Why Conscious AI Would Be Bad for the Environment

This is a guest post by Griffin Kiegiel.

Since the meteoric rise of ChatGPT in 2021, artificial intelligence systems (AI) have been implemented into everything from smartphones and electric vehicles, to toasters and toothbrushes. The long-term effects of this rapid adoption remain to be seen, but we can be certain of one thing: AI uses a lot of energy that we can’t spare. ChatGPT reportedly uses more than 500,000 kilowatt-hours of electricity daily, which is massive compared to the 29 kilowatt-hours consumed by the average American household.

As the global temperature and ocean levels rise, it is our responsibility to limit our collective environmental impact as much as possible. If the benefits of AI don’t outweigh the risks associated with increasing our rate of energy consumption, then we may be obligated to shut down AI for the sake of environmental conservation. However, if AI becomes conscious, shutting them down may be akin to murder, morally trapping us in an unsustainable system.

AI consumes a lot of energy

The New York Times recently reported a huge surge in electricity demand in the United States. While there are several contributing factors—such as an increase in the number of remote jobs and electric vehicles—one key factor is the construction of hundreds of new data centers. Since 2019, Virginia alone constructed 75 data centers to meet rising demand for cloud services and large-language models like ChatGPT. Following suit, utility companies across the country are proposing building several new power plants, which rely primarily on fossil fuels.

This is the exact opposite of what we need to be doing right now. There is no doubt that humanity’s industrialization of the natural world has negatively affected the global climate, and it’s clear that we need to be reducing the amount of energy we consume. Shutting down AI and halting the construction of new power plants by reducing electrical demand would be a step in the right direction.

Unfortunately, it’s unlikely that AI development will slow down. The big tech companies are currently in an “AI race”, investing hundreds of billions of dollars in AI development in an effort to create the most profitable AI possible. This technology is certainly profitable, and as long as that continues to be the case, we can expect competition to drive the advancement of AI systems.

Conscious AI?

Virtually no one believes that today’s AI systems are conscious, but this may not be the case for long. As the quest to build the world’s first artificial general intelligence continues, these systems are getting better and better at presenting themselves as conscious beings.

The nature of consciousness, despite centuries of philosophical study, remains largely mysterious. As such, we struggle with identifying consciousness in other beings. Our best tool for identifying conscious beings seems to be the principle, “If it acts like it’s conscious, then it probably is conscious.” Of course, consciousness comes in degrees. Dogs seem, in some sense, less conscious than humans, insects less conscious than dogs, etc. But it’s highly plausible that all conscious beings are moral patients, worthy of respect.

Given the high rate at which AI is improving, it’s plausible that AI will one day be able to behave like a conscious being. Once this happens, the case for conscious AI (CAI) becomes much stronger. Unless we make significant advancements in our understanding of consciousness before then, there will be little reason to deny that beings who behave in a conscious manner are conscious.

It seems unlikely that CAI would consume less energy than current AI. While there are efforts to create sustainable AI, more complexity will plausibly require more power, and a conscious system would likely be more physically complex than a non-conscious system. Even if we were able to increase their efficiency, demand would likely increase in response, maintaining the same level of energy consumption.

Let’s avoid this moral dilemma

The day CAI is created will be the day a new species is introduced to the world. This species will likely be more human-like than any other conscious species we’ve encountered, and as their creators, we may be obligated to maintain their existence. This would mean maintaining the energy-guzzling power stations and data centers that they rely on, worsening the climate crisis. That is, unless you’re comfortable with committing genocide for the sake of the environment.

If it came down to us or them, I’m sure most of us would choose to shut down CAI, but it would be a much more morally fraught choice than the one we can make today. Today we can choose to shut down a computer to reduce our energy consumption and take steps towards a truly sustainable future. If CAI arrives, we may have to choose between a sustainable future and ending the lives of conscious beings. With all of the other problems a shifting climate is sure to bring, let’s not add another to our plate.

---

Griffin Kiegiel is a Ph.D. Candidate in Philosophy at Wayne State University. His current interests include AI ethics, environmental ethics, and theories of consciousness.