Primer partnered with Vibrant Data Labs, a nonprofit that uses data to create the first-ever comprehensive map of climate change funding. Our interview with Eric Berlow of Vibrant Data Labs shows how ‘following the money’ reveals both our biggest opportunities and threats to turn climate change around.

It’s no secret that the future of our planet is in trouble. The recently reported IPCC report concluded that countries aren’t doing nearly enough to protect against significant disasters to come. 

But in order to solve a big problem like climate change, and to understand if our current response is working, we need to see where private funding in the sector is going. What issues are getting money, and which organizations are getting that funding? What other trends might emerge? 

Applying NLP to climate change

That’s where natural language processing comes in. Using Primer’s NLP technology, we partnered with Eric Berlow at Vibrant Data Labs to produce the first-ever climate change funding map. Primer’s Engines analyzed data on over 12,000 companies and nonprofits funded in the last five years. Using organizations’ descriptions provided by Crunchbase, grant applications provided by Candid, and in partnership with the Cisco Foundation, we generated one of the first-ever data-driven hierarchies of climate topics to better understand our current response, alongside any potential gaps. Using this topic hierarchy, we can see what projects organizations are working on – and where. That helps us see what’s missing in the bigger picture. And to solve a problem like climate change, a big picture view is what’s needed.

Watch our interview with Eric Berlow on why following the funding is crucial for the climate’s future. 



“The Coronavirus pandemic was like the trailer to the climate change movie where if the have-nots are really impacted, everybody gets affected by it. Climate change is one of those problems that is an all-hands on deck problem. You cannot just solve a little piece of it. “ – Eric Berlow



Learn more about Primer’s partnership with Vibrant Data Labs here, and learn the technical piece behind the work here and here.

For more information about Primer and to access product demos, contact Primer here.

I have a strange new pandemic ritual. Every Monday, I gather my team of machine learning engineers in a video meeting to kick off the week. But before we get down to business, I use a giant neural network to generate questions that help us go deeper than small talk. We vote on our favorite question and then take turns answering it.

  • What would you do if you knew you could not fail?
  • If you were guaranteed that you would live forever, would you still be afraid of death?
  • What is the number one thing you wish you could change about how people see you?

Those are actual questions the model generated. They indeed lead to deep talk.

We’ve been working remotely at Primer since the earliest days of the pandemic. This ritual is one of our solutions to the problem of feeling disconnected to each other. I can tell you: It really works. 

Here’s how I do it. 

First, I create a prompt consisting of numbered questions. For example:

  1. What is the meaning of life?
  2. If you were the last person on Earth, who would you choose to join you?
  3. What animal would you be for a day?

You can find lists of such questions all over the internet, probably inspired originally by this 2015 New York Times article, 36 questions for a private conversation starter. I find that a list of at least 20 questions works best.

Then I feed this prompt to a big autoregressive language model. I use GPT-J-6B, the open source model created by Eleuther. I just copy-paste the prompt and run the model in Primer’s developer interface. The model’s inference takes about 5 seconds.

Photo credit: John Bohannon, generated with VQGAN+CLIP with the prompt “A beautiful rainbow in the style of van Gogh.”

What comes back is machine-written text that continues where mine left off:

4. What thing would you never let anyone see you do?
5. What color are your dreams?
6. If you could have one superpower, what would it be?

Yes, these neural networks know how to count. My team uses the numbers to vote for their favorite.

I admit it may seem bizarre to use a machine to help people move beyond small talk and connect. But this is artificial intelligence at its best: Working together with humans to help them do what they do best. 

I generated 365 Deep Talk questions so you can try it yourself.

How mapping bottom-up climate action can drive more strategic climate solution responses and help us adapt thoughtfully. 

At the recent COP26 climate summit, a Minister from the Pacific island country of Tuvalu announced that it would be seeking to retain legal status as a country even if its entire territory were to become submerged [Reuters]. He was standing thigh-deep in the ocean in an area that used to be dry land. His speech made it clear—the effects of climate change are here today.

When it comes to tackling the climate crisis, we typically think about solar power, electric vehicles, and carbon capture to mitigate future climate change. But Tuvalu’s story is the tip of the iceberg of climate adaptation — the messier, less-defined problem of how human civilization will respond to the changes that are here now and predicted to accelerate over the next 30 yrs — even if mitigation efforts are successful.

In order to help define this messy space, Primer recently partnered with Vibrant Data Labs, a social impact data science group, to make sense of this broader and more diverse climate landscape. Crunchbase and Candid provided data on over 12,000 companies and nonprofits funded in the past 5 years that are addressing climate-related topics. Primer’s natural language processing (NLP) engines mined these organizations’ descriptions to generate one of the first-ever, data-driven conceptual hierarchy of topics to better understand the shape of our current response, and its potential gaps. This unique perspective comes bottom-up from how the private and social sector organizations on the ground describe what they do — not by what is most spoken about in the news or social media.

Our analysis suggests that while new technologies are emerging to address climate mitigation, existing organizations that have historically tackled structural inequities (e.g, gender equity, migrant rights, homelessness) are uniquely poised to address climate adaptation challenges which permeate every aspect of civil society. Our sample showed these organizations are beginning to add a climate lens to their work on diverse social issues.

Defining the Climate Space

We created a hierarchy of interrelated topics based on the company descriptions. Using this hierarchy, we are able to surface the broad topics in climate work and also drill down into specifics. 

Examining the topics in this way revealed there are two major branches: one dealing with topics related to preventative technologies (Environment And Renewable Energy) and the other with topics addressing the human impact of change (Public And Social Benefits).  This computational technique led to a close split between mitigation and adaptation. It’s exciting that our method could organize these topics in a way that gets close to how a human would do the task.

