How to Stop Climate Change — by ChatGPT

Three months into my layoff, I’m finally getting interviews and feeling some breathing room to explore other things I’ve wanted to do with my time.

A few lessons into the IBM GenAI Engineering course, I learned something I hadn’t considered: AI can perform Root Cause Analysis.

There’s only one thing I really want to get to the root of:

How did we go so wrong in rising to the challenge of slowing climate change?

And is there anything left we can do about it?

Surprisingly, ChatGPT had some answers that seem hope-inducingly effective, even in our polarized political climate. I only had to ask.

Of course, climate change work is and has always been an all-hands-on deck effort. What tickles me about ChatGPT’s ideas is they sound fun — and I can imagine broad swaths of Americans from different backgrounds enjoying getting involved as well.

So, listen up Climate nonprofits, concerned citizens, former boomtowns, churchgoers, CSR officers, and elected officials. There’s something here for all of us.

How AI would save the planet

And offset all its power-hungry data centers

Below, I’ll post my prompt and ChatGPT’s plan for reversing our national momentum on climate inertia (consolidated of course from the excellent research and ideas from many groups, all attributed).

But first, here are a few highlights I found particularly insightful:

1. Depolarize the Message: Rather than “existential threat,” focus on a unifying narrative of responsibility, ingenuity, stewardship, and disaster resilience.
2. “Pre-Bunking.” Teach in schools & public forums how to spot fake experts & cherry-picked data — using games and quizzes to build “mental antibodies.”
3. Robust Energy-Town Support. Where mines and refineries close, provide job guarantees, wage insurance, pension protection, and local-tax backfills tied to new industrial projects (transmission build-outs, well-plugging, etc).
4. Energy Benefits Advisors. Across utilities & governments, provide pre-filled applications and support for rebates, efficiency upgrades, and energy assistance.
5. Fun Projects: Church- or school-led “comfort and cost” drives (HVAC filters, shade trees, cool roofs), volunteer “methane-leak detection & fix” days.
6. Smart Communicaitons. Event-linked attribution briefs via local news within 48 hours of extremes; “Your Home, $X Saved” mailers, pastor talking points.

WHAT I LIKE ABOUT THIS APPROACH

Having started and run my own climate-action nonprofit for three years back in 2007, I’ve watched with concern as climate momentum has shifted behind the scenes mainly to the energy/tech sectors, while the politically intractable national conversation around climate change (and hope for collective action) has seemed largely abandoned.

I like that ChatGPT’s climate ideas engage people in politically neutral, yet uplifting, activities. We don’t have to say we’re “fighting climate change” to do the right thing. We can just work to make our communities’ health, comfort, and safety better.

The full conversation

For anyone with the bandwidth to implement some of these climate-supportive ideas, please see below for the full plan.

As mentioned, AI of course pulled this either in full or in part from the work of many excellent nonprofits, academic institutions, NGOs, and think tanks around the world. But I like what AI did with it — amalgamating all this work into a feasible, somewhat inspiring plan.

And if you ARE working on this and need help — let me know! I might be available to get involved: [email protected].

What I asked ChatGPT

Prompt Engineering for Social Good?

I just learned you can perform Root Cause Analysis. I would like you to diagnose/perform an RCA on why about half the people in the United States do not believe in Climate Change and/or do not consider it an existential threat for all humanity.

Based on the results of your RCA, I would like you to propose effective interventions for this point in time (2025) that can effectively undo the prevalence of this misinformation and bring people to collectively taking responsibility for and taking necessary steps to stop climate change globally.

(Or join my conversation with ChatGPT at https://chatgpt.com/share/68ffe14b-16ac-8013-a0f7-bcf8743fdc69)

What ChatGPT returned

Encouraging as always. But also kind of inspired.

Awesome brief. Here’s a practical, evidence-based Root Cause Analysis (RCA) of U.S. climate disbelief/low-salience — and a matched intervention plan for 2025 that aims to move attitudes and behavior at scale.

* Problem statement (2025 snapshot)

“Disbelief”: Roughly 15–16% of U.S. adults say global warming isn’t happening. (The Guardian)

“Low perceived threat”: About half of Americans do not see global warming as a serious threat to them personally, and concern ranks below other environmental issues. (Gallup.com)

Salience is highly partisan, with Republicans far less likely to prioritize climate or support many climate policies. (Pew Research Center)

* Root cause analysis (RCA)

R1. Partisan identity & motivated reasoning
Climate views are now strong signals of political identity; group loyalty and “identity-protective cognition” shape information acceptance, independent of factual knowledge. (SSRN)

R2. “Solution aversion”
Skepticism rises when proposed remedies (regulation, taxes, mandates) conflict with a person’s ideological values or interests. This is experimentally demonstrated. (PubMed)

R3. Information ecosystems & organized doubt
Decades of coordinated messaging (think tanks, industry allies, media) seeded uncertainty about the science and amplified economic-harm frames. (Environmental Defense Fund)

R4. Psychological distance & risk perception
If climate harms feel distant (in time, place, or to “other people”), urgency stays low. This effect is well-documented in the risk-perception literature. (Frontiers)

R5. Value worldviews & religion
Highly religious Americans — especially White Evangelicals — report lower concern; younger Evangelicals show more concern than older cohorts, signaling a generational shift. (Pew Research Center)

R6. Local economic identity
Communities tied to fossil-fuel economies fear loss; perceived economic threats depress support for climate action even when general climate concern exists. (CGEP)

R7. Policy backlash framing
Rules framed as “mandates” (e.g., future EV sales targets) trigger opposition — even among people who accept warming — because they cue identity and cost concerns. (Pew Research Center)

R8. Mixed signals about immediacy
While more Americans now report climate-linked extreme weather, the connection isn’t always communicated with clear, credible attribution. (The Guardian)

* Strategic goal

Shrink “low-threat/low-priority” segments, harden resistance to misinformation, and convert passive agreement into visible, collective action — without deepening polarization.

