigating x-risk, and collaborates closely with the newly formed Center for Human Compatible AI at UC Berkeley. By operating as an independent legal entity with its own relationships to donors, advisors, and supporters, BERI is able to make decisions more quickly, and under a broader mission statement.
The city of Berkeley has historically been home to many forward-thinking movements and cultures, and thus seems like an ideal place to nurture careful and responsible thinking about the future. Thus, BERI currently views its role in the world as a local stimulus for global thinking. That is, while our mission is to help the world at large, our best opportunities for impact involve fostering a local community of experts and doers in and around the city of Berkeley and the Bay Area more broadly.
What is BERI doing about other x-risks besides AI?
Right now, BERI is focused on x-risk work relating to AI, because that is our main comparative advantage. While BERI is able understand the existence of other existential risks besides AI, current BERI personnel do not have the expertise to evaluate effective interventions to reduce those risks. For example, if given money to reduce existential risks from synthetic biology, we would not currently feel comfortable administering those funds. For now, BERI is treating AI x-risk reduction as its “first customer”, because we feel competent to address it.
In the long run, BERI’s mission statement leaves open the possibility that we might expand our activities in any way the organization’s leadership considers the best use of its resources to reduce x-risk. If that occurs, we aim to be transparent about the change in scope of our activities, in keeping with our above concerns about the accuracy of public knowledge about the total work going into x-risk reduction.
How to contact us
If you wish to leave us a message, you can email firstname.lastname@example.org. However, note that BERI is not currently engaged in much public communication, as we are currently focused on developing our own operations, and supporting researchers who have already been nominated to us by our affiliates. As such, you should expect a very slow response if you try to contact us.