COVID-19 response as XRisk intervention

post by polyphony (tyleralterman) · 2020-04-10T06:22:19.969Z · LW · GW · 3 comments

Contents

  Training Ourselves
  Forging alliances
  Establishing credibility
  Growing the global risk movement
  Creating XRisk infrastructure
None
3 comments

Thank you to the reviewers of this post: Seán Ó hÉigeartaigh, Robin Hanson, Anders Sandberg, Allison Duettmann, Andre Ornish, Jessica Taylor, and Blake Borgeson.

---------

Effective Altruists and rationalists are some of the only people in the world tracking existential risks (XRisks). And so it’s understandable that some of us would ask: “Is this a distraction?” Surely a crisis that ends many lives pales in comparison to crises that could end all of them.

However, this line of reasoning might make long-termists lose sight of a large opportunity: we can build XRisk prevention capacity *through* pandemic response efforts. Through fighting COVID-19, we can:

1. Train ourselves

2. Forge alliances

3. Establish credibility

4. Grow the global risk movement

5. Create XRisk infrastructure

I’ll go through these by category.

Training Ourselves

Now that we’re dealing with a live global catastrophe, there’s a chance to develop our skills and ability to coordinate for future global catastophes.

Here are some particularly relevant areas for skill growth:

Forecasting: Create predictions of what might happen and be proven right or wrong.

Scenario-planning: Run inside-view and outside-view simulations of your proposed COVID-19 intervention, act accordingly, and then improve your simulation abilities based on what actually happened.

Coordination: Pandemic response efforts will often require that you collaborate with a wide variety of different actors, just like with other global risks.

Persuasive argumentation: This skill can be built through convincing others that your project is worth supporting, or through influencing decision-makers.

Networking: Navigate the complicated world of crisis-response to find others to aid and work with. Your project might require you to form connections with epidemiologists, with front-line responders, with public officials, with manufacturing plants, with fellow activists, and more.

Project management: Develop the ability to set and meet deadlines at a moment when they really matter. Learn how to lead other people. Find and use the best tools for tracking tasks. Document and curate best practices (this can help other projects as well).

Perhaps the greatest opportunity here is for EAs and rationalists to improve their existing comparative advantage: applied epistemology. A live crisis affords epistemic growth opportunities unmatched by many other contexts. You need to work with spotty information and a rapidly unfolding situation. You must process signal from unreliable sources. You need to make decisions over uncertainty.

This is an opportunity to:

• Get rapid feedback on your decisions and predictions.

• Ground abstract models in concrete experiences.

• Understand how various canonical tools fare in a real-life situation (back-of-envelope calculations, conducting meta-analyses on RCTs, doing reference-class forecasting)

In other words, it’s an opportunity to practice rationality with immediate practical upside.

Forging alliances

Just about every sector of society is mobilizing itself to fight the pandemic right now. This gives EAs and rationalists an “in” to form alliances that can be useful for preventing future global risks. In just a few weeks of working on the crisis, my extended network has come to include crisis-response experts, epidemiologists, philanthropic foundations, lobbyists, manufacturing firms, and intelligence community members.

Let’s say you care about AI safety in particular. Some alliances you might forge now include:

• Policy experts at the White House’s Office of Science and Technology Policy (OSTP). OSTP is now working with AI teams at Microsoft and Google to analyze thousands of scientific papers on COVID-19. Many policy experts are stretched thin and would value additional research support. And in one notable case I’m unable to post publicly, EAs are already offering it!

• Researchers across various AI groups like Deepmind that are working on AI-based pandemic-response initiatives.

• Competent volunteers you meet through activist efforts. For instance, the best initiatives tend to have at least one solid project manager or ops person. Nearly every AI safety initiative benefits from having competent PMs and ops people, so this is a chance to source new talent.

Establishing credibility

You know what’s useful for convincing people that you can competently address a global crisis? Competently addressing a global crisis. Useful resources are given to those who have demonstrated the sorts of prescience and intelligence that EAs and rationalists can demonstrate right now. Credit is already being assigned to rationalists and Silicon Valley people whose public material demonstrated better tracking of the virus than most established institutions. One attempt I like in this direction is epidemicforecasting.org from a team at the Future of Humanity Institute.

Notable demonstrations of competence on particular problems can often grant more general credibility. E.g., as with Nassim Taleb’s prediction of the 2008 financial crisis and Nate Silver’s predictions during the 2008 election.

Conversely, the EA and rationalists communities stand to *lose* credibility for inaction during the pandemic. The optics of tepidly responding to a global crisis after being the community that always ranted about global crises are...not good optics.

Of course we could also lose credibility by getting in the way of trained professionals during this one. But I believe EAs are pretty good at being helpful instead of hinderful.

Growing the global risk movement

The category of “global catastrophe” is no longer just an abstract idea. It’s a tangibly experienced reality. This ought ideally make it much easier to get more resources devoted to GCRs and XRisks in the future.

It is easy to imagine useful movement-building efforts. For example, now would be a good time to volunteer as marketing staff for Toby Ord’s new book, The Precipice.

With this crisis, there may be the potential to cause larger culture-shifts. Historians speculate that the Enlightenment emerged from the Thirty Years War and that the Renaissance emerged from the Plague. This new plague offers an opportunity for shifting culture toward a new renaissance – ideally one which has long-termism as a central value.

