The Intergovernmental Panel on Climate Change (IPCC) recently released its sixth assessment report (AR6). AR6 tells us that humans are causing climate change, that the problem is worsening and that confidence is growing that many negative ecological impacts are attributable to climate change. We also have growing confidence that low-probability, high-impact events are made more likely by increased warming; the relationship between greenhouse gas (GHG) emissions and warming is linear; and climate change is unlikely to be stabilized without achieving net-zero global emissions. Despite imperfect knowledge, policymakers must make risk-informed decisions in climate policy, especially pertaining to high-impact, low-probability climate impacts outlined in AR6.

But the AR6 also contains some good news: a 2.8 to 5.7 degrees Celsius increase in global temperature by the end of the century would require a significant increase in global emissions. A scenario with rising and then declining emissions has a warming range of 2.1 to 3.5 degrees Celsius, though a 1.5-degree Celsius warming target is affirmatively out of reach. The worst-case scenarios, such as a collapse of the Atlantic Meridional Overturning Circulation (AMOC), are not impossible, but to reach 4-5 degrees Celsius of warming requires a turn for the worse in global emission trajectories.

AR6 provided less “cautious uncertainty” than previous assessments, while flagging new evidence on areas with high uncertainty and high consequence. This is notable because uncertainty in climate science has made it difficult to translate accessible, accurate and useful information for the general public and policymakers. The issue is not so much the limitations of climate science, but rather how it is conveyed by scientists and interpreted and reframed by non-scientific audiences. Poor communication of uncertainty exacerbates polarization of climate opinions, including unsupported alarmism and unwarranted skepticism. Reviewing the communication of uncertainty in AR6 presents on opportunity to bolster society’s literacy of climate science.

Understanding Uncertainty in AR6

AR6 has caused a stir by putting to bed, once and for all, that there is any level of reasonable uncertainty whether observed warming impacts globally are the result of human or natural causes—anthropogenic climate change is now a matter of scientific fact, full stop.

It is unequivocal that human influence has warmed the atmosphere, ocean and land. Widespread rapid changes in the atmosphere, ocean, cryosphere and biosphere have occurred. (AR6, P. SPM-5)

This assertion does not mean that all observable impacts of climate change have a high level of certainty, which is what some politicians and advocates claim. Conversely, the misperception that scientists are claiming unequivocal certainty fuels skepticism that such reports are politicized, not evidence based. Translating differences in scientific uncertainty across climate impacts in accessible terms is imperative to making reports like AR6 comprehensible to various audiences, while minimizing unnecessary controversy.

AR6 uses specific nomenclature to characterize uncertainty: five levels of confidence qualifiers ranging from “very low” to “very high.” The following table depicts terms of assessed likelihood and the probability of their occurrence.

Qualifying Term Likelihood of Outcome Examples
Virtually certain 99-100% Sea level rise over 21st century.
Very likely 90-100% Atlantic Meridional Overturning Circulation weaking over 21st century.
Likely 66-100% Increased overland precipitation since 1950 caused by human factors.
Very unlikely 0-10% Global land and ocean sink turn into a source by 2100.
Extremely unlikely 0-5% Hot extremes over the past decade would have occurred without human influence.

TABLE 1. Select Terms for Assessed Likelihoods of Occurrence

Assessed likelihood, however, only provides half the information. The consequence of an event is the other half of the equation. Once accounting for that, sometimes the risk profile of a low-probability, high-impact effect is greater than that of a high-probability, low-impact effect.

Compared to previous IPCC assessments, AR6 has higher confidence in attributing recent and geographically granular effects to anthropogenic climate change. Science still cannot attribute a single event, such as a hurricane, to climate change. But there’s greater statistical confidence that the frequency and magnitude of observable phenomena, like recent and future extreme heat events, are exacerbated by human activity. Fortunately, these relatively more certain effects are relatable to the general population, which makes them ripe to help climate science communication.

Unfortunately, the most severe climate risks have high uncertainty and are unrelatable to the general population. These include potentially catastrophic effects that are far in the future, which are often described as “fat tail” events because their low probability but high impact stretches the tail of a risk assessment curve. Such disaster-type events are particularly susceptible to probability neglect. A potential catastrophe plays into the alarmism narrative, while its low probability nature can also exacerbate dismissive bias. This is problematic because tail events matter most; they often dominate economists’ calculations of the damages GHG emissions may cause.

AR6 presented new evidence on tail events. These include ice sheet collapse, abrupt ocean circulation changes, forest dieback and compounded extreme events (concurrent heatwaves, droughts, wildfire, floods). Of particular interest is that AR6 noted that the Atlantic Meridional Overturning Circulation is “very likely” to weaken this century, but scientists only have “low confidence” in the magnitude. Considering the consequences—a “very likely” abrupt shift in regional weather patterns and water cycle in Africa, Asia and Europe—the fact that AR6 saw “medium confidence” that an abrupt collapse will not occur before 2100 still leaves cause for concern. Although a collapse before 2100 is unlikely, the AMOC presents total risk that could far outweigh most other high probability effects.

Other tail events to watch include feedback loops, whereby climate processes dampen or amplify an initial perturbation. As climate science evolved the past two decades, prospects for negative feedback loops that would counteract warming have diminished. AR6 provided new information on positive feedback loops, such as finding it to be “very unlikely” that global land and ocean GHG sinks could turn into a source of emissions by 2100. But whether this probability is closer to 1 percent or 10 percent could have a major impact on the cumulative risk profile of climate change.