* Interventions (what to do in 2025)

I. Reframe from ideology to immediate, shared benefits

1) Health-first framing (smoke, heat, allergies, heart-lung disease) with local data; pair every risk with a concrete efficacy step (filters, cooling centers, tree canopy plans). Health frames test strongly across parties. Track shifts in “very/somewhat worried” and “already harming people” in YPCCC county-level estimates. (Yale Climate Communication)

2) Pocketbook framing over planetary doom: emphasize lower bills, reliability, and American manufacturing/jobs from upgrades (heat pumps, insulation, transmission). Use bipartisan-supported policy elements (methane leak sealing, home efficiency tax credits) as gateways. (Resources for the Future)

II. Reduce solution aversion

3) Choice architecture over mandates: spotlight options (rebates, credits, co-ops), not top-down requirements. Where mandates exist, pair with flexible compliance paths and highly visible consumer benefits. (Pew Research Center)

4) Conservative-consistent messengers: veterans, farmers, linemen, small-town doctors, pastors, insurance adjusters. Message: responsibility, ingenuity, stewardship, disaster resilience — not partisanship. Evidence shows messenger fit beats message content when identity is salient. (SSRN)

III. Inoculate against misinformation (pre-bunking)

5) Pre-bunk common deception techniques (fake experts, cherry-picking, “it’s natural cycles,” “too expensive”) before audiences encounter them, using brief interactive content (games, quizzes) shown to build “mental antibodies.” Deploy across schools, local news, and platforms. (University of Bristol)

6) Rapid-response myth-busting that pairs myth + debunk + alternative explanation, and always ends on action(what to do now). Don’t repeat myths in headlines.

IV. Make climate proximate, not distant

7) Event-linked attribution briefs within 48 hours of extremes (floods, smoke, heat): short, plain-language “how much more likely/intense” plus “what local fixes help.” Coordinate with weathercasters and emergency managers. (The Guardian)

8) Neighborhood-level salience: use the Yale Climate Opinion Maps to target outreach where “worried” is low but “policy support” is surprisingly high — these are ripe for activation. (Yale Climate Communication)

V. Convert belief → participation

9) Visible, collective projects: church- or school-led “comfort and cost” drives (HVAC filters, shade trees, cool roofs), volunteer “methane-leak detection & fix” days with utilities, farm windbreaks. Publicly post dollars saved and risk reduced — not just CO₂ tons. (Bipartisan policy support is strongest where co-benefits are concrete.) (Resources for the Future)

10) IRA-era benefits navigators: one-stop help desks (libraries, clinics, credit unions) that pre-fill rebates/credits and schedule installations. Meet people at trusted institutions; measure conversion rates and bill savings.

VI. Engage religious communities without culture-war

11) Stewardship & care messaging crafted with clergy; highlight that most U.S. Christian leaders do accept human-caused warming, and lift up intra-faith exemplars (energy-saving retrofits, disaster mutual aid). (PNAS)

VII. Just transition for fossil communities

12) Place-based transition compacts: job guarantees, wage insurance, pension protection, and local-tax backfills tied to new industrial projects (transmission build-out, well-plugging, CCS where appropriate). Use Columbia’s fossil-community economic findings to target sequencing and supports. (CGEP)

VIII. Depolarize the narrative

13) Two-message track:

National frame: energy security, reliability, U.S. competitiveness.

Local frame: health, bills, safety from extreme weather.

Avoid “existential threat” as the only frame; pair with high-efficacy steps to reduce fatalism.

* Implementation blueprint (6–18 months)

Segment & aim: Use YPCCC maps to identify three county archetypes: (A) high worry/low action → activation, (B) moderate worry/high policy support → conversion, © low worry/high exposure (heat/flood) → salience building. (Yale Climate Communication)

Channels: Local TV weather, farm bureaus, chambers, faith networks, clinics, libraries, emergency-management lists; limit national political cues.

Cadence: Pre-bunks quarterly; attribution briefs after each major event; benefits-navigator pop-ups monthly.

Partners: Electric co-ops, builders, HVAC contractors, insurers, hospital systems, veteran & faith groups, county EMAs.

Products: 60-sec pre-bunk videos; “Your Home, $X Saved” mailers; templated pastor & coach remarks; storm-afterfact cards (“9% more rain, 40% more likely — here’s what helps”). (The Guardian)

* Measurement & learning agenda

Attitudes: Track county shifts in “very/somewhat worried,” “already harming people,” and “human-caused” via YPCCC and replicate local polling where active. (Yale Climate Communication)

Behavior: Rebates claimed, upgrades installed, utility-bill deltas, enrollment in tree-shade/cooling programs, heat-illness ER visits.

Resilience: Flood/heat losses vs. historical baselines; insurance claims; outage durations.

Misinformation resilience: Short pre/post tests of tactic recognition (e.g., “fake expert,” “cherry-picking”). (University of Bristol)

Policy acceptance: Support for efficiency tax credits, methane-leak rules, grid upgrades, CCS tax credits — issues that poll competitively across parties. (Resources for the Future)

Design: Use A/B and stepped-wedge RCTs for media frames, messengers, and benefit-navigator formats.

* Why this should work

It targets the real blockers (identity, solutions, distance) rather than re-arguing the physics. (SSRN)

It meets people where they are (health, cost, reliability) and uses messengers they already trust. (Resources for the Future)

It hardens against misinformation with pre-bunking, the approach with the strongest empirical support for lasting resistance. (University of Bristol)

It converts abstract concern into visible local wins, which in turn create social proof and durable norms.