ChatGPT Vs Printed Media

The Ethical, Emotional and Carbon Costs of Information Sharing in the Age of Artificial Intelligence

Introduction: A Familiar Debate

Artificial intelligence is the latest in a long line of technologies to face scrutiny over its environmental impact. Every major technological shift—whether it was the rise of the internet, the advent of search engines, or the explosion of cryptocurrencies—has prompted concerns about carbon emissions and energy consumption. Now, AI tools like ChatGPT, DeepSeek, and Gemini are under the same microscope.

Critics argue that AI requires immense computing power, leading to excessive carbon emissions. This is a valid concern. But what if the debate is missing a crucial piece? What if, instead of focusing only on the carbon cost of AI, we compared it to the systems it might replace?

In this article, we will examine the environmental and ethical costs of AI-driven information sharing. We will compare the carbon footprint of large language models (LLMs) with the energy demands of traditional information-sharing methods—particularly printed books.

We will also explore the broader economic implications, asking whether the rise of AI represents a fundamental problem or an opportunity to rethink how knowledge is distributed and valued.

The Weight of History: Technology and Carbon Emissions

Over the past 100 years, with the growing realisation that human activity has been causing the planet to warm up, technological advancements have often been met with concern about their impact on planet Earth’s ability to sustain human life. The industrial revolution led to massive carbon emissions but also made goods and services more widely available. A slew of Innovations from the advent of the combustion engine to the growth of the internet have revolutionized our lives but also increased the ways by which we are damaging the environment which supports our ability to live.

Now, AI is at the centre of the conversation. Large language models require enormous amounts of data to be trained, and their computations demand significant energy. OpenAI’s ChatGPT-4, for example, is estimated to consume roughly 500 tonnes of carbon emissions in training alone. That’s a lot. But how does it compare to the carbon footprint of traditional methods of sharing knowledge?

A Carbon Footprint Comparison: ChatGPT vs. Printed Books

To understand the environmental impact of AI, we need to compare it with something familiar: books. It might seem a bit of a leap initially but one of the things I have personally been using AI for in the past year is to gain rapid access to the information buried inside factual content books. I will, for example, ask ChatGPT to summarize a specific book in 1000 words or less. I’ll get it to read that summary to me, make an assessment on which information I need to zoom into, then ask ChatGPT to expand on those areas, digging deeper until I feel that the information is understood.

And whilst printed books are one of the oldest and most respected ways of sharing information; tangible, durable, with a deep emotional and cultural significance – none the less they also have a substantial environmental cost in terms of carbon emissions.

Let’s take the best-selling book of 2023, This Woman, which sold around 1.5 million copies. The production of a single book, depending on whether it is a paperback or a hardcover, emits between 2.5 and 4 kilograms of carbon. When multiplied by 1.5 million copies, this results in a carbon footprint of roughly 4,500 tonnes.

By comparison, the annual carbon footprint of ChatGPT-4, including its training and daily operations, is estimated at 1,500 tonnes. This means that a single best-selling book emits three times as much carbon as the entire annual usage of ChatGPT-4.

And that’s just one book. The publishing industry as a whole emits an estimated 150,000 tonnes of carbon per year. If we are serious about reducing emissions, should we not be looking at alternatives? AI-driven information sharing could, in some cases, be a more carbon-efficient way of accessing knowledge than printing and distributing millions of books.

The Problem of Unread Books

There is another issue with printed books: waste. Studies suggest that 60% of all books remain unread. Whether bought on impulse, received as gifts, or stocked in the back rooms of libraries, many books never serve their intended purpose. That means their environmental cost—extraction of raw materials, printing, distribution, and storage—often results in little knowledge being transferred.

Of course, books have cultural value beyond their direct use. A shelf full of books provides an aesthetic, intellectual, and even emotional experience. But if we are talking purely about the efficiency of information sharing, AI provides a way to access knowledge without the carbon footprint of producing physical objects that may never be read.

The Ethical Cost: Who Gets Paid When AI utilises the Information Held in Books?

