For some time now, I've been struggling to decipher the essence of what the tone of an AI newsletter should be.
A breeze? Serious? Like an otaku?
But who would have asked? no one.
Until it silently hit me.
It wasn't about the subject matter. But the day I left.
Because today is not just another Saturday. It's the weekend.
Hmm. I had to take a shower to get rid of the stench from writing these lines. Bururu.
Because that is What does AI sentences sound like?. Once you recognize a pattern, you can't ignore it. And I don't know about you, but that's myself I feel sick!
To put it bluntly, I wrote the first few sentences in style A.I.. This is ironic, of course, because humans use that style, so does the AI (actually LLM).
That doesn't matter now, because anything creative the AI touches becomes inert and lifeless. Commenter hacker news do your best:
“I can’t help but think about the Midas-like nature of the LLM, because everything it touches is something that messes things up or makes people want to avoid it, like:
– Ghibli studio style graphics,
– Infamous em dashes and bullet points
– Customer service (make sure to use Klarnas “Support” these days…)
– Oracle stock price ;)- Imagine if your company, one of the most robust and unassailable technology companies in the world, succumbed to its CEO's insane dedication to LLM.
– Internet Content – We are now combing through all the Internet sources we don’t know about…
– And now chips?
Where does it stop? When do you decide to retire all technology as it is?”
No matter which side of the AI divide you are on, there is no denying that 2026 begins with slowing AI momentum and growing headwinds against AI.
As a fan of systems thinking, I like to imagine what could happen by studying feedback loops.
A feedback loop is simply a force that occurs when the output of one part of a system becomes an input to another part, thereby reinforcing (positive) or weakening (negative) existing dynamics.
A negative feedback loop is also called a balancing loop. This is because they weaken or balance the strength of existing loops.
Consider AI. Yes, it's the system itself. But it is also part of a larger, more complex system that includes countries, financial markets, investors, citizens, regulators, and economies.
AI momentum has slowed considerably compared to the same period last year. With a few exceptions, the new model is not significantly improved. Consumer enthusiasm has clearly waned. Companies aren't talking as excited about pilots and budgets. And even sophisticated investors are worried about the AI bubble bursting.
These are all feedback loops that work against the existing momentum of AI.
Many of you may know Gartner Hype Cycle:
The graph shows adoption. But if you want to look at it through the lens of momentum, I look at it like this.
Imagine the release of ChatGPT in 2022 as a “pebble” rolling off the top of a mountain. The pebble triggered an AI avalanche, increasing its mass and momentum.
But the terrain of 2026 will not be all downhill. It's getting flat. There are even more opposing forces.
Consumers are, at best, tired of AI, or in many cases bored with it.
“Why do Americans hate AI?”
More and more citizens are rising up against AI data centers.
“Big Tech's rapidly expanding data center plans face fierce community opposition.”
“As Google searches for land for its power-hungry data centers, Dalit farmers in Visakhapatnam find their fields becoming beacons for growth they may not survive.”
Even experienced and respected programmers are tired.
“Fuck you! The co-creator of the Go language is understandably furious at this thank you email.”
Savvy investors now feel confident enough to speak out against and bet against AI.
“The AI boom is in the early stages of a bubble, says Bridgewater founder Ray Dalio.''
“'Big Short' Investor Michael Varley Heaps Misery on Tech Stocks After Oracle Fails to Close AI Debt Deal”
The market too. Even as companies like OpenAI and Anthropic plan big IPOs.
“AI trading loses momentum as investors invest in broader S&P 500 stocks”
“Oracle stock falls as AI spending exceeds profits”
What AI breakthrough will break through these headwinds in 2026? New models? Better adoption? Dramatically more people paying for AI?
I'm not sure. I understand so far method In 2026, there will be more balancing loops for AI than reinforcement loops.
If you disagree, please write and let me know.
training data
(aka interesting links and reading material about AI)
Get ready for game slop. AI “World Model” is aiming.
Pay attention to what they do, not just what they say. Amazon and google We're fighting back against the AI.
Interesting explanation of why and how Nvidia acquires Groq (No, not the unhinged one).
indian Bandar Apna Dost (“Monkeys Are Our Friends”) is clearly the global king of YouTube slop. Annual income: $4.25 million. An estimated 21-33% of YouTube is currently bullshit or brain rot.
It turns out that optimizing to please humans has turned AI chatbots into “goofballs.” Similar feedback loops may be responsible make coding assistant worse?
Who needs the software that AI can generate to be carefully designed, stable, and well-maintained? disposable software In terms of scale?
What can the rest of the world learn from this path? Is China regulating AI?
a Great review of Claude Opus 4.5 Posted by a web developer (apparently a passenger on the AI hype train). I was really impressed.
AI labs, data centers, and hyperscalers want to grow faster than the existing power grid can handle. So they want to Generate your own (mostly dirty) power on-site.
zero shot community
“I think we can all agree that AI will have a huge impact (for better or worse) on our professional and personal lives, so I think it’s important to take the time to understand how the technology works, its limitations, potential impacts, etc. And what better way to do that than by reading a book? With that in mind, I picked up the following books. prediction machine. The authors describe AI as a predictive technology. For example, if you ask Alexa, “What is the capital of India?” Alexa will answer “Delhi,” but Alexa doesn't actually know the capital. Instead, we predict that when people ask such a question, the most expected answer is “Delhi.” This also explains why GenAI tools like ChatGPT sometimes make mistakes with basic math. They don't really understand addition and subtraction. What they do is predict what people are most likely to be looking for when they ask such questions.
Another interesting argument from the authors is that decision making involves prediction, judgment, action, and outcome. AI reduces predictive costs. Decreasing prediction costs affect the value of others, increasing the value of complements (data, judgments, actions) and decreasing the value of substitutes (human predictions).
This book is packed with insights, including when AI offers the greatest benefits (note that it is not without error), the risks it poses (bias, hallucinations, bad data), and the broader impact AI will have on society. Highly recommended for anyone interested in the impact of AI on business and strategy. ”
– Vishal Thibadewal (linkedin)
