Lots of people—way more than usual!—are reading and sharing last week’s article about a new study finding that Big Oil’s use of AI to increase production could trigger 3.3 to 13.3 times more climate pollution than the emissions from AI data centers.
The attention is promising. As the study authors told us, expanding the climate conversation about AI beyond data centers to include “enabled emissions” is the necessary first step toward making sure AI doesn’t lock us into dangerous, irreversible warming.
But despite all this reaction, I’m not seeing many people ask what I believe to be the most important question raised by the research:
Why are these huge technology companies making custom AI tools to help the fossil fuel industry produce more oil and gas in the first place?
Microsoft, Google and Amazon have all promised both to fight climate change and to develop AI responsibly. Microsoft says it will become carbon negative by 2030, and that the systems it builds “must also support the long-term health of the planet.” Google says its AI safeguards are designed to mitigate “unintended or harmful outcomes,” and it aims to reach net zero emissions across its operations and value chain by 2030. Amazon says responsible AI includes preventing harmful outputs and misuse, and it has pledged to reach net zero carbon by 2040.
And yet, Microsoft, Google and Amazon have all created custom AI or machine-learning tools for fossil fuel companies to help them increase production. So how are these companies going to respond to research suggesting their actions are achieving the exact opposite of what they promise?
Remember: In the scenarios the researchers modeled, global emissions fell only when AI produced no productivity gains for the fossil fuel industry. It was not enough for AI to help renewable energy, too.
So do Microsoft, Google and Amazon accept that finding? And if so, what are they going to do about it?
I think these are the most important accountability questions that can be asked about this issue right now. So I’m trying to get answers. On Tuesday, I emailed the press teams at Microsoft, Google and Amazon to ask them to respond to the research. In essence, I asked whether they accept the study’s findings; how their work increasing fossil fuel production aligns with their climate and responsible-AI commitments; whether they track or disclose the emissions their technology enables; and whether the findings will prompt them to stop making custom AI tools that help fossil fuel companies produce more.
One thing I expect to hear from these companies is that their AI technology is actually helping oil companies reduce the emissions intensity of their operations. But emissions intensity is not the same thing as overall emissions. AI may help an oil company emit less pollution per barrel of oil. But if the company is producing more barrels than it otherwise would have, total emissions can still rise. That’s why I specifically asked the tech companies whether they’re looking at the whole picture: not just whether AI makes each barrel cleaner to produce, but whether it helps produce more barrels overall.
Now, why am I telling you this? Admittedly, it’s not common practice for reporters to tell their audiences about questions they’ve asked corporations before receiving answers. But I’m doing it because I think it might increase our chances of getting a response.
These days, it has become excruciatingly common practice for companies to ignore reporter requests, particularly when it comes to climate change. I’m hoping that knowing people are watching will inspire Microsoft, Google and Amazon to actually participate in the discussion.
I’m genuinely hoping these companies will take the research seriously, take our inquiry seriously, and provide thoughtful answers. Assuming they will, I’ve given them what I think is a reasonable amount of time to look into everything: A deadline of midday on August 25.
For what it’s worth, a Microsoft media relations representative swiftly responded to confirm receipt. Google also sent an automated confirmation. Amazon has not yet responded to confirm they got the questions, but that’s not always common practice.
The other reason I’m telling you this is because I realized that simply making people aware of the problem of enabled AI emissions is not enough. People also have to know what the solutions are in order to effectively demand change.
One obvious solution is for these tech companies to stop making custom AI tools for the fossil fuel industry. Another is for them to voluntarily track and disclose the emissions their technology enables—and count those emissions when making claims about their climate progress.
If they don’t do those things voluntarily, governments could step in. They could require enabled emissions disclosure, and place limits on AI applications designed to expand fossil fuel extraction. (The study found that carbon pricing alone, even at an extremely high level, reduced but did not reverse the emissions increase. Some direct constraint on AI-enabled fossil fuel expansion would still be necessary).
But if Microsoft, Amazon and Google are truly the climate champions they claim to be, they shouldn’t need to be forced.
I look forward to updating you next week after we get answers. And if we don’t, that will be a story in itself.
Other stories I'm following:
The Trump administration is officially opening untouched national forests to logging. On Tuesday, the Trump administration moved to roll back an environmental rule that blocks road construction, logging, and mining on tens of millions of acres of public forest land. While Trump is framing the move as necessary to control wildfires, as we’ve previously reported, that’s a scam! This is a brazen giveaway to the logging industry that, if anything, will worsen wildfires.
Data center gas plants will boost U.S. power emissions by 20 percent. This is why it’s still necessary to talk about the climate impact of data centers, via Bloomberg:
Data center developers are turning to bespoke natural-gas power plants, a development that promises to dramatically increase carbon emissions and make it harder for US technology companies to meet their lofty climate goals.
Ninety-nine proposed plants tracked by BloombergNEF would emit about 318 million metric tons of carbon dioxide annually if run at industry-standard rates, according to a Bloomberg News analysis. The entire US electric power industry emitted about 1,485 million metric tons of carbon last year, according to Energy Information Administration data, meaning one slice of data center infrastructure has the potential to lift US power sector emissions by 20%, and as much as a third should the new plants run flat out.
Extreme heat more dangerous for people 60 and over than previously thought. From the Guardian:
Older adults will bear the brunt of increasingly frequent sustained spells of extreme heat globally as the world continues to warm up, according to a study in The Lancet Planetary Health. …
“Our results suggest accounting for age-stratified vulnerability, intolerable heat will be experienced at much larger scales and for longer durations, affecting far more people than previously thought,” according to the study, which was led by health academics from Stanford University in California.
Older adults are known to be more in danger because of the effects of very high temperatures. They sweat less, their hearts work harder in the heat and their blood vessels are less able to widen, limiting their capacity to stay cool.
…The study suggested older adults were “at risk of harmful and potentially fatal heat stress” at just 1.5C of warming above preindustrial levels, compared with 4C for younger adults.
Some personal news…
It’s been a busy last few days over in my household, for the most exciting reason: We have a new family member!
Margaritaville—Margo for short—came home with us on Friday from the Brandywine Valley SPCA. She’s a one-year-old pitbull mix (we think?) who loves everything and everyone, but especially Mr. Possum.
Five days in, we’re already in love—and Margo doesn’t seem to mind us, either. Stay tuned for more pics soon. (That’s the point of having a dog after all, right?)
Looks like the Jevons paradox is in play with AI increasing the efficiency of the fossil fuel industry.
Sid Madison
Congratulations on the new addition - Margo is beautiful! LOVE her ears! Pitties are the best - I lost mine in Dec 2020, and still miss her every day.