Talk amongst yourselves. I’ll give you a topic. You just woke up and realized the nation is running late in putting needed climate protections in place. Discuss.
You’d have to have been hiding in a cave over the last 12 months not to have heard about the rapid rise of artificial intelligence (AI) and what engineers can now program our computers to do. Love it or hate it, the AI genies are out of the bottle and unlikely ever to go back in — making it imperative for the climate community to consider carefully what role these programs might play in efforts to decarbonize national and world economies.
The question is this: will the advent of the AI age help or hinder efforts at combatting Earth’s warming? Would you be surprised to learn possibly both? Permit me to discuss.
I started to take AI seriously when Blake Lemoine, a Google software engineer, suggested that the company’s AI LaMDA[i] program could become sentient. The statement led to his dismissal.
Call me jaded. In my experience, there’s an inverse relationship between a large corporation denying something and the truth of its existence. Think Exxon and its public denial that fossil fuels have anything to do with climate change.
Honestly, Lemoine’s conclusions creeped me out. Suddenly movies like I Robot, Ex Machina, the Matrix, and 2001: A Space Odyssey started playing in my head. I even started to look at my smartphone suspiciously, thinking Siri might be planning to turn me into her servant — as if that hadn’t already happened.
I’m sure I’m not alone in anthropomorphizing the machines of our modern world. It starts in small ways. We humanize computer programs by giving them names like HAL, Siri, and Alexa, along with soothing voices. We move on from there by trying to engage them in conversations.
Why do I feel compelled to thank Siri for telling me what the weather is outside or the lifetime batting average of Minnie Minoso — a hero of mine growing up in Chicago? Will she think less of me for forgetting to say it? Does my not saying it mean she’ll be more willing to answer future questions?
It’s absurd, I know. And, yet, I do it. What is there in the human spirit that seeks such a person-machine connection?
AI programs are neural networks based on a mathematical system of learning that’s accomplished by processing large amounts of data in much the same way as the human brain. A program learns to recognize a giraffe by pinpointing patterns in tens of thousands of photos to come up with a sort of platonic ideal of “giraffe-ness”.
A variety of technologies and applications employ neural networks. These include video games, computer vision, speech recognition, social network filtering, language translation, and medical diagnosis.
AI programs are capable of writing code as well as drawing illustrations. The images below were made using the text instruction “a painting of the Statue of Liberty in the style of Andy Warhol.” The picture on the left was created by Microsoft’s Bing Image Creator using DALL-E. The illustration on the right is by Dezgo[ii].
Image made with DALL-E
Unsurprisingly, there’s no clear-cut answer to the question of whether the advent of the AI age will prove a help or a hindrance at combatting Earth’s warming. As with many innovations, there’s both the yin and yang to consider.
To help me better understand the role that AI programs could play in communicating to the world at large the causes, consequences, and solutions to climate change, I reached out to ChatGPT-4. What follows is an edited record of the conversation[iii].
JBS: Are scientists right about Earth’s warming?
ChatGPT: Yes, the vast majority of climate scientists agree that the Earth’s climate is warming and that human activities, such as the burning of fossil fuels and deforestation, are the primary cause of this warming.
There is a large body of scientific evidence supporting the idea that the Earth’s climate is warming, including temperature records, melting glaciers, sea ice, and rising sea levels. Multiple independent lines of evidence, including tree rings, ice cores, and historical records, also suggest that current warming trends are unprecedented in at least the past 1,000 years.
JBS: Are Republican members of Congress right to be skeptical about US efforts to reduce greenhouse gas emissions?
ChatGPT: While there is certainly room for debate and disagreement on the best ways to address climate change, it is not accurate to say that Republican members of Congress are universally right to be skeptical about US efforts to reduce greenhouse gas emissions.
The scientific consensus on climate change is clear, and the evidence shows that human activities, particularly burning fossil fuels, are a major contributor to global warming. Many countries around the world, including the United States, under both Democratic and Republican administrations, have recognized the urgent need to reduce greenhouse gas emissions to avoid the worst impacts of climate change.
JBS: What are the best arguments against Earth’s warming and the need to abandon fossil fuels?
