AI Bubble Will Diffuse, Not Pop
- Tarasekhar Padhy
- Feb 18
- 5 min read
(Disclaimer: Profanity.)
The value of AI is inflated beyond belief. A technology that hasn’t proven its worth beyond certain individual-level tasks is dominating the stock market.
And it’s primarily because there are plenty of circle jerks like the OpenAI→Oracle→Nvidia throughout the AI industry. OpenAI rents cloud infrastructure from Oracle to rent and deploy its models. Oracle needs Nvidia’s chips to build the necessary hardware infrastructure. And finally, Nvidia pumps money into OpenAI through data center investments. [1]
Consequently, AI-disbelievers are singing endless songs about how the bubble will burst to bring terrible economic conditions, where we will be hunting for rats in the sewers.
On the other side, there are lazy cunts who thought investing in AI is the safest get-rich-quick scheme and have put significant amounts of borrowed money. These type of people already lost their own savings to crypto. They need AI to be successful, or they will be destitute ultra pro max.
As always, the truth lies in the middle.
Let’s look at both sides of the conversation to figure out what’s to come.
AI is powerful
Since ChatGPT became publicly available, I have been using LLMs for various tasks in my content writing profession. It has led to visible productivity gains, and I can technically say that “I am making money with AI.”
Initially, the time savings weren’t that high, but it did reduce mental fatigue considerably. Later, more recently, to be honest, I’ve been mixing it up with other LLMs to save huge chunks of time.
I use ChatGPT for research, Claude for structural vision, and Gemini as a collaborative writing tool when creating an outline. Then, for drafting, I’d give the edge to Claude Opus 4.6 and the second position to Gemini 3 Thinking Model.
With such an AI-enhanced workflow, I can produce a 2000-2500-word publish-ready draft that requires nuanced research from scratch in 3 hours. Before AI, it definitely took the whole day and then some.
Similarly, I know web developers, software engineers, and designers who use different generative AI solutions to work faster. These folks are now doing more with less, which is increasing their income.
But that is not the case for everyone.
Power only a few can wield
To use AI tools at a level where it gives you an advantage in terms of speed and quality of work, you need two things: a deep knowledge of the task you want to do and the ability to use relevant AI tools effectively.
I’ve been writing content for various audiences for more than five years at this point. This has given me a fingertip feel of the nuanced steps from ideation to publication. Consequently, I can translate them better to LLMs.
The same applies to coders and designers.
The professionals who have immersed themselves in their craft will be aware of the creative and patterned elements. The creative elements will serve as a structure that becomes a prompt to generate new content, such as text, images, videos, and code.
Furthermore, they need to experiment with their hypotheses in multiple AI tools to find the right fit. There’s also a bit of trial and error while incorporating such solutions into their workflow.
And none of that is easy.
While it might appear that everyone is an expert on LinkedIn, the reality is grimmer.
Personally, I’ve encountered content writers who have more experience than I but struggle to extract the maximum value from AI tools. Yes, they can use AI for one-off tasks, such as finding details about a business or a product and generating rough outlines.
However, that’s barely enough if you are looking to save hours and double your output.
As I mentioned in the previous section, AI helps me wrap up content pieces in 3 hours, tops, that used to take around a whole day. Now, in the same time frame, I can produce three complex content pieces for three separate clients who are completely unique from one another.
Despite the context switch and sky-high quality expectations, LLMs are making it all possible.
However, based on the conversations I am having with other writers and the pace-quality balance of their work, I can confidently conclude that they ain’t shit.
Of course, part of it can be attributed to my mentality, which is “I am the best but never good enough,” encouraging me to keep pushing. At the same time, it wouldn’t be possible without the technology enabling me to do it all.
I don’t want to wank myself off too much here. The point I am driving is that you need to experiment→fail→repeat with these tools on top of being good at what you do to derive value.
And the majority of knowledge workers are stuck with the basics. Are they more efficient today because of AI? Yes. Have they elevated their skill level to stand out among their peers? No.
The experts have left the chat
People who know the true applications and limits of AI in its present state include power users like myself and professionals like data scientists and AI researchers. Such individuals operate in small circles because that’s what deep work demands.
Unfortunately, the majority of public discourse around these generative and predictive models is dominated by influencers, consultants, analysts, early adopters, and dick-riders.
These are the talentless bunch who lack firsthand experience with AI tools, and their whole schtick is to “guide” others in making AI-oriented decisions, such as which company to put money in and which tool to adopt to improve a process.
While the potential of AI is underutilized, its perceived impact is blown out of proportion by these retards who display more passion than an aging whore who’s being fucked by a rich simp.
Consequently, companies invest heavily in AI and force their workforce to use those tools to boost productivity and output. To make matters worse, they fire experienced and top-performing employees to cut costs and justify AI investments.
On the flip side, the innovators are gradually finding ways to extract more value out of LLMs and other generative platforms. As I mentioned above, some are easily able to 2-3x their productivity.
These gains come slowly, which is not at all proportionate to the level of faith placed on these tools by senior executives and business leaders.
What’s next: Slow burn and new dogs
Businesses and individuals who rely on their experience and leverage trial-and-error methodologies to incorporate AI solutions into their processes iteratively will win.
Dumb retards who mass adopt AI tools and force their non-thinking staff to use them for predefined tasks in predefined manners will get cooked. Because these tasks can be automated at scale with agentic AI, which is, apparently, the next big thing.
But none of it is going to happen overnight. AI will take over more jobs after power users find ways to do those things effectively at scale.
I am already seeing organizations struggling to keep up with the market’s evolving needs because they aren’t using AI well enough. Especially in the content marketing industry. Their place is being usurped by new players who use LLMs as if their lives depend on it.
Again, none of it happened overnight. In fact, the aforementioned change took years!
It’s already happening in small scale. Perhaps it might happen on a larger scale, too. Maybe a trillion-dollar corporation, such as Microsoft, will lose this race and die.
Fingers crossed.
Resources
