Which energy juggernaut is worse AI or Al?
The title of this post is misleading click-bait. Thanks to the font design, it is not obvious that I am referring to Artificial Intelligence vs. Aluminum (or Aluminium for my readers outside of North America). Two seemingly unrelated things. Yet, there are still interesting parallels. To keep it more obvious, in the content of this post, let's use AI vs AL.
The dreaded AI. We already know it is here to take our jobs. We also know it is hallucinating, it is extremely biased and it is a guaranteed road to extinction. Now, you can also add another monstrosity to it. It eats electricity like a baby snacking on cookies.
Just read the trending articles:
AI already uses as much energy as a small country. It’s only the beginning - Vox
How much electricity does AI consume? - The Verge
A.I. Could Soon Need as Much Electricity as an Entire Country - The New York Times
If the titles are not recreating in your mind the scene from The Matrix where the dark clouds are spread over the human-abandoned wasteland, I don't know how insensitive you are to this looming and almost guaranteed future.
From these articles you will learn:
- The International Energy Agency (IEA) estimates electricity consumption associated with data centers, cryptocurrency, and artificial intelligence represented almost 2 percent of global energy demand in 2022.
- Training a large language model like OpenAI’s GPT-3, for example, uses nearly 1,300 megawatt-hours (MWh) of electricity, the annual consumption of about 130 US homes each year.
- The average smartphone uses 0.012 kWh to charge — so generating one image using AI can use almost as much energy as charging your smartphone.
Let's now focus on aluminum. It is used in transportation, packaging, construction, electric engines... the list is long. Its mass production started around 1890 thanks to the availability of electricity. Producing 1 ton of aluminum requires 17,000 kWh. In 1900, the production of aluminum was 6,800 tons. That translates to 111,600 MWh or 0.1116 TWh.
In 2022, the reported production was close to 70 million tons, which used about 903 TWh of electricity. That would make the country of Aluminum the 6th largest country in the world by electricity consumption after China, the US, India, Russia and Japan. Yet, when was the last time you saw an article lamenting the unsustainable usage of energy for aluminum? Is there archived news coverage from 1900 hyperventilating that 0.2% of global electricity is spent on aluminum?
Here’s another comparison. In the US, out of all the electricity created, 39% goes to residential use and 35% goes to commercial use. But you should know another fact — 65% of electricity produced in the US is lost within the electrical system!
Here’s another bit of trivia — if all personal vehicles in the US were converted to electric cars, there would be a 25% electricity shortfall.
The point of all of this? With every new shiny thing comes the excitement about all the possibilities and a bright future. You also get the 'wagging fingers' showing you where everything can go wrong. Both sides have valid points and should be listened to. But when presenting numbers, somehow — and perhaps conveniently — they omit the larger context and either minimize or maximize the impact for the right effect.
Reviewing the numbers, perhaps one should start asking questions such as:
- Since we have Google Search, what was the impact on productivity? What problems were we able to resolve to improve our lives?
- Wouldn't it make sense to increase the use of AI to improve the efficiency of the power grid? Imagine how much more electricity would be suddenly available without building new power generators.
- How can we generate an order of magnitude more energy than we do now?
Since the dawn of any civilization, harnessing more energy has contributed to the growth and advancement of society. More power allowed for better materials, better transportation and better communication.
Every time humanity discovered new ways of generating more power, things improved. You might find the work of Kardashev, Sagan and Kaku of interest. They defined 'methods of measuring a civilization's level of technological advancement based on the amount of energy it is capable of harnessing and using.'
If we care about efficiency, making a lightbulb that uses 10W instead of 60W is nice. It provides this warm, positive emotional feeling while changing almost nothing.
But if we discover (with the help of AI) how to generate electricity using fusion, that will change almost everything.
The recurrent pattern? Try to focus on things which matter and can make a difference.