The AI boom is not going anywhere anytime soon, and so are its many side effects.
Generating AI compute requires a serious amount of power, and with the advent of more powerful models, things have only been going downhill from here on, as companies tap (and invest billions) into these technologies, which appear to be rapidly running out of resources to sustain them.
The latest in a series of changes involves the renewal of ‘peakers’, a form of ‘dirty power’ that results in substantially more pollution output, for short term gains. It does beg the question though, whether this recent boom is even worth sustaining.
Increased Electricity Demand for AI Forces the Revival of Less Than Ideal Sources of Energy
Chicago’s Pilsen neighborhood recently saw the reactivation of an ancient oil-fueled power plant in order to meet increased power demands. The reasoning can be directly tied down to the bogging down of the existing electricity output by the emergence of AI.
AI is computationally expensive, which in turn requires the usage of massive amounts of electricity to power its cycles. With the recent AI boom, many have been turning to (temporary) greener pastures, and have diverting energy to these datacenters instead.
This has caused a deficit in the energy available to the general public, and these ancient power plants have been brought back to life to compensate for the same.
It should be mentioned here that these ‘dirty peakers’ aren’t exactly new, and have been around since the 60s, having being sidelined into working as backup sources of power generation, often acting as the last line of defense against widespread power outages caused due to uncontrollable factors such as storms and the like.
Ignoring the so called ‘economic case’ of keeping them around, there is little doubt as to the environmental effects of these peakers. Peakers use fossil fuels to generate electricity, and are as a consequence a lot ‘dirtier’ in their power output, relentlessly releasing toxic fumes and pollution onto neighboring areas, which can casue serious health effects in the long term.
They are also not as efficient as modern power plants, and require far more work to maintain and generate the same amount of power.
This move was reiterated by the current US administration, who advocated for the renewal of peaker plants in the face of this manufactured crisis. It’s quite a disappointing move really, and the short term benefits are outweighed by the overall negative effects, which are pushing us further into the past.
Until the AI bubble pops, expect more and more resources to get sidelined in favor of generative AI, which rarely is useful for the average user (or at least in a way that is energy conservative and/or ethically sound).