The topics underneath our mitigation branch (Environment and Renewable Energy), are what one may expect: “Water”, “Nature and Conservation”, and “Energy, Storage, and Energy Waste”. Adaptation work is much more varied and therefore harder to define. Our analysis can help us paint a crisper image of this emerging landscape.

This topic hierarchy shows organizational distribution on climate change topics, with higher convergence at the top vs the bottom.

Climate Change as a Social Issue

The Intergovernmental Panel on Climate Change (IPCC) defines adaptation as “the process of adjustment to actual or expected climate and its effects” [IPCC]. Using this definition, we see some top level branches that are climate adaptation. As the earth warms and extreme weather becomes the new normal, Disaster Relief and Preparedness will be critical to serving areas affected.

A less obvious topic might be Community And Neighborhood Development. The subtopics within it seem like quite standard areas pertaining to social issues such as: Health, Affordable Housing, Human Rights, Government Advocacy, and Gender Equality. Looking deeper into the language of these organizations we can see how they are incorporating a climate lens to their work.  For example, here is the description of one of the organizations working in Gender Equality:

MADRE is an international women’s human rights organization founded 37 years ago.  We partner with grassroots women to create social change in contexts of war, disaster and injustice. To advance women’s human rights, MADRE supports grassroots women to meet their basic needs as a bridge to develop their advocacy and political participation….

…. Our actions to confront climate change today will decide the futures of our planet and of generations to come. You can join the women leading the way.    Climate change is a global threat, and poor, rural and Indigenous women are hardest hit. They are impacted first and worst by the food shortages, droughts, floods and diseases linked to this growing danger. But they are more than victims. They are sources of solutions, inventing innovative, locally-rooted responses.    Through our Women Climate Defenders initiative, MADRE’s grassroots partners enable their communities to adapt to climate change. They build clean water systems to guard against drought and seed banks to preserve future harvests.

This is an example of how an organization that has been addressing women’s human rights for 37 years can contribute today to building climate resilience in the most vulnerable communities. It also highlights how climate adaptation requires addressing diverse, interdependent topics.

We can dive deeper into the Gender Equality data to understand the key topics that organizations in this field are working on today. A quick glance at this chart shows a wide range and diversity of topics in the climate adaptation cohort, from Human Rights to Infrastructure to Youth Organizing and Antidiscrimination.

The topics which co-occur most frequently with gender equality cover a range of socially minded topics which are not all tightly related to gender equality.

Let’s compare it to a topic from our climate mitigation set, Nature Conservation and Environment

The topics which co-occur most frequently with nature conservation and environment are very conceptually similar and mostly related to climate mitigation.

Organizations in this cohort work on Sustainability, Renewables, Water Conservation, Sustainable Agriculture, and Wildlife Preservation. It seems that most of these issues are more proximate to each other.

To further peel back the layers on current climate solutions, let’s take a deeper look at “crowding” or “spreading” of focus areas by organization. With NLP, we can approximately measure organizational “topic coherence” which tells us if a given organization optimizes on breadth or depth, and exactly how far apart the topics are within that cohort. We created a score from 0 to 1 that calculates how similar an organization’s topics are to each other—we call this the “organization focus score”. Organizations that focus on a narrow set of topics will have scores closer to 0. We can then extrapolate to the topic level to measure how narrowly focused the organizations in each topic are. When we plot this out from 0 to 1, we see topics relating to climate adaptation (Public and Social Benefit) are being addressed by organizations that are more broadly focused than the organizations addressing climate mitigation (Environment and Renewable Energy) topics.

The topic coherence score measures how closely related the topics a given organization works in are to each other. A score of 0 closer to 0 indicates the topics are very similar and a score closer to 1 means they are all very dissimilar. Climate adaptation topics (Public and Social Benefit) contain organizations with a more diverse set of focus areas than the climate mitigation topics.

Our analysis reveals that, while an organization working on mitigation will typically be working on a single, defined solution, organizations working on climate adaptation are fighting on multiple fronts.  


“In an interconnected world, it is exactly this messiness that funders need to embrace”, says Vibrant Data Labs’s Eric Berlow. “Traditional venture capital tends to fund focused, scalable solutions, with easy-to-measure outcomes, like renewable energy. But the climate crisis is an ‘all hands on deck problem’. Winning on one corner of the problem is an important piece; but if structural and systemic inequities in climate adaptation are not addressed, like the people of Tuvalu above, we all lose. We all feel the climate impacts of supply chain shocks, forced migration, civil unrest, and war. The most recent IPCC report suggests these trends will exacerbate over the next 30 years even if renewables and carbon capture solutions are successful. Climate funders will have to adopt a more holistic multi-pronged approach to rise to this challenge.”

Conclusion

As climate change becomes more and more a central part of our lives, understanding the landscape of solutions and providers gives us perspective on the magnitude of the space. We used NLP to analyze the work of over 12,000 companies to better understand where private and public organizations were focusing their efforts. In doing so, we highlighted the broad set of topics that are climate related and illustrated that many organizations working across the diverse social sector are now adding climate solutions to their efforts to enhance equity and resilience.

In a coming post, we will present our partner, Vibrant Data Labs’ story in which they take this analysis a step further to highlight the solution areas that are receiving the most funding.

Natural language processing (NLP) automates the work that would previously have required hundreds of researchers. These machines read and write – and they’re changing the future of business.

With NLP, we can now analyze information at machine speed, but with human-level precision. As reading and writing is automated through machine learning, companies are freed up to focus on the challenges that are uniquely human.

Read more here