While AI might reduce carbon emissions compared to traditional publishing, it introduces another problem: compensation for knowledge creators. When we buy a book, we support the author, the publisher, and the entire industry that brought it to market. When we ask ChatGPT to summarize that book, who gets paid?

This is a crucial issue. AI can summarize and distribute knowledge efficiently, but it currently does so without compensating the originators of that knowledge. If I use ChatGPT to generate insights from a book, I am benefiting from someone’s intellectual labour without financially contributing to them.

This is part of a broader trend. The rise of AI has already made it harder for writers, musicians, and artists to earn a living by obscuring the attribution of source material in training data, by providing tools which imitate cheaply their skills and by using their data without compensation.

The Bigger Picture: AI and Economic Displacement

This concern ties into a larger discussion about AI and the economy. Many jobs—writers, artists, coders, even legal and medical professionals—are already being disrupted by AI. Yet, despite this displacement, there has been little serious discussion about economic restructuring.

One potential solution coming out of academia is Universal Basic Income (UBI), the basic premise being that everyone receives a basic income sufficient to survive without needing to work for it. In a world where AI handles more and more tasks, this guaranteed income could allow people to focus on creative and intellectual work without worrying about survival. Imagine a system where an author writes a book, publishes it digitally, funded by a UBI system financed through taxes or profit sharing of AI-driven economies. The book never needs to be printed, reducing carbon emissions, but the author is still able to survive financially.

However, UBI remains politically contentious. One or two governments have experimented with the idea and a handful of mavericks in the corporate sector have openly advocated for it – but mainly, it could be argued, to provide a narrative that dissuades governments of the need to place legal controls on AI.

Most governments and corporations seem resistant to even acknowledging the scale of AI-driven job displacement, let alone implementing solutions. And so, at this moment, we are left in an awkward space: AI improves efficiency and accessibility but undermines the economic models that support knowledge creation.

Emotional Attachments and Resistance to Change

Whenever I present these ideas in lectures or discussions, they provoke a strong emotional reaction. People gasp at the suggestion of moving away from printed books, just as other people do when asked to consider giving up their SUVs.

This highlights an important human trait: emotional attachment to familiar tools and traditions. We explore the world through books, through physical media, through objects that shape our experiences. Asking people to move away from these things feels like asking them to give up a part of their identity. It’s hardly controversial to say that part of the reason why certain demographics are so willing to go against the established science, proclaiming that human induced climate change is a hoax, is because doing so provides a narrative that allows them, guilt free, to keep driving the gas guzzlers that they love.

But if we are serious about reducing carbon emissions, we are going to have to challenge our emotional attachments to such things, we all have a part to play. Progress is often uncomfortable, but when the evidence points to a more sustainable alternative, it seems only reasonable to consider it rationally.

A Thought Experiment: What If We Designed an Efficient Information System from Scratch?

If we were starting from scratch—without the history of printed books, search engines, and publishing houses—how would we design an optimal system for sharing knowledge?

We would probably want a system that:

1. Delivers information instantly without the environmental cost of physical production.

2. Compensates knowledge creators fairly so they can continue producing valuable content.

3. Minimizes waste, ensuring that shared information is actually used.

AI-powered tools, in theory, could meet the first and third criteria. But the missing piece remains economic fairness. In my opinion, until AI models find a way to support the people whose knowledge they distribute, they will remain controversial primarily for they undermine the creative economy.

Conclusion: Rational Progress, Not Reactionary Resistance

AI, like every new technology, has both costs and benefits. Its environmental impact should absolutely be scrutinized. But that scrutiny must be comparative. We cannot look at AI’s carbon footprint in isolation—we must compare it to the systems it seeks to replace.

When viewed through this lens, AI offers a potential improvement over the carbon-heavy publishing industry. However, it raises ethical concerns about how knowledge creators are compensated.

The challenge ahead is not whether we should embrace AI or reject it. The real challenge is how we shape it to ensure it benefits society as a whole. That means making difficult decisions—not just about technology, but about how we value knowledge, creativity, and fairness in a rapidly changing world.

Simon Wilkinson
BRiGHTBLACK