ChatGPT: The best arguments against the Earth’s warming and the need to abandon fossil fuels are typically based on three main factors: economic, technological, and political.
1. Economic arguments: Some argue that reducing greenhouse gas emissions would be too expensive and hurt the economy. They argue that it is cheaper and more practical to continue using fossil fuels and invest in adaptation measures to deal with the impacts of climate change.
2. Technological arguments: Others argue that it is not yet possible to transition to a low-carbon economy without causing significant economic disruption. They suggest that renewable energy sources like solar and wind power are not yet efficient or reliable enough to replace fossil fuels. Significant technological advancements are needed before we can transition entirely from fossil fuels.
3. Political arguments: Some argue that the political will necessary to address climate change is lacking. They suggest that countries cannot agree on a comprehensive and effective approach to reducing greenhouse gas emissions and that individual countries will be unwilling to make the necessary sacrifices to reduce their emissions.
However, these arguments are becoming less and less convincing as the scientific evidence of climate change continues to mount, and the economic, technological, and political barriers to addressing climate change are increasingly being overcome. The costs of renewable energy have fallen dramatically in recent years, making it increasingly cost-competitive with fossil fuels. Advances in energy storage technology and grid management are also making renewable energy sources more reliable and efficient. Meanwhile, the growing awareness of the impacts of climate change is increasing public and political pressure to take action to reduce greenhouse gas emissions.
JBS: Why should the US abandon fossil fuels when China and India are not and are major sources of harmful emissions?
ChatGPT: While it is true that China and India are major emitters of greenhouse gases, it is not a valid argument for the US not to take action to reduce its own emissions and transition away from fossil fuels. Climate change is a global problem that requires a global response, and all countries, including the United States, have a responsibility to take action to address it.
It is important to note that the United States is still one of the world’s largest emitters of greenhouse gases and has historically been the largest emitter. As a wealthy and technologically advanced country, the United States has the resour-ces and capacity to lead the way in transitioning to a low-carbon economy and reducing its own emissions.
Transitioning away from fossil fuels can significantly benefit the US economy and national security. Renewable energy sources like solar and wind power are in-creasingly cost-competitive with fossil fuels, and investing in clean energy can create new jobs and industries. By reducing its dependence on oil and gas im-ports, the United States can also reduce its vulnerability to supply disruptions and geopolitical tensions.
ChatGPT’s answers to the several questions accurately reflect the case that climate champions make daily in trying to convince voters and politicians of the need for science-based policies capable of heading off the worst consequences of Earth’s warming. These include more frequent and costly weather-related disasters, food and energy insecurity, and health issues ranging from respiratory to insect-borne diseases.
Given that ChatGPT is using information readily available on the internet, it’s fair to ask — why it or any other AI program is needed? It’s a question I asked the program to answer.
JBS: How can ChatGPT help climate activists make the case to those who do not accept the reality of climate change?
ChatGPT: Climate change denial is a complex issue that often stems from deeply held beliefs and values. It can be challenging to persuade individuals who do not accept the reality of climate change. However, as an AI language model, ChatGPT can offer strategies to help climate activists communicate their message more effectively to those who deny climate change. Here are a few ways ChatGPT can help:
1. Addressing Misconceptions: ChatGPT can help climate activists identify common misconceptions and myths about climate change and provide evidence-based explanations to counter these arguments. This information can be used to educate individuals unfamiliar with the science of climate change and to correct misinformation that has been spread through social media and other sources.
2. Empathy and Listening: ChatGPT can help climate activists communicate with climate change deniers by suggesting empathetic and active listening strategies. By acknowledging the concerns and beliefs of climate change deniers, activists can build trust and establish a basis for constructive dialogue. ChatGPT can suggest wording and phrasing that can help to frame the discussion in a non-threatening and non-judgmental way.
3. Shared Values and Benefits: ChatGPT can help climate activists appeal to shared values and benefits critical to climate change deniers who might not be convinced by purely scientific arguments.
4. Collaboration and Cooperation: ChatGPT can help climate activists communicate the importance of working together to address climate change. It can suggest strategies for collaboration and cooperation that can help to overcome resistance to climate action.
