I stopped writing and building for more than six months because the ChatGPT-fueled artificial intelligence “revolution” made me feel useless – and worse about myself than ever before.
Then, I had an epiphany. One that took far too long to come but lots of testing and time were necessary to conclude:
Using large language models (ChatGPT, Gemini, etc) is not just overrated but is actually making you a dumber, less successful person. It’s a literal illusion that has paralyzed the minds of millions of people, including myself.
That said, this article is not anti-AI at all. I actually love AI.
What I don’t love is what it’s, inadvertently or not, doing to people’s motivation and perception. This article is my attempt at making sense of what’s really going on.
Recently, I’ve spoken with dozens of people about it. There’s a serious doom and gloom in the air with some people even feeling quite depressed.
But with so much information and capabilities at our fingertips, people should be feeling supercharged, creating things never before possible, right?
We are told you can now 10x your productivity. And that this is bigger than anything the world has ever seen.
So, why are some of us feeling dumber and more “stuck” than ever before?
I think I can answer that.
First: What An LLM Actually Is
This section isn’t a boring filler explanation. It’s unbelievable how few people know what it actually is yet use it all the time.
Simply put, LLMs like ChatGPT use something called Transformer architecture, which is a really smart breakthrough that looks at an entire body of text to determine which words are most important for other words.
Prior to this, we were limited to looking at one word at a time. Now, companies can train their models on absurd volumes of text (taken at any cost) over and over and over until it can predict which word should come next.
It is, factually, a next word predictor. That’s all it is – it’s just really, really good at it because billions of dollars have been spent to train on every byte of data they can possibly find.
Effectively, this now gives us the ability to communicate with text better than ever. Where this actually supercharges productivity:
- Searching and researching
- Translating
- Creating drafts
- Coding (to an extent)
- Summarizing
- Contextualizing
And don’t get me wrong. This ability is a legitimate improvement for customer service, legal and medical research, note-taking, and just about anything where being able to “talk to a dataset in natural language” is helpful.
That is what we have. That is what the current artificial intelligence “revolution” is built upon.
Think of it as a friend who has a really, really good memory of everything people say, but he has no idea what any of those words mean. So, when you talk to him, he can respond with sentences that make total sense – but will never have an actual new idea or interpret what you said in a nuanced way.
Now imagine that friend has memorized everything ever posted online and replies immediately. He’s very valuable for some things (the list above), but he has the same limitations.
What An LLM Isn’t
Because these chat bots will relentlessly predict the next word over and over no matter what you ask it, a couple of things happen:
- It’s addicting. Instant gratification is always addicting.
- It appears more intelligent than it actually is.
Let me reiterate: I said searching and researching is one of its values. And it is! I use it all the time to learn things.
It is great at teaching you things you don’t know.
But here’s where the trap emerges. The addiction to its readiness and the overestimation of its intelligence makes us push it harder.
We begin to expect it to somehow magically be something more than a word predictor trained on existing information. And we keep pushing it, thinking we just need to prompt it a little better. A little more. A little different.
As if it’s magically going to be sentient and come up with anything new or groundbreaking.
It is literally impossible. It is not:
- Good at brainstorming (contrary to what you may think)
- Good at writing copy
- Good at drafting a business plan
- Good at therapy
- Good at analyzing competitors
Here’s the harsh reality. If ChatGPT gives you any of those things, and you think it’s good, one of two things are true:
- You’re below average at that thing and can’t spot why it’s mediocre.
- You’re lazy, and the magic of getting something – anything – instantaneously is more acceptable than creating something special that takes time and thought.
Interestingly, those in category #1 are actually leveling up while those in #2 are leveling down.
Everyone’s talking about Ai taking over every job in the next 5 years, and all I see is a really uneventful, boring race to the middle.
It’s going to leave things wide open for those of us who understand building something of actual value requires more than prompting a chatbot that was created to predict the next word – with bias toward you, no less.
The AI Epidemic Nobody’s Talking About
People email me, the motivation completely sucked out of their souls, to say it’s depressing because it’s just not worth pursuing any business since Ai will simply do it better in the near future.
Then the other half of people are so-called vibe coding some nonsense, thinking they’re at the forefront of the future. They’re launching things that are cool to tell their friends about but actually make and will always make $0 per year because they solve no actual problem.
I know this because in the past year, I’ve been both of those people.
But, after months of using every imaginable LLM-based tool, I saw that we’re being fooled by something that’s certainly not new:
The promise of easy. In particular, easy money.
