- ChatGPT-type AIs are very energy-intensive
- ChatGPT and its rivals will result in at least 4 or 5 times more computation per search.
- Tech giants seek to reduce their electricity consumption
No one wants to miss the boat. After Microsoft and Google which announced the integration of their own conversational agent on their search engines, the Chinese Tech giants Baidu, Alibaba, and NetEase have confirmed that they are working on their own language generation models.
If the euphoria seems in order, the deployment of these AIs is not without posing a major problem in the fight against climate change. Indeed, these technologies require significant computing power for their training and operation.
ChatGPT represents scaling
Asked by WiredAlan Woodward, professor of cybersecurity at the University of Surrey, UK, points out: “Indexing and searching for content on the internet already takes up huge resources, but incorporating AI requires a different kind of firepower”.
He adds : “It requires processing power as well as efficient storage and searching. Every time we see a step change in online processing, we see a significant increase in the power and cooling resources required by large processing centers. »
Asked by our colleagues, Martin Bouchard, co-founder of the Canadian data center company QScale, confirms this idea. According to him, it will take 4 or 5 times more calculation per search on Bing and Google.
While Internet use already accounts for almost 4% of global greenhouse gas emissions, and even if digital is not the most polluting industry, the impact of these new systems could so very quickly feel.
Solutions to consume less?
For now, Microsoft has not responded to requests from the American media. We do know, however, that the Redmond firm has set itself the goal of having a negative carbon footprint by 2050.
On the Google side, this problem seems to have been thought out upstream. Jane Park, spokesperson for the company, explains: “Our results show that combining efficient models, processors and data centers with clean energy sources can reduce a system’s carbon footprint by up to 1,000 times. »
In any case, it is urgent to deploy these new, less energy-consuming solutions. The big technology companies indeed seem very eager to launch these tools and this should be the case in the coming months.
