The AI Bubble Is Real. But Not For The Reasons You Think.

The AI bubble is real, but not for reasons that some doomers would assume it is.

The technology, even an LLM or AI model alone, is far from pointless technology as some people like to make it seem — but most people are severely illiterate with these tools, and computer systems in general.

The launch of ChatGPT in 2022, which we will speak intensively on in this article, was perhaps one of the most irresponsible moves from the standpoint of creating a new rat race of un-informed people with near zero-literacy on how this stuff works.

For for years now, we have been bombarded with content about new AI this, new AI that. New models. New tools. Agents. With that, the pressure that if you’re not using it, you’ll be left behind.

In reality, many people are now "building with AI" for the sake of building instead of starting with an actual problem to solve. People are in many cases now, over-building AI systems without any real grounding in computer system principles, let alone LLM's, AI models, and how these are designed to work effectively.

Result: Billions of dollars flooding into these companies across over a billion monthly users, many of whom don’t know how to use the technology effectively — and therefore don’t see as much return as they could have otherwise if they knew how to use it. 

That is the bubble. Illiteracy, people and businesses burning money on tools they don't know how to use well, and depending on corporations for the opportunity to do so.

What Actually Happens Next?

The inevitable outcome of this bubble is a pop. That's obvious. Consumers on average are realistically paying 4 to 70 times what they should actually be paying for the same service they recieve right now if corporations were to charge propely for compute. It's possible that come late 2027, prices are substantially higher for the average consumer.

The likely result: average people are squeezed out of cloud subscriptions to the point of not using them at all — or they will be forced to use them in a dramatically different way in order to stomach the investment.

The third option, and the dark horse that most people don't see coming, is local, independent AI solutions that run everything on your own hardware. The "Local AI Exodus" is a strong lever that will inevitably play a key role in the stage of disillusionment as the bubble pops — people realize they can get the same thing, for dramatically cheaper, not to mention private, and under their own control.

Why "AI" Doesn't Vanish After This

Machine learning and computer system techniques have been deployed, utilized, and were incubated in the military industrial complex (1960's) as well as corporate environments (1990's) for decades before a consumer ever talked with a chatbot. We're talking 60 years.

No, for long, this did not look like the AI models or LLM's of today, but the fact remains: sophisticated systems for pattern recognition, signal analysis, and automated execution have been in the works for decades. LLM's are merely an iteration in the field. Useful in some cases, not in others, the deployment matters.

Anyway, for four years behind closed doors starting in 2017, LLM’s were useful, but not a hyper engaging chatbot you spoke at 3am with... This technology was used to build real systems. They were integrated into real workflows. They were used with the same level of tact, maturity, and discipline that had been exercised by engineers for almost 60 years prior with previous iterations in machine learning technology.

The Launch Of ChatGPT To The General Public: The Irresponsible Move? Probably.

Everything changed in 2022 with the launch of ChatGPT.

At its core, this took the same technology that engineers were working with since 2017, but the big difference is that it carried a level of behavioral “enhancements” and guardrails that had never been seen before.

Suddenly, what was once a serious tool in serious environments was hyper-engaging, and designed to appeal to the user as much as, if not more than it was designed to be of genuine utility to the user.

Notably, it created the arrangement for people to send their most personal data and deeper windows into their active reasoning to boxes sitting thousands of miles away from them.

People were given no real container for literacy, maturity, or efficacy with the tool. They were given a black box chatbot with no understanding of how it works, where exactly their personal data was going, and how to use it effectively.

This, in a nutshell, is what led us here. Lots happened along the way, but this it the key event that most people have never examined.

The tool was decent despite some manipulation, but the container was totally off for the average person to adopt it with maturity and tact.

Independent AI Ownership, Literacy, and Escaping The Bubble Before It Eats Your Lunch?

We percieve every single issue with modern AI as an ownership issue or a literacy issue. Nothing else. Thats why we wrote two manuals on it.

— AI cannot be anything but neutral until it is owned, and steered by a corporation with institutional backing.

— AI cannot be anything but neutral until someone using it has no idea on how it's most effectively used.

Cognitive atrophy, sensitive data leaking outside of your control, "hallucinations", and more: are a symptom of who owns it, as well as how competent/literate the user is.

We teach independent AI ownership on a modular machine you can control for decades. You are not locked within someone elses terms. It is the greatest definition of freedom with the technology, and using it as you please. The manuals teach you assembly, deployment, and security for your own private system, start to finish. No data centers are needed.