Centre for the Study of Existential Risk Four Month Report October 2019 - January 2020

By HaydnBelfield @ 2020-04-08T13:28 (+8)

The Centre for the Study of Existential Risk (CSER) is an interdisciplinary research centre within the University of Cambridge dedicated to the study and mitigation of risks that could lead to civilizational collapse or human extinction. Our research focuses on Global Catastrophic Biological Risks, Extreme Risks and the Global Environment, Risks from Artificial Intelligence, and Managing Extreme Technological Risks. Our work is shaped around three main goals:

Our last Six Month Report was in September 2019. Since then, we have continued to advance existential risk research and grow the field. This report outlines our past and future plans and activities. Highlights of the last four months include:

We send short monthly updates in our newsletter – subscribe here.

Note this report was written in February, before many COVID-19 changes. We have attempted to indicate which events and recruitment has been postponed, but may not have caught every reference.

READ AS PDF

Contents

  1. Current and upcoming activities
    1. Recruitment
    2. Upcoming events
  2. Review of recent activity
    1. New staff
    2. Policy and industry engagement – Impact.
    3. Public engagement – Field-building
    4. Visiting scholars – Field-building
    5. Academic engagement – Field-building
    6. Publications.

1. Current and Upcoming Activities

1.1 Recruitment

We recruited up to three Research Associates and one Senior Research Associate, from any relevant field of expertise or experience, who can contribute to our existing strands of work, or who might lead the development of work in additional areas of, or approaches to, existential or global catastrophic risk. These positions are funded by a grant which runs through to August 2024.

We also plan to start recruitment for additional administrative support as soon as we are able.

Summer Visitors / Interns

Last summer we hosted four visiting researchers. This worked well, and we are exploring running a remote scheme this year. Outputs from last year’s visits included: Ross Gruetzemacher co-authored a paper with Shahar that he presented at the 2020 AIES conference, and a paper with Jess which he presented at EA Global London 2019, and contributed substantially to the development of Shahar’s AI scenario exercise; Amritha Jayanti wrote a report with Shahar on ‘accountability gaps’ and lethal autonomous weapons, and secured a position at Harvard’s Belfer Centre; Nathaniel Cooke contributed substantially to Luke Kemp’s ‘Rise and Fall’ database; and Siebe Rozendal wrote a useful report on Eight high-level uncertainties about global catastrophic and existential risk.

1.2 Upcoming Events

2. Review of recent activity

2.1 New staff

We have recruited for three positions for our new Templeton World Charity Foundation project A Science of Global Risk. Two of the new staff have started over the last couple of weeks, and the third will join us in early March.

Lara Mani - Postdoctoral Research Associate, Communication and Outreach for a Science of Global Risk – started January 2020

Lara will work towards building an empirical evidence base for a variety of outreach and communication techniques adopted to present global risk – and to understand how an improved knowledge of global risk can translate to action. Her PhD in Geoscience (University of Plymouth) involved developing video games for communicating volcano risk in the Eastern Caribbean, from which she has developed a strong interest and pioneering expertise in the evaluation of communications strategies and techniques.

Catherine Richards - Part-time Research Assistant for Lord Martin Rees – started January 2020

Catherine has an engineering and business background in the energy and water sectors, and is completing a PhD(Eng) as a John Monash Scholar at the University of Cambridge where she is working on Global Systems Engineering. She was selected as one of 30 women under 30 to watch by Forbes magazine. As well as contributing her own research experience and outputs to CSER’s ongoing work Catherine will be supporting the Science of Global Risk programme and will be helping Martin Rees to leverage his expertise and network to better support CSER and our work.

Clarissa Rios Rojas - Postdoctoral Research Associate, Public Policy for Global Risk – startec March 2020

Clarissa is an Eisenhower Fellow and the Founder and Director of the Ekpa’palek NGO. She has worked at the Geneva Center for Security Policy, contributed to projects at the UN Biological Weapons Convention and UN Office for Disarmament Affairs, and is a member of the Global Young Academy. Clarissa has a PhD in Developmental Molecular Biology (University of Queensland). She also contributed to our 2019 Bioengineering Horizon Scan. Clarissa will be working to help us identify and address the many challenges of making public policy for global risk, with a focus on policy co-creation with key stakeholder groups in order to improve policy design and uptake.

Sean OhEigeartaigh and Jose Hernandez Orallo have also recruited a Postdoctoral Research Associated for their Paradigms of AI grant:

John Burden - Postdoctoral Research Associate, Paradigms of Artificial General Intelligence and their Associated Risks – starting July 2020

John Burden has a Bachelors’ and Masters’ degree in Computer Science from the University of Oxford, and is completing a PhD at the University of York. His current research focuses on reinforcement learning, and his research interests include learning in AI systems, AI safety, and generality.

