AI Impacts Wiki

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== Resisted Technological Temptation: Geoengineering ======

//Published 30 March, 2023. Last updated 30 March, 2023.//

Stratospheric aerosol injection could stop or reverse global warming with direct costs of \$1-10 billion per year, compared to global benefits of roughly \$1-10 trillion per year by mid-century and at least tens of billions of dollars per year for large countries.

Large countries do not implement geoengineering because not enough research has been done to know that the risks are small. Geoengineering research is prevented by a majority of climate scientists' opposition, which makes it harder to get funding, and by certain environmental groups that advocate against specific experiments.

===== Details =====

This case study is part of the [[responses_to_ai:technological_inevitability:incentivized_technologies_not_pursued:resisted_technological_temptations_project|resisted technological temptations project]]. The goal of this project is to understand situations where some actor might have expected to capture substantial value from pursuing a technology, but did not as a result of concerns about downsides that would not directly affect that actor. 

==== Background ====

Geoengineering is "the deliberate large-scale manipulation of the planetary environment to counteract anthropogenic climate change."((Royal Society. //Geoengineering the Climate: Science, Governance and Uncertainty.// (London, 2009) p. ix. [[https://royalsociety.org/-/media/Royal_Society_Content/policy/publications/2009/8693.pdf]].)) This page will focus specifically on the most commonly discussed type of geoengineering: injecting aerosols into the stratosphere to reflect sunlight.

The benefits of geoengineering are avoiding the damages and risks of global warming, including sea level rise, ice melting, some extreme weather events, changes to crop productivity, some biodiversity loss, and potentially crossing climate tipping points. Risks from geoengineering include small changes in the color of the sky, effects on plant growth and solar power, destruction of ozone for some aerosols, possible shifts in rainfall patterns, getting locked in to using a technology for the foreseeable future, and unknown unknowns. These risks are much smaller than the risks of allowing climate change to continue. A sufficiently aggressive transition away from fossil fuels would also prevent the risks of climate change, but would cost significantly more than geoengineering.

This page is accompanied by [[http://aiimpacts.org/wp-content/uploads/2023/04/Why-Has-Geoengineering-Been-Rejected.pdf|a longer report]], which includes more details and justification for everything described here.

==== Considerations for Particular Actors ====

Implementing geoengineering would directly cost \$1-10 billion per year.((IPCC Special Report. //Global Warming of 1.5°C.// (2018). Ch 4.3.8.2: Economic and technological feasibility. p. 348-349. [[https://www.ipcc.ch/sr15/chapter/chapter-4/]].)) This might be a net benefit for some actors:

  * **Global Organizations.** Most people who advocate for geoengineering seem to prefer that it be done by a broad consortium of countries. //Benefits:// The cost of geoengineering could be shared among more of the people who stand to benefit. Having many countries involved might reduce conflict over the appropriate amount of geoengineering. //Additional Barriers:// Building an international agreement to implement something controversial is a difficult coordination problem. International law surrounding geoengineering is currently a hodgepodge of treaties designed for other purposes, some of which seem to discourage and others encourage geoengineering. //Stakes:// Along with the direct costs, there are also globally distributed indirect costs, from reduced solar power production and possibly reduced crop growth, which likely cost about \$100 billion per year or less. The benefits would be most of the avoided cost of climate change, on the order of \$1-10 trillion per year by mid-century.

  * **Large Countries.** Large countries have fewer coordination problems than global organizations and have the resources to implement geoengineering. //Benefits:// Some large countries currently spend tens of billions of dollars per year on climate change prevention and mitigation and estimate the net gains of these policies to be tens of billions of dollars per year more.((For example, the UK spends about £15-18 billion per year to prevent climate change and estimates the cost of climate change to its economy as £55 billion by mid-century. \\ //Climate Change Act 2008 Impact Assessment.// The National Archives. (2009) p. 5. [[https://webarchive.nationalarchives.gov.uk/ukgwa/201ccccc30123162956/http://decc.gov.uk/assets/decc/85_20090310164124_e_@@_climatechangeactia.pdf]].
)) //Stakes:// The net benefits a large country would receive by implementing geoengineering in addition to or instead of its current climate change prevention policies would be on the order of \$10 billion per year. This group seems to be the most likely to pursue geoengineering.

  * **Coastal Cities.** Coastal cities are at a high risk of flooding due to sea level rise. //Benefits:// The flood losses of some individual coastal cities would be greater than the cost of geoengineering if no other mitigation were done. //Alternatives:// For an individual city, building levees and dikes costs a similar amount or less than geoengineering. //Stakes:// The net benefits would likely be less than the net benefits of an alternative.

  * **Private Oil Companies.** The annual profit of the largest oil companies is \$10-25 billion. //Benefits:// Their market might be eliminated by other efforts to prevent climate change. To roughly estimate the scale of the value of avoiding these costs, we can look at what they are willing to pay to avoid them. Some oil companies have been willing to spend tens of millions of dollars per year on climate denial (not enough for geoengineering) and some have been willing to spend a few billion dollars per year on green investments (maybe enough for geoengineering). //Additional Barriers:// Oil companies might face backlash if they started doing geoengineering and lose access to some of their important markets. //Stakes:// The net benefits seem marginal, so it is not surprising that other factors dominate.

  * **Individuals.** The largest private fortunes are \$100-250 billion. There does not seem to be a clear incentive for an individual to pursue geoengineering. //Stakes:// The net benefit is likely negative.