Importantly, this may not happen automatically. It’s possible that we’ll miss the opportunity to draw the connection between COVID-19 and the broader category of global risk. It’s also very possible that the broader culture will fail to blame COVID-19 in part on short-termism, and miss out on valuing EA ideas around long-termism. Here, we must helpfully intervene.

Global institutions like the IMF and IBRD emerged from Breton Woods immediately after World War II. Toward the end of the pandemic, there is going to be a short window for policy proposals and new global institutions. Likewise, there will be a limited window where it’s possible to introduce narratives that change the broader culture. To meet these windows, we need to start policy and narrative work during the crisis.

Creating XRisk infrastructure

There are plenty of exciting ideas out there for creating long-term XRisk infrastructure out of COVID-19 but not very many people acting on them. Here are just a few:

• Improving prediction markets: Prediction markets like Metaculus could benefit from more quality forecasters, increased organizational capacity, and extended influence media (e.g., media coverage and credibility amongst policy-makers).

• Increasing community world-modeling capabilities: This might simply include stimulating more quality analysis via a crisis which requires understanding many relevant parts of the world. Or it might include the building and improvement of tools like Guesstimate.

• Creating prize competitions: One idea I’ve seen is creating a prestigious prize like the Nobel for outstanding contributions to global risk reduction. One could start awarding these during the current crisis and build it into an ongoing institution post-crisis. This could make it not only more acceptable, but also widely virtuous to make contributions to global risks such as future pandemics, nuclear war, and AI safety.


In summary, there is ample opportunity to treat COVID-19 as an XRisk capacity-building intervention. If you find yourself interested in acting on this opportunity, here are a few resources:

The Effective Altruism Coronavirus Discussion Facebook Group

Coronavirus Research Ideas for EAs [EA · GW] from Peter Hurford

A database of EA responses to COVID-19, organized by Michal Trzesimiech

The Resilient Socieities Initiative. We’re interested in hearing proposals from EAs and rationalists even if they’re at the idea stage.

3 comments

Comments sorted by top scores.

comment by Rohin Shah (rohinmshah) · 2020-04-11T17:33:20.075Z · LW(p) · GW(p)

Copied from the EA Forum [EA(p) · GW(p)]:

I can't tell whether you're arguing "some small subset of EAs/rationalists are in a great position to fight COVID-19 and they should do so" vs. "if an arbitrary EA/rationalist wants to fight COVID-19, they shouldn't worry that they are doing less because they aren't reducing x-risk" vs. "COVID-19 is such an opportunity for x-risk reduction that nearly all longtermists should be focusing on it now".

I agree with the first (in particular for people who work on forecasting / "meta" stuff), but not with the latter two. To the extent you're arguing for the latter two, I don't find the arguments very convincing, because they aren't comparing against counterfactuals. Taking each point in turn:

Training Ourselves

I agree that COVID-19 is particularly good for training the general bucket of forecasting / applied epistemology / scenario-planning.

However, for coordination, persuasive argumentation, networking, and project management, I don't see why COVID-19 is particularly better than other projects you could be working on. For example, I think I practiced all of those skills by organizing a local EA group; it also seems like ~any project that involves advocacy would likely require / train all of these skills.

Forging alliances

Presumably for most goals there are more direct ways to forge alliances than by working on COVID-19. E.g. you mentioned AI safety -- if I wanted to forge alliances with people at OSTP, I'd focus on current AI issues like interpretability and fairness.

Establishing credibility

I agree that this is important for the more "meta" parts of x-risk, such as forecasting. But for those of us who are working closer to the object level (e.g. technical AI safety, nuclear war, climate change), I don't really see how this is going to help establish credibility that's used in the future.

Growing the global risk movement

You talk about field-building here, which in fact seems like an important thing to be doing, but seems basically unrelated to the COVID-19 response. I'd guess that field-building has ~zero effect on how many people die from COVID-19 this year.

Creating XRisk infrastructure

Agreed that this is good.

Overall take: It does seem like anyone working on "meta" approaches to x-risk reduction probably should be thinking very seriously about how they can contribute to the COVID-19 response, but I'd guess that for most other longtermists the argument "it is just a distraction" is basically right.

I'd probably change my mind if I thought that these other longtermists could actually make a large impact on the COVID-19 response, but that seems quite unlikely to me.

comment by avturchin · 2020-04-10T12:53:12.177Z · LW(p) · GW(p)

Current pandemic is also a test run of a global catastrophe which demonstrates how humanity is ineffective in preparedness and responding, how it was deaf to warnings and how greed, ignorance and overconfidence prevented us to act effectively. If it were more dangerous pandemic, say, bird flu with 60 per cent mortality but transmitted by birds, we may be already devastated.

In some sense, it demonstrates a failure of all previous pandemic prevention and even x-risks prevention efforts. We didn't create anything tangible which we could use now, not even an adequate stockpile of masks or consensus about their efficiency.

comment by romeostevensit · 2020-04-11T04:56:48.173Z · LW(p) · GW(p)

Why aren't more models open source on github? Or are they and I just need to be on the right email lists?