Member-only story
Google’s AI Denial
What a load of bull.
There is no such thing as a free lunch, and that has never been more true than with AI. Sure, AI models can be incredibly powerful, but they take so much energy to build, maintain and operate that they are barely profitable and have a vast carbon footprint that is only set to grow exponentially. This fact was laid out letter by letter in Google’s recent annual environmental report, which found their 2023 CO2 emissions were 48% higher than in 2019, despite the company’s goal of being net-zero by 2030! Why? Well, over the past five years, Google has dramatically expanded its use of energy-intensive data centres. As Data centres are a crucial part of AI infrastructure, and Google has desperately tried to solidify itself as a forerunner in the AI industry, many people pointed the finger at AI for Google’s worryingly increasing emissions. However, one of Google’s chief scientists, Jeff Dean, claims this simply isn’t the case and that they are on track to reach net-zero in only six years. Sadly, all the evidence points to Google and this chief scientist being clouded by denial.
So, what evidence does Dean have to back up his position?
Well, in a recent interview, he stated, “There’s been a lot of focus on the increasing energy usage of AI, and from a very small base that usage is definitely increasing.” He elaborated “I think people often conflate that with overall datacenter usage — of which AI is a very small portion right now but growing fast — and then attribute the growth rate of AI-based computing to the overall datacenter usage.”
Now, it’s no wonder Dean would defend AI like this; he is currently the lead of Google’s AI division, after all. But the fact he doesn’t back up this position with any solid figures is, at the very least, suspicious. After all, if anyone would know off-hand data to back up such a position, it is Dean.
Firstly, there is a significant amount of data that suggests Dean is wrong here. For example, it has been found that up to 20% of US data centre energy use was driven by AI last year. Whereas, back in 2019, AI used practically none. Considering Google is developing AI faster than most other AI companies, this fact alone points to a significant amount of their increased emissions coming from AI.