==== Anti-Geoengineering Efforts ====

The efforts against geoengineering have operated on three levels: in published papers, in outdoor experiments, and in implementation. Each level precedes and precludes the later levels.

Prior to 2006, there was a taboo against publishing scientific papers about geoengineering. Crutzen, an editor of the journal //Climate Change//, broke this taboo and people began to consider geoengineering to be a legitimate object of scientific inquiry.((Crutzen. //Albedo Enhancement by Stratospheric Sulfur Injections: A Contribution to Resolve a Policy Dilemma?// Climate Change **77**. (2006) p. 211-219. [[https://link.springer.com/content/pdf/10.1007%252Fs10584-006-9101-y.pdf%20%20]].)) Prior to 2006, only 1-2 papers on geoengineering were being published per year, while by 2009, over 50 papers were being published per year.((Oldham et al. //Mapping the Landscape of Climate Engineering.// Philosophical Transactions of the Royal Society A. (2014) [[https://royalsocietypublishing.org/doi/10.1098/rsta.2014.0065#d3e1677]].))

Currently, no outdoor experiments for stratospheric aerosol injection are being done. Since 2010, there have been two attempts at doing these outdoor experiments, but neither experiment happened. There seems to be two main causes:
  * Most climate scientists oppose geoengineering.((Only 31% of climate scientists support research on geoengineering technologies. \\ Dannenberg & Zitzelsberger. //Climate experts' views on geoengineering depend on their beliefs about climate change impacts.// Nature Climate Change **9**. (2019) p. 769-775. [[https://www.nature.com/articles/s41558-019-0564-z]]. \\
The body of the paper aggregates the beliefs of scientists at the IPCC and negotiators at the Paris Climate Accords. The disaggregated information comes from Supplementary Table 2(c):
[[https://static-content.springer.com/esm/art%3A10.1038%2Fs41558-019-0564-z/MediaObjects/41558_2019_564_MOESM1_ESM.pdf]].)) They write the IPCC reports which do not recommend geoengineering experiments. Government funding agencies, which seem to use recommendations from the IPCC reports and local climate scientists, rarely fund geoengineering research.
  * When geoengineering experiments are proposed, certain environmental groups((Especially the ETC Group and the Friends of the Earth.)) rally to oppose them. These environmental groups work with indigenous groups, when possible, to convince advisory committees and governments to not allow the research to proceed.

No country or international consortium has attempted to implement geoengineering. Two of the most common arguments against it, found in both scientific and policy papers,((Huttunen et al. //Emerging policy perspectives on geoengineering: An international comparison.// The Anthropocene Review 2(I). (2015) p. 14-32. [[https://journals.sagepub.com/doi/pdf/10.1177/2053019614557958]].)) do not seem sufficient to completely explain this:
  * There is a 'moral hazard' that geoengineering might mean that many countries will not do anything else to prevent climate change. As more greenhouse gases are emitted, the amount of geoengineering required and the risks associated with it would increase. But current prevention efforts are not what you would expect from a rational actor, so it is not clear how implementing geoengineering would affect other climate change prevention policies.
  * Geoengineering requires global governance to keep it from causing conflicts between countries, and this governance structure has not been created yet. But large countries are often willing to pursue their own interest when international law is uncertain.
It seems as though the most effective argument against implementation currently is:
  * Not enough research has been done. This sort of major decision should only be made with a good scientific understanding of what will happen.

The efforts against geoengineering began with a taboo against discussing it in scientific papers. Although that taboo has been broken, there are still no outdoor geoengineering experiments. Large countries seem to be unwilling to pursue a technological fix for climate change without outdoor experiments that quantify its effectiveness and risks.

==== Comparison to AI Capabilities Research ====

The existence of other technological temptations, such as geoengineering, suggests that it is possible for human-level AI to be feasible and remain uncreated. A comparison between AI capabilities research and other technological temptations might provide insight into the conditions under which some valuable technologies are realized and others are not.

There are several important differences between geoengineering and AI:
  * There is a clear understanding of what it would take to do geoengineering.((Smith & Wagner. //Stratospheric aerosol injection tactics and costs in the first 15 years of deployment.// Environmental Research Letters **13**. (2018) p. 124001. [[https://iopscience.iop.org/article/10.1088/1748-9326/aae98d/pdf]].)) The path to human-level AI is much less clear.
  * Getting climate wrong may pose devastating risks, but it is less likely to pose an existential risk than getting AI wrong.((Toby Ord estimated the existential risk from climate this century to be 1/1000, and the existential risk from AI this century to be 1/10 in //The Precipice//.))
  * Most climate scientists seem to be more opposed to geoengineering research than most AI researchers are to research that causes progress towards human-level AI.

There are also some important similarities:
  * The potential global benefits are likely orders of magnitude greater than the direct costs.
  * An individual actor could implement something with global consequences.
  * Current governance is unclear or non-existent.
  * The cost of building AGI seems likely to be large enough to limit the number of actors, but not so large that it can only be pursued by the largest countries.
  * Technological progress and economic growth could make it easier for more actors to implement it, although aircraft prices are not falling as quickly as computer prices.

Here are a few key takeaways from this technological temptation:
  * Preventing research can prevent implementation.
  * A high status scientist was able to break a taboo on a certain area of research.
  * The scientific consensus impacts what funding is available.
  * Small groups of advocates can successfully block individual projects.

//Primary Author: Jeffrey Heninger.//

===== Notes =====