Artificial intelligence is not a miracle. It’s not a digital god or a living brain that’s about to rewrite human evolution. It’s software. Clever software, sure, but still software. It doesn’t feel, it doesn’t think, and it certainly doesn’t “know” anything. It predicts. It guesses based on patterns. That’s it.
AI Is Just Software, Not Sorcery
AI tools like ChatGPT, Gemini, or Claude are not mystical forces. They’re built on algorithms, mathematical models, and huge piles of data. The “intelligence” part is an illusion created by how fast and accurately these systems can process that data.
It’s the same principle that makes your calculator good at sums or Photoshop good at editing.
The difference is scale.
AI tools seem smarter because they have access to more information and can make educated guesses that sound human. But it’s still doing what software has always done: following instructions.
Large language models use transformer architecture with statistical pattern recognition to predict the next most likely word based on training data, not logical comprehension.
They do it bloody brilliantly but they haven’t a sodding clue what they are doing or why!
The Hype Machine Keeps Spinning
What makes AI seem magical isn’t the tech itself, it’s the marketing.
Every startup, tech giant, and “AI influencer” out there is pitching the idea that AI will transform every part of our lives. They’re shouting about “AI-powered” everything, from emails to toasters, because it sells.
But the truth is much of this hype is rebranded automation. That’s it. Rebranding things it’s always done! It’s no different from the buzzwords that came before.
Remember “cloud,” “blockchain,” and “disruption”? The words change, the pattern doesn’t.
The Hardware Is Impressive, But the Cost Is Enormous
Behind all the software wizardry sits an uncomfortable truth.
These systems chew through electricity like there’s no tomorrow. Data centres filled with racks of high-end GPUs (they are the brains and all the muscle) run around the clock, sucking up power – and oceans of water to keep it all cool.
Every time you ask an AI model a question, it’s not floating through some digital ether. It’s being processed on machines that produce heat, need maintenance, and leave a massive carbon footprint.
Meanwhile we’re made to feel guilty for wanting a plastic straw with our milkshake!
US data centre power demand is projected to more than double by 2035, rising from 35 gigawatts in 2024 to 78 gigawatts. Training models like GPT-4 (and we’ve moved on a lot since then) required around 30 megawatts of power. (pdf)
How much power is that?
A lot!
Water consumption tells a similar story. Data centres use water primarily for cooling servers through evaporative cooling systems, with around 80% of that water lost to evaporation.
Yes. Millions of gallons of clean water evaporated but let’s not mention that. It makes for shit headlines!
Scottish data centres are now using enough tap water to fill 27 million half-litre bottles annually, quadrupling since 2021.
Some water companies have already refused data centre applications for water because catchments are closed to new abstraction. For all the talk about the future, AI is still grounded in very physical, very resource-hungry reality.
Data centres currently account for just over 1% of global electricity demand and heading for over 1% of CO2 emissions.
That’s more than the entire UK
And it is getting worse by the day!
People Still Make AI Useful
The magic isn’t in the code, it’s in the people using it well. Just like owning a camera doesn’t make you a photographer, using ChatGPT doesn’t make you a writer or strategist.
The skill lies in knowing what to ask, how to interpret the output, and how to refine it into something that sounds like you. The smartest use of AI isn’t about replacing human creativity, it’s about amplifying it. Those who treat AI like a collaborator instead of a miracle tool get far better results.
Accept AI for What It Is
AI is a tool, not a teacher. It can help, guide, and accelerate things that already exist, but it can’t create purpose, emotion, or genuine human understanding.
The danger isn’t that it’s going to “take over,” it’s that we might hand over too much, believing it’s smarter than it really is. The more people remember that AI is just software, the more effectively we can use it.
So, next time someone tells you AI is changing the world, remember this: so did spreadsheets, search engines, and smartphones. Each one reshaped how we work, but none of them made us less human.
The problem isn’t the technology, it’s how blindly we worship it.
Straight Answers About The Reality Of AI
Is AI actually intelligent?
No. AI does not think or understand anything. It predicts patterns based on data and algorithms, not emotions or reasoning.
Why do people believe AI is magical?
Because of marketing hype. Companies promote AI as a revolutionary force when most of it is just advanced automation with clever branding.
Does AI use a lot of electricity?
Yes. AI systems rely on huge data centres that consume enormous amounts of power and water for cooling, making them energy intensive.
Can AI replace human creativity?
No. AI can assist and speed up creative work but cannot replace original thought, emotion, or human understanding.
What is the best way to use AI tools?
Use AI as a helper, not a replacement. The best results come from combining your expertise with AI’s ability to process information quickly.
Sources
Large Language Models: What You Need to Know in 2025
hatchworks.com
What Are Large Language Models? A Beginner’s Guide for 2025
kdnuggets.com
How ChatGPT Learns via Patterns, Not Logic
linkedin.com
How Does ChatGPT Work? An Easy-to-Understand Guide
skywork.ai
How Do Large Language Models Work? How AI Generates Text
coursera.org
Power for AI: Easier Said Than Built
about.bnef.com
AI: Five charts that put data-centre energy use and emissions into context
carbonbrief.org
Report: Water use in AI and Data Centres Executive summary (pdf)
assets.publishing.service.gov.uk
Scottish data centres powering AI already using enough water to fill 27 million bottles
bbc.co.uk
Taking action on other significant water-using sectors and emerging demands
gov.uk
Transformer architecture: redefining machine learning
toloka.ai
How Transformers Work: A Detailed Exploration of Transformer Architecture
datacamp.com
Transformer (deep learning architecture)
en.wikipedia.org