Every day, we see more news about AI and how it is set to change our lives. Traditional information platforms, like Google, are slowly being replaced by AI agents, and AI-generated content is appearing everywhere. It is undeniable that large language models (LLMs) are challenging our status quo and reshaping the way we live and work.

This text is divided into two parts. First, we will demystify how large language models operate at a fundamental level, breaking down the math and algorithms in an accessible way. Second, we will examine the state of AI today who is building it, how much they are investing, its environmental costs, and its broader societal impact. The goal is that by combining technical understanding with critical context, readers will gain the tools to engage with AI knowledgeably

As a business student, large language models can seem far from my field of study, bordering more on computer science than business. I have been highly critical of generative AI for reasons beyond the underlying technology, but I also believe it is important to examine this technology even if we are not data scientists. Through this text, I will try to explain the technical aspects of AI in a way that is accessible to most people outside the field and connect them to a holistic view of AI across political, economic, and social domains.

We cannot discuss our future with LLMs or their role in society if we do not understand how they work. Understanding how large language models operate at a fundamental level acts as a demystifying tool, allowing us to be more critical of the technology we are using.

Having a critical stance against AI does not mean that we are rejecting progress or new technologies. Only by examining the current state of things through a critical lens can we make better-informed decisions.

Whenever a new technology emerges, sensationalism spreads across most of the media we consume. This pattern repeats itself with every new innovation. Sensationalism can be harmful because it encourages blind adoption of technology that we do not fully understand. More importantly, the hype and promises shift the conversation away from the real, uncomfortable questions that we should be asking.

We should be asking ourselves these uncomfortable questions because billions of dollars are being spent on this technology. This investment will influence areas of our lives even if we are not active participants in its development. For example, will society benefit in the long term after the burst of the AI bubble, as it did with the dotcom bubble? During that period, although there was considerable economic damage, society still benefited from the technology that emerged.

Another important concern is sustainability. The rising global spending on data centres and the increasing energy demands to train larger models raise questions about whether this technology will genuinely improve our lives or if it is primarily a struggle for power. While there have been breakthroughs that allow current LLMs to consume less energy, current incentives are not focused on sustainability. We are building infrastructure that harms the environment for a technology that the general public does not yet understand.

I want this text to serve as a starting point where we can learn about LLMs at a fundamental level and also explore how AI and governments are using LLMs for power, as well as the challenges and benefits of these models. My goal is to provide readers with an arsenal of information that allows them to address AI more critically.

This text assumes a basic level of linear algebra and calculus, also it’s recommended to study some set builder notation.

However, I will break down the concepts as much as possible to make them accessible. I highly encourage readers to engage with this journey, even if their field is not technological. Current LLMs are becoming a part of many areas of our lives, regardless of the field, and understanding it at a deeper level will allow us all to become more aware of our usage of it in the future.