Communicating Uncertainty in Climate Science

Climate change is best framed as a global risk management exercise. Risk has two components: the likelihood an event will occur and the consequence of that event. Climate change presents a massive number of discrete cause-effect relationships, each with varying degrees of likelihood and consequence, and the scientific community’s understanding of these risks is constantly evolving. This makes dynamic risk communication vital.

Evolving social norms play a major role in influencing risk communication. Throughout the 2000s, the journalistic norm of presenting balanced reporting actually led to biased coverage of climate change. For example, a media story that quoted views consistent with the scientific consensus received comparable coverage to a contrarian viewpoint from a scientific outlier. This inaccurately suggested that the scientific community was divided on anthropogenic contributions to climate change. Now, we find ourselves in an era of contested expertise, which has deterred experts from openly communicating the uncertainty of their knowledge out of concern of their audience’s reaction.

Cognitive heuristics also influence the efficacy of climate risk communication. Risk attitudes often exhibit “probability neglect,” in which individuals focus solely on the outcome and disregard probability. When the probability of an event is low, the opposite can be true: individuals ignore a potential outcome because they effectively perceive the probability of occurrence as negligible. Such black-and-white interpretations miss the grey area that scientific evidence usually provides, often leading to polarized public attitudes on climate change causes and effects.

Humans are naturally oriented toward information that confirms existing opinions, which is problematic for adjusting attitudes with evolving scientific evidence. This “confirmation bias” can cement polarized public attitudes even further, irrespective of new reports like AR6. These biases are present at the individual level and are amplified by social groups. The latter has become particularly concerning in the age of social media, where low-credibility content spreads quickly and can result in the proliferation of misinformation.

This doubles-down on the importance of researchers and social influencers communicating climate science uncertainty clearly. This includes clarifying whether uncertainty relates to the determination of cause-effect relationships or the magnitude of the effect. It’s also important to convey why uncertainty exists, such as variations in natural climate phenomena, measurement shortcomings or limited knowledge of the underlying biophysical or social processes like emissions projections. Communication initiatives like the Yale Program on Climate Change Communication have tested a variety of social science and data visualization techniques to improve the accuracy of public opinion.

Opportunities Ahead

In light of the framing and rollout of AR6, it is important for stakeholders to focus on better uncertainty communication in the following ways:

  1. Improve communication of tail events. Framing climate issues as an insurance problem may improve relatability. Individuals are routinely exposed to low probability risks like auto accidents and home fires. Paying a premium to mitigate that risk is usually worthwhile, especially if the event would prove financially catastrophic. To maximize the welfare of current and future generations, we must think about what premium is appropriate to pay to reduce the likelihood of catastrophic climate effects.
  2. Policymakers must become accustomed to making decisions under uncertainty. Policymakers face uncertainty in numerous other decisions, such as preventing terrorism or funding disaster relief agencies before a disaster occurs. It is easy for policymakers to understand the risk of abrupt risks like earthquakes or nuclear war as a low-probability, high-consequence risk, but it is often a struggle to communicate to policymakers more distant, incremental risks such as climate change. Nevertheless, the risks are just as real, and the same risk-based calculus should be applied to environmental risk as well.
  3. Research and communication on climate change’s more relatable impacts. One of the lead authors on the regional assessment chapter of AR6 noted that the report’s greatest impact is to “bring things closer to home,” such as what climate change means for those living in the Midwest or tropical islands. Indeed, AR6 highlights that the scientific community has higher confidence in many near-term regional impacts, such as drought, fire and extreme precipitation in central and western North America. This may help calibrate public engagement tools, such as describing how familiar events like extreme weather are more likely to occur in a specific region. AR6 provides an interactive atlas tool for flexible spatial and temporal analyses that interpreters could use to translate science into takeaways for geographically specific audiences.
  4. Climate communication must enhance the audience’s trust in the messenger. This underscores the imperative of avoiding politicized communication that undermines public acceptance of the scientific community. For example, U.N. Secretary General Antonio Guterres said the document is “code red for humanity” and that “this report must sound a death knell for coal and fossil fuels before they destroy our planet.” Yet AR6 implies no such things, and this statement has already contributed to discrediting the IPCC and AR6 by influential climate skeptic groups who inaccurately interpret scientific uncertainty. Instead, IPCC studies like AR6 should be marketed for what they are: the most authoritative information to-date on climate change causes and effects that informs policymaker considerations. 


AR6 fine-tuned our relatively high confidence in the moderate impacts of climate change. But the biggest risks—both environmental and economic—lie in the low-probability, high-consequence tail effects. This underscores the imperative of communicating climate science uncertainty accurately and effectively.

We operate in a media and political environment that rewards polarization, and rarely has this been more apparent than in climate science. An objective assessment of AR6 is that its high-quality science has improved our understanding of climate change, increased our certainty as to its anthropogenic nature, and articulated our varying levels of confidence in various potential climate impacts.

Policymakers need to become accustomed to making decisions with imperfect knowledge. But just as other problems have uncertainty, climate change should be approached from a risk-based framing. AR6 demonstrates high risk from low probability climate impacts. Policymakers should prioritize these tail effects. Just as individuals pay a premium for car insurance in case of a severe accident, policymakers will need to discern the appropriate premium to avoid worst case scenarios in climate change.

Image credit: Mike Mareen

Featured Publications