DeepSeek vs. ChatGPT and Gemini: Which AI Model is Truly Better for the Environment?

Chinese entrepreneur Liang Wenfeng recently launched DeepSeek’s new AI app, surprising many in the tech world. This app operates at a significantly lower cost than popular models like ChatGPT and Gemini. As a bonus, it could be better for the environment, especially since many tech companies, including Google and Meta, have faced criticism for their heavy energy consumption and even revived coal plants to meet demand.

In theory, any AI tool that uses fewer resources should be more eco-friendly. However, experts remain cautiously optimistic about DeepSeek’s impact. David Rolnick, an assistant professor at McGill University, mentioned that we have very little data on the environmental effects of ChatGPT and DeepSeek. “Any numbers about their environmental impact are largely speculative,” he said, explaining that the differences in energy consumption depend on usage rather than the algorithms themselves.

DeepSeek might signify progress in sustainability. Climate scientists worry about the rising energy demands of data centers in the U.S., which are anticipated to double or even triple by 2028. Rolnick pointed out that the environmental concerns linked to large AI models stem from their energy use during training, water usage for cooling, and the environmental costs of hardware production.

Rahul Sandhil from MediaTek highlighted DeepSeek’s potential. This “open-weight” model can be downloaded and run locally, provided users have the right hardware. If successful, some regular AI tasks could shift away from data centers and run on personal devices instead. Additionally, DeepSeek’s open-source nature stands in contrast to the challenges faced by companies like OpenAI, which have struggled for profitability.

Adoption will be crucial. If users don’t switch to DeepSeek, its environmental benefits will be moot. Furthermore, according to Anthony Cohn from the University of Leeds, DeepSeek is limited to text and doesn’t yet handle other input types like images or video, which requires much more computational power.

From a financial perspective, how DeepSeek conserves resources raises questions about sustainability in AI. Julie McNamara from the Union of Concerned Scientists urged regulators to scrutinize energy forecasts tied to fossil fuel infrastructures. She argued that AI doesn’t necessarily have to demand massive energy investments or negatively impact the environment, contrary to some claims.

McNamara emphasized that electricity demand will likely increase regardless of data center needs, and society must evaluate how much we want to rely on large AI models. Rolnick noted that many tasks traditionally assigned to AI could easily be managed by simpler methods, like Google searches or human input. “The environmental advantages of one AI model over another are minor compared to evaluating if we even need these algorithms,” he said.

Addressing the relevance of DeepSeek, McNamara expressed concern over fossil fuel companies using AI’s energy demands to justify new power plants. This moment could reshape conversations surrounding fossil fuel use and public health standards. “It’s crucial to question the narratives surrounding AI and its energy consumption as we strive for a more sustainable future,” she added.

Source link