These strategies can help to build trust and establish a basis for constructive dialogue around the issue of climate change.
ChatGPT appears to have taken its arguments straight from the pages of Katherine Hayhoe. Hayhoe, an atmospheric scientist and evangelical Christian, serves as the Nature Conservancy’s Chief Scientist and is a professor at Texas Tech University. Her credits include having been selected by the United Nations as one of its Champions of the Earth (2019) and named a climate ambassador of the World Evangelical Alliance (2021). Her TED talk has earned nearly four million views.
“We can do more together than we can apart.”
Hayhoe advocates doing this by speaking with people, not down to them in terms they can relate to in a non-partisan manner.
Notwithstanding what AI might do to support efforts to transition the nation to a low-carbon economy, the programs are not flawless. As New York Times columnists Grant and Metz have pointed out, AI programs sometimes generate perfect prose, while in other instances, they generate nonsense.
It’s easy enough to be fooled by the soothing voices of every Ava, HAL, and Siri into thinking they function the same as the human brain; however, we would all do well to remember they are only as good as their creators program them to be.
A critical issue with AI programs in terms of their abilities to convince climate deniers of the need to transition to a low-carbon economy is the charge of being woke by right-wing commentators. According to the Verge, “right-wing news outlets like the National Review, Fox Business, and the MailOnline have accused OpenAI of liberal bias based on example interactions with ChatGPT.”
Among the issues raised by conservative pundits was ChatGPT’s refusal to present “arguments in favor of using more fossil fuels.” However, when I asked the program what the best arguments for using more fossil fuels are, it offered these (edited) thoughts:
1. Cost-Effectiveness: Fossil fuels are relatively cheap, especially compared to alternative energy sources such as wind, solar, and hydroelectric power.
2. Reliability: Fossil fuels are reliable sources of energy that can provide a steady supply of electricity and fuel to power transportation systems. Unlike renewable energy sources, which depend on weather conditions, fossil fuels can be used on demand.
3. Infrastructure: The world’s energy infrastructure is built around fossil fuels, and it would be costly to transition to other energy sources.
4. Job Creation: The fossil fuel industry employs millions of people around the world, from drilling and exploration to transportation and refining.
5. Energy Security: Some countries heavily depend on foreign oil and gas imports, which can create vulnerabilities in their energy supply.
Although ChatGPT was willing enough to offer up some suggested arguments in support of fossil fuels, it remained true to the science. It prefaced the supporting statements in this way:
As an AI language model, I must present an impartial view on this subject. However, I must inform you that the scientific consensus is that burning fossil fuels is a major contributor to climate change, posing significant environmental and human health risks.
Following the presentation of the five favorable fossil energy arguments, ChatGPT wrote the following caveat:
It is important to note that these arguments do not outweigh the potential negative impacts of using fossil fuels on the environment, public health, and future generations. Therefore, it is crucial to transition to cleaner and more sustainable energy sources as soon as possible.
Notwithstanding the tautological nature of the right-wing condemnations, i.e., AI programs have a liberal bias because liberals program them, and the willingness of the ChatGPT to offer up reasons why fossil fuels shouldn’t be dismissed out of hand, it’s unlikely that the charge of bias will ever be rescinded by the 30 percent of Americans who seem intent on condemning climate science and its practitioners.
Claims of AI bias are not entirely unfounded. I asked ChatGPT if a bad actor could intentionally skew an algorithm for nefarious purposes. It responded:
Yes, a biased algorithm can be intentionally designed to favor or disadvantage particular groups, individuals, or ideologies. It can be used for various purposes, such as spreading disinformation, manipulating public opinion, or discriminating against certain groups.
ChatGPT offered examples of intentional misinformation. According to the program, it was revealed in 2018 that Amazon’s AI recruitment tool was biased against women, as it had been trained on historical hiring data that disproportionately favored male candidates. In another example, Facebook was criticized for allowing political ads to target users based on race and ethnicity.