The CSER team

2.2 Policy and industry engagement – Impact

We have had the opportunity to speak directly with policymakers and institutions across the world who are grappling with the difficult and novel challenge of how to unlock the socially beneficial aspects of new technologies while mitigating their risks. Through advice and discussions, we have the opportunity to reframe the policy debate and to hopefully shape the trajectory of these technologies themselves. Researchers continued their extensive and deep collaboration with industry. Extending our links improves our research by exposing us to the cutting edge of industrial R&D, and helps to nudge powerful companies towards more responsible practices.

2.3 Public engagement – Field-building

We’re able to reach far more people with our research:
12,202 website visitors over the last two months.
7,000+ newsletter subscribers.
8,542 Twitter followers.
2,701 Facebook followers.

Cover image from BBC Science Focus issue which quotes Laura Holt and Simon Beard

Promotional image for Professor Brian Cox’s interview with Professor Martin Rees which was broadcast by the BBC

2.4 Visiting scholars – Field-building

2.5 Academic engagement – Field-building

As an interdisciplinary research centre within Cambridge University, we seek to grow the academic field of existential risk research, so that it receives the rigorous and detailed attention it deserves.

Zia Mian delivering the Blavatnik Public Lecture ‘A Conceivable Horizon of Horror’ (30 Oct 2019)

Grethe Helene Evjen delivering the Blavatnik Public Lecture ‘Svalbard Global Seed Vault - Saving Seeds for Eternity’ (18 Nov 2019)

We continued our weekly ‘Work-in-Progress’ series:

The launch of the campaign for a Future Generations Bill, based on CSER research and enabled by the APPG for Future Generations CSER founded.

2.6 Publications

Peer-reviewed papers:

“The study of existential risk—the risk of human extinction or the collapse of human civilization—has only recently emerged as an integrated field of research, and yet an overwhelming volume of relevant research has already been published. To provide an evidence base for policy and risk analysis, this research should be systematically reviewed. In a systematic review, one of many time-consuming tasks is to read the titles and abstracts of research publications, to see if they meet the inclusion criteria. We show how this task can be shared between multiple people (using crowdsourcing) and partially automated (using machine learning), as methods of handling an overwhelming volume of research. We used these methods to create The Existential Risk Research Assessment (TERRA), which is a living bibliography of relevant publications that gets updated each month (www.x-risk.net). We present the results from the first ten months of TERRA, in which 10,001 abstracts were screened by 51 participants. Several challenges need to be met before these methods can be used in systematic reviews. However, we suggest that collaborative and cumulative methods such as these will need to be used in systematic reviews as the volume of research increases.”

“This paper examines and evaluates the range of methods that have been used to make quantified claims about the likelihood of Existential Hazards. In doing so, it draws on a comprehensive literature review of such claims that we present in an appendix. The paper uses an informal evaluative framework to consider the relative merits of these methods regarding their rigour, ability to handle uncertainty, accessibility for researchers with limited resources and utility for communication and policy purposes. We conclude that while there is no uniquely best way to quantify Existential Risk, different methods have their own merits and challenges, suggesting that some may be more suited to particular purposes than others. More importantly, however, we find that, in many cases, claims based on poor implementations of each method are still frequently invoked by the Existential Risk community, despite the existence of better ones. We call for a more critical approach to methodology and the use of quantified claims by people aiming to contribute research to the management of Existential Risk, and argue that a greater awareness of the diverse methods available to these researchers should form an important part of this.”

“On certain plausible views, if humanity were to unanimously decide to cause its own extinction, this would not be wrong, since there is no one whom this act would wrong. We argue this is incorrect. Causing human extinction would still wrong someone; namely, our forebears who sacrificed life, limb and livelihood for the good of posterity, and whose sacrifices would be made less morally worthwhile by this heinous act.”

“We model the Western Roman Empire from 500 BCE to 500 CE, aiming to understand the interdependent dynamics of army size, conquered territory and the production and debasement of coins within the empire. The relationships are represented through feedback relationships and modelled mathematically via a dynamical system, specified as a set of ordinary differential equations. We analyze the stability of a subsystem and determine that it is neutrally stable. Based on this, we find that to prevent decline, the optimal policy was to stop debasement and reduce the army size and territory during the rule of Marcus Aurelius. Given the nature of the stability of the system and the kind of policies necessary to prevent decline, we argue that a high degree of centralized control was necessary, in line with basic tenets of structural-demographic theory.”

“This chapter treats disaster response policies directed at the economic recovery of private households. First, we examine problems of disaster-induced financial distress from a legal and economic perspective. We do this both qualitatively and quantitatively, and focussing on residential loans, using the victims of the 11 March 2011 tsunami as our example. Then, using doctrinal and systematic analysis, we set out the broad array of law and policy solutions tackling disaster-induced debt launched by the Japanese Government. On this basis, we assess the strengths and weaknesses of these measures in terms of their practical adequacy to prevent and mitigate financial hardship and examine them against multiple dimensions of disaster justice. We conclude with suggestions for improving financial disaster recovery by taking a prospective approach, preventing the snowballing of disaster-related losses, which we argue represents a equitable and effective way forward in allocating resources following future mega disasters.”

