patrick.net

I feel like this question doesnt get asked enough. We are told Daylight Saving Time exists to save energy, don't waste electricity. Yet at the same time, the government is pushing massive AI expansion that consumes orders of magnitude more electricity, and no one seems concerned.
It’s hard not to feel like we are living in an absurd, upside-down world. And honestly, if anything needs to go, it's Daylight Saving Time.

That is my argument for getting rid of Daylight Savings, its original purpose is gone, it's not there. It's a useless relic of the past like a feather pen and a floppy disk.

Below is an answer from AI, they don't even hide it:

AI uses significant and rapidly growing amounts of electricity, primarily in data centers for training models and running applications (inference), with estimates suggesting it could soon rival the consumption of small countries, requiring massive new power infrastructure and impacting global emissions, though the exact figures vary by task (video is high, text is lower) and model efficiency.
Key Energy Consumers
Training Models: Training large language models (LLMs) like GPT-3 consumes enormous power, with one estimate for a single GPT-3 run using gigawatt-hours (GWh).
Inference (Usage): The energy needed for users interacting with AI (like asking ChatGPT a question) is a huge factor, with Google noting it accounted for 60% of their AI energy use from 2019-2021.
Data Centers: Massive data centers housing AI servers need vast energy for computation, plus significant power for cooling and memory.
Specific Energy Usage Examples
Video Generation: A single AI video can use around 90 Watt-hours (Wh), much more than text (0.047 Wh) or images (2.9 Wh).
ChatGPT (Inference): Estimates suggest it could use 564 MWh per day for OpenAI's infrastructure.
Google Gemini: A single median query uses about 0.24 Wh, but this is just the chip, not the full data center process.
Future Projections & Impact
Country-Level Usage: Some analyses predict AI server electricity needs could reach 85-134 TWh annually by 2027, comparable to countries like Argentina or Sweden.
Data Center Growth: Data centers are among the fastest-growing emission sources, with new facilities planned to be much larger and power-hungry.
Total Data Center Consumption: Global data center electricity use (including AI) was around 415 TWh in 2024, about 1.5% of global consumption, but growing fast.
The Bottom Line: AI's energy use is a major, rapidly increasing demand, driven by powerful hardware, massive data centers, and the sheer scale of user interactions, leading to substantial energy bills and greenhouse gas emissions, even as some models become more efficient.