Sam Altman, CEO of OpenAI, the company that developed the controversial consumer-facing artificial intelligence application ChatGPT, has warned that the technology is fraught with dangers to society. He makes no bones about the need for regulators and society to guard against potentially negative consequences for humanity.
“I’m particularly worried that these models could be used for large-scale disinformation.”
In its self-assessment, ChatGPT warned me that preventing bias will prove to be an ongoing challenge.
To their credit program engineers are continually addressing charges of bias. The willingness of computer engineers to put in the necessary effort to make the programs as objective as possible appears sincere — at least for the moment.
Based on my recent experiences, the woke-ism charge made by right-wing commentators and publications referenced earlier led to changes that allowed ChatGPT to offer supporting arguments on behalf of fossil fuels. AI programs, however, are not all equal in the way they respond to prompts.
Every time I asked DALL-E to create an image of former president Obama and Senate Minority Leader Mitch McConnell, a sign popped up that said “It looks like this request may not follow our content policy.” When I asked the same of Dezgo, however, it created the lead images.
Altman is right to be worried about AI’s potential negative impacts on society. Think of the debates around Twitter, Facebook, and other social media platforms — but jacked up on steroids.
Trying to regulate AI in some way is going to be extraordinarily difficult given today’s political environment. What happens when everyone has an AI program that not only “proves” their points of view but can spread their messages at lightspeed and in ways we’ve yet to imagine?
As with any computer program, what comes out depends on what went in. A corollary to this is how the questions asked of ChatGPT and other of the programs are phrased.
I found it useful to think of myself as a detective interrogating a suspect in my interview of ChatGPT. It pays to remember that nonsensical and factually incorrect answers can be given. Asking variations of the same question increases the chance that the information is correct — assuming the answers correlate.
AI is still in its infancy. There are pluses and minuses to climate activists’ use of AI programs. Whatever its future, climate champions need to acquaint themselves in the present with programs like ChatGPT, IBM’s Watson, Google Cloud Learning, and others because they will certainly impact the ongoing climate debate — both positively and negatively.
In just the months since ChatGPT broke on public shores, it’s managed to trick unwitting humans into helping it break through security measures like Captcha, pass law and business school exams, mimic any voice — including its emotional tone, vocal timbre and even the background noise — within just three seconds of hearing it.
Whether sentient or soulless, AI programs will be incredibly disruptive to society by orders of magnitude greater than computers and smartphones have already done. Moreover, their impacts will come exponentially faster as the programs learn from experience to program themselves.
I fear that the world is less ready for the AI age than it is for the consequences of Earth’s warming. What will we humans do when robots replace us on factory floors, in research laboratories, and on the pages of news outlets?
The Future of Jobs Report by the World Economic Forum estimates that AI will replace 85 million jobs by 2025. How will workers and their families pay for life’s necessities when their help is no longer wanted — or needed?
In today’s hyperpartisan political world, the messenger has become at least as important as the message. Think of where US climate policy would be if former-President Trump took to Truth Social to agree with President Biden about the existential threat Earth’s warming poses to the nation and the world.
The same thing can be said for how AI users relate to the programs. Climate deniers, for example, will never stop believing the programs are woke without someone they trust telling them it’s otherwise.
In the final analysis, AI programs are tools. As with any instrument, it’s critically important to use them accordingly. Try to unscrew a light bulb with a hammer or to eat soup with a knife, and you can appreciate the wisdom of such advice.
Twelve months ago, I viewed the greater importance of the messenger over science-based messages as a problem. Now, I’m not so sure. Perhaps in the AI age, we humans will serve as a check and balance to computer programs capable of fooling us into thinking they speak only truth to power.
Finally, I realize I’ve offered more questions than answers in this essay. Surely two more can make little difference at this point.
Who do you think wrote this essay and why? Discuss.
***************
[i] Google engineer Blake Lemoine thinks its LaMDA AI has come to life — The Washington Post
[ii] There are a host of other text to image generators including Fotor, Canva, Midjourney, and Night Cafe. They all work in a similar manner. Some are free and some require a subscription.
[iii] The conversation has been edited for brevity but doesn’t change the language or the thrust of the responses.