“This paper explores the tension between openness and prudence in AI research, evident in two core principles of the Montréal Declaration for Responsible AI. While the AI community has strong norms around open sharing of research, concerns about the potential harms arising from misuse of research are growing, prompting some to consider whether the field of AI needs to reconsider publication norms. We discuss how different beliefs and values can lead to differing perspectives on how the AI community should manage this tension, and explore implications for what responsible publication norms in AI research might look like in practice.”

Reports:

“The international governance of global catastrophic risks (GCRs) is fragmented and insufficient. This report provides an overview of the international governance arrangements for eight     different GCR hazards and two drivers. We find that there are clusters of dedicated regulation and action, including in nuclear warfare, climate change and pandemics, biological and chemical warfare. Despite these concentrations of governance their effectiveness if often questionable. For others, such as catastrophic uses of AI, asteroid impacts, solar geoengineering, unknown risks, super-volcanic eruptions, inequality and many areas of ecological collapse, the legal landscape is littered more with gaps than effective policy.” 

“This report highlights how the body of emerging GCR research has failed to produce sufficient progress towards establishing a unified methodological framework for studying these risks. Key steps that will help to produce such a framework include moving away from a hazard-focused conception of risk, typified by the majority of quantitative risk assessments that we analyse, and toward a more sophisticated approach built on a mature understanding of risk assessment and disaster risk reduction and preparedness. We further suggest that a key barrier to the development of a mature science capable of comprehensively assessing the drivers of GCRs has been the political, philosophical, and economic context within which the field has arisen, as demonstrated by five distinct “waves” of GCR research. We believe that a suitably committed funder that transcends these contextual boundaries could have a transformative impact on the discipline, and with it our understanding of GCRs and their drivers. We propose that the Global Challenges Foundation is in a uniquely strong position to play this role.”

“Recently the concept of transformative AI (TAI) has begun to receive attention in the AI policy space. TAI is often framed as an alternative formulation to notions of strong AI (e.g. artificial general intelligence or superintelligence) and reflects increasing consensus that advanced AI which does not fit these definitions may nonetheless have extreme and long-lasting impacts on society. However, the term TAI is poorly defined and often used ambiguously. Some use the notion of TAI to describe levels of societal transformation associated with previous 'general purpose technologies' (GPTs) such as electricity or the internal combustion engine. Others use the term to refer to more drastic levels of transformation comparable to the agricultural or industrial revolutions. The notion has also been used much more loosely, with some implying that current AI systems are already having a transformative impact on society. This paper unpacks and analyses the notion of TAI, proposing a distinction between TAI and radically transformative AI (RTAI), roughly corresponding to societal change on the level of the agricultural or industrial revolutions. We describe some relevant dimensions associated with each and discuss what kinds of advances in capabilities they might require. We further consider the relationship between TAI and RTAI and whether we should necessarily expect a period of TAI to precede the emergence of RTAI. This analysis is important as it can help guide discussions among AI policy researchers about how to allocate resources towards mitigating the most extreme impacts of AI and it can bring attention to negative TAI scenarios that are currently neglected.”

“Technological advances are bringing new light to privacy issues and changing the reasons for why privacy is important. These advances have changed not only the kind of personal data that is available to be collected, but also how that personal data can be used by those who have access to it. We are particularly concerned with how information about personal attributes inferred from collected data (such as online behaviour), can be used to tailor messages and services to specific individuals or groups. This kind of ‘personalised targeting’ has the potential to influence individuals’ perceptions, attitudes, and choices in unprecedented ways. In this paper, we argue that because it is becoming easier for companies to use collected data for influence, threats to privacy are increasingly also threats to personal autonomy—an individual’s ability to reflect on and decide freely about their values, actions, and behaviour, and to act on those choices. While increasing attention is directed to the ethics of how personal data is collected, we make the case that a new ethics of privacy needs to also think more rigorously about how personal data may be used, and its potential impact on personal autonomy.”

“The second of the Centre for the Study of Existential Risk’s international conferences provided a timely opportunity for the Centre, along with the wide communities working on existential and global catastrophic risks and in related fields, to reflect on our work so far and to deepen and broaden our learning from other disciplines. This allowed us to both focus on some of the practical challenges of the task that we have set ourselves and identify what we are doing well. Hence, the Conference served not only to address important issues facing the existential risk research community, but also to establish and maintain the connections with other communities that have important contributions to make to the developing ‘science’ of existential risk research.”

“Countries can begin to close the production gap by aligning their energy and climate plans. Governments have a range of policy options to regulate fossil fuel supply, including limits on new exploration and extraction and removal of subsidies for production. Some countries are already demonstrating leadership: Belize, Costa Rica, Denmark, France, and New Zealand have all enacted partial or total bans on oil and gas exploration and extraction. Germany and Spain are phasing out coal extraction. Non-state actors and subnational governments can also help facilitate a transition away from fossil fuels, by mobilizing constituencies and shifting investment to lower carbon options. Individuals and institutions have already pledged to divest over USD 11 trillion from fossil fuel holdings. Several governments are planning for a “just transition” that aims to minimize disruption for affected workers and communities.”

The CSER team pictured in Summer 2019.