“You can code ANYTHING now without knowing code,” these vibe code platforms tout all the while, in reality, they’re the only ones making actual money from the movement.
It always goes back to the same thing: people want to make a lot of money by doing very little.
That core trait of human beings, combined with the absurd promises and hype from those leading the charge, is what is driving this revolution.
This revolution based around a text predictor.
And I am no longer listening to the people so obsessed with AI that they might as well be AI itself.
You just don’t get it. You’re missing the boat. In 5 years, society will be completely transformed.
The Soul-Destroying Hype
Going back to how I started this article, I honestly felt like writing was useless. I was so impressed with what ChatGPT could give me, in mere seconds, that I felt my own writing was legitimately obsolete.
I told a friend that one of my biggest advantages in business (not to mention joys in life) was quickly becoming useless. After all, you could take this very article, feed it to an LLM, and tell it to write something on any subject in my style.
And it would do a pretty good job.
Combine that reality of today with the incessant tribal shouting from the rooftop, out of the mouths of every CEO, that AI is about to replace everything that ever made us special… and you have a pretty dire outlook.
We’re told that AI is going to replace virtually every job, from data entry up to surgeons, in the next few years. Journalists cling to small sound bytes like “AI will cure cancer,” something so absurdly speculative. Or respected CEOs sharing videos of robotics that are supposedly groundbreaking but are so ridiculously far from being part of our daily lives that it’s laughable.
Don’t get me wrong. I think technological advancements are exciting, and people way more intelligent than I am are doing cool things.
But, please, operate in reality and ignore the loud noise from the self-serving minority that make you believe you’re useless and can’t possibly build something that other people will want and even pay for.
We’re being promised a future nobody even wants.
“Imagine a world where everything is simply done for you” isn’t what anybody’s seeking. And even if AI does get to the point where virtually everything we’re manually doing right now becomes automated, so what?
Like we have throughout the history of time, we just adapt and want new things, which leads to new businesses – run by humans.
How New Technologies Actually Impact Society
I’m not going to bore you with a history lesson on the printing press, the industrial revolution, the Internet, or mobile phones.
But we can use common sense.
As I write this, I have landscapers here to plant a bunch of flowers, bushes, and whatnot. Looking out the window, I see one of them using some kind of machine to dig the holes super fast.
I can’t help but think when the local hole-digger heard about this new invention decades ago, he thought he was screwed.
Why would anyone hire me to dig a hole when this machine can do it faster and probably even better?
I won’t insult your intelligence by presenting this like it’s some eye-opening reveal, but what ended up happening is that guy learned how to use the tool to dig more holes, faster than ever.
His years of digging holes manually didn’t suddenly mean nothing. His foundational understanding of digging holes became more important than ever as he was now able to do more and increase his productivity.
And that’s how many people rightfully look at LLMs.
But many people are pushing it beyond what it’s capable of, fooled by the addictive illusion created from its confident responses.
Will AI continue to get more capable? Yes, of course. It has been for decades after all.
Technological shifts take time. They work themselves out while making some businesses and careers obsolete but giving rise to new ones that we can’t even predict.
Staying Grounded In Reality
Whatever you’re working on, ignore the hype and focus on reality.
If the capabilities of large language models were quietly brought into products of all kinds, we would have all said basically noticed how awesome it is and then continued on with our lives, using it where it makes the most sense.
We would have just communicated with text a whole lot better than ever – and loved it.
But, instead, we were given the world’s greatest illusion: a chat interface that responds like magic and feels like science fiction along with comments that it’ll cure cancer, that it’s bigger than electricity, and that society is about to change forever.
Because of a text predictor.
One that is quite average when it comes to generating content and ideas that will make you successful. And, to reiterate, if it gets better, it’s better for everyone.
The easy way never lasts long for this very reason.
Using ChatGPT to plan or run your business is a bad idea. It’s extremely good at making you think you’re progressing when you’re not.
That’s why you have 100 threads with dead end brainstorming sessions that resulted in nothing at best – a massive waste of time more likely.
You’re aimlessly deliberating with a non-reasoning text-predictor that has been built to serve you – not challenge you, not come up with new ideas because it literally can’t.
Your next brilliant business idea, ad copy, marketing strategy, or piece of content is going to come from you and other humans. While everyone races to the middle because it’s easier, put the text-predictor away and rediscover your mind.
If you disagree with all of this, that’s great. In a world of robots validating our every thought and offering to “whip up a 1 page PDF” for that, I welcome your humanity.
