Is it possible that Artificial Intelligence will destroy the planet?

It’s hard to talk about it Artificial intelligence without thinking about an army of robots that will dominate the world. And what we saw in science fiction movies became reality too quickly: today there is fast technology that responds to us, creates poetry, simplifies our lives and creates content by itself.

And while some are convinced that AI will pose a danger to humanity when it becomes intelligent enough to stop following human orders, There is another risk of this technology that is less talked about: its electrical and environmental cost.

As technology journalist Steve Lohr explains in The New York Times , We are only just experimenting with AI, but its cost is already colossal: big tech companies are spending immeasurably to build and rent data centers and machine rooms.

Last year alone, it is estimated that nearly $105 billion was invested in these giant, and obviously energy-intensive, facilities.

All this shows that If artificial intelligence does not end humanity when it is revealed, it is likely that it will still destroy the planet, but due to its excessive demand for electricity and its contribution to climate change.

Many experts have been warning the world and big businessmen for some time that Artificial intelligence has “obscene” energy demands.

Yet, for journalist Steve Lohr, approaching this problem from another angle could be interesting: for him, Paradoxically, AI could repair the damage caused by its creation by itself.

“AI has the potential to accelerate scientific discovery and innovation in one field after another, increasing efficiency and reducing carbon emissions that contribute to global warming in sectors such as transportation, agriculture and energy production.” wrote the expert in an article published in New York Times.

As the journalist explains, generative AI – like ChatGPT, Copilot, among others – requires a lot of energy to operate: “A ChatGPT query requires nearly 10 times more electricity than a normal Google search, according to a recent estimate.”

And now that many other companies have followed ChatGPT’s pace (and countless other tools are being built with generative AI), It is likely that in a few years, the energy consumption required by global data centers will multiply.

According to a forecast by the International Energy Agency demand is expected to more than double by 2026.

This is worrying, because If data centers choose to source their energy from polluting fossil fuels like coal and natural gas, greenhouse gas emissions are likely to increase.

This is why the journalist of New York Times assured that, To avoid destroying the planet and contributing to pollution, AI should be a “green technology.”

Moreover, If AI is used wisely, it could even reduce greenhouse gas emissions by 5 to 10% by 2030, according to a study by Boston Consulting Group ordered by Google .

And it can also go further: for example, the Drew Endy, a bioengineering researcher at Stanford University He told the same media that AI could give “a boost to biological design” and help discover DNA formulas suitable for more efficient and less polluting agriculture.

In other words, applying AI as a tool to heal the planet could be the best alternative to avoid its destruction.

Another example of how this technology can helping the environment is that it is used to find fundamental metals for the technology industry but also for the fight against climate change, such as copper (which is needed to create electric vehicles), or find geothermal energy for power plants.

A recent case is that of Zanskar, a startup from Salt Lake City, USA, which is trying to increase the success rate of geothermal projects through AI: as Zanskar co-founder Carl Hoiland explains, 90% of projects started from scratch fail because they drill in the wrong places.

But, By combining AI with geological data, the success rate can be doubled or even tripled.

With the same optimism and looking at the historical context, journalist Lohr assured that The efficiency of the technology is likely to increase at an even faster rate than that of AI.

As happened with the famous computing “clouds”, initially (in the 2000s) it was thought that energy consumption would be overwhelming. However, the production of global data centers multiplied before this happened, and energy consumption increased by only 6%. %.

Something similar could happen with AI, experts believe.

“Once the mania dies down, other incentives come into play. “The industry has a strong incentive to become more efficient.” said to New York Times the former scientist of Lawrence Berkeley National Laboratory Jonathan Koomey.

According to the journalist, Currently, large technology companies are already looking at how to optimize their artificial intelligence systems.

For example, they would locate their facilities in northern countries, so that the cold air from outside could serve as a refrigerant, in addition to looking for alternative energy sources to be able to reduce electricity consumption and not destroy the world, at least not in this way.