A group of activists from diverse climate, tech, and anti-hate speech organizations sent a letter to Senator Chuck Schumer demanding that AI’s impact on climate change be addressed.
Schumer already has his hands full hosting the AI Insight Forum which held its first session this week. That forum saw AI regulation discussions mainly focused on addressing ethical, legislative, and civil rights issues.
The letter, a little ominously titled “Final letter to Sen. Schumer on Climate & AI”, was endorsed by some notable names like Amazon Employees for Climate Justice, Greenpeace USA, and The Tech Oversight Project.
The organizations behind the letter each have their own narrow interests, but the main thrust of the letter focuses on the danger that AI poses to the environment.
The first key area the letter mentioned was the “enormous energy requirements and the carbon footprint associated” with creating and running LLMs.
The second area of concern was “the ease and speed with which people and organizations can use LLMs to produce and distribute climate disinformation.”
Today, tech CEOs are descending on Capitol Hill to spread their propaganda about AI. Since we haven’t been invited inside, we’re knocking down @SenSchumer‘s door with this message: All AI policy tackle climate change, head-on. pic.twitter.com/peMXzfNk5C
— Climate Action Against Disinformation (@caadcoalition) September 12, 2023
The examples the letter listed as evidence of these dangers were well-referenced and made valid points. On the face of it, AI really isn’t great for the environment.
The letter quoted reports that “Google consumed twice as much energy as the entire city of San Francisco” in 2020 and that the company’s “on-site water use rose roughly 20 percent in 2022.” They didn’t mention it, but Microsoft’s water usage doesn’t make for good reading either.
Addressing climate disinformation, the letter stated, “Researchers have been able to easily bypass ChatGPT’s safeguards to produce an article from the perspective of a climate change denier that argued global temperatures are actually decreasing.”
To address these and related dangers the authors of the letter urged Schumer to add a number of climate-related provisions to the AI legislation he’s working on.
AI climate change demands – Are they workable?
Demands such as requiring companies to self-report on energy use and emissions related to AI development sound reasonable. Some companies like Google are already doing that.
The demands regarding “climate disinformation” may be more tricky as there is a lot of debate regarding what constitutes “disinformation”. The arguments fall pretty much left and right of the US political spectrum, so getting bipartisan support for this is unlikely.
The immediately measurable impact that AI is having on natural resources and carbon emissions is undeniable. But AI is also having positive effects on the environment too.
The huge jump in processing power and GPU efficiencies means that today’s models are less resource-hungry to train and use than those from a year ago.
Should we be holding AI back now to reduce its impact on the environment? Or will its advancement inevitably lead to a net positive effect on our climate?