Clarification of the thesis on where AI will not be able to replace humans, and where it will.

"Industries where uncertainty cannot be reduced by the volume of tests (money), but only by creativity - humans will remain. Everywhere else - AI will take jobs."

1. Example #1: Modern marketing. You can optimize advertising with a budget, buy all the traffic, and achieve maximum results. There is little creativity in this task, all processes are clear, and the methodology is also clear. Here, AI will eventually push humans out.

2. Example #2: Any kind of arbitrage - reselling goods, algo-trading, dropshipping. Regular, routine process with negligible predictable margin and high turnover. Trading bots have long been working in trading. Here too, AI will eventually push humans out.

3. Example #3: Writing business applications. Well, this has already been discussed.

4. Example #4: Disease diagnosis. Everything has long been formalized here, data collection and working with patterns. AI will eventually push humans out 100%.

Thus, any task where brute force plus data analysis can lead to optimal results - AI will push humans out everywhere. Transport, communication, operational management, logistics, warehouse accounting, education, psychological counseling, procurement, commodity sales, short-term forecasting, business application development, etc.

However, there are tasks where there is no methodology and an infinite number of options, an unlimited amount of money - cannot lead to anything.

Example #5: In creativity, there is no technology to create a guaranteed hit. All hits are anomalies and serendipity, including literature, cinema, music, games.

Example #6: Fundamental science.

In these areas, it seems that AI will not be able to replace humans for now, because although AI can reason, explain, and write code, its creative abilities are still zero. That is, all the pictures, texts, and code that AI creates are not creativity, but essentially imitation in a broad sense. However, a fundamental scientific problem cannot be created by imitation. A new scientific theory cannot be created by iterating over old ones.

And it's not even about the level of AI intelligence. Even if AI is 100 times smarter than a human, there is still no reliable method to create a hit. Not methodologically. Because a hit is not just an anomaly.

All hits have long been taken apart, studied under a microscope, all components and patterns are known, but there is a problem. Knowing all this still does not allow you to guarantee a hit or scientific discovery. That is, the sum of technologies does not lead to a breakthrough.

This is the case when

1=1,

1+1=1,

1+1+1=1, and the infinite sum of particulars still does not increase the result by a jot. Do you understand?

No physicist has yet been able to pass through a wall. Even a thin one made of rice paper.

This is the case about which one of the greats said (quote approximate, from memory):

"The trick is not to think about what everyone has seen but cannot explain. The trick is to think about what no one has thought about before."

And for now, AI cannot precisely do this: think about what no one has thought about. But a human can.

That's what I do every day, but for the most part, I don't write about it because people would think I'm crazy. I, of course, don't care what people think about me, it's just that there are things about which it makes no sense to speak publicly until a certain moment. Until you understand exactly yourself - why talk about it.

I talk about the most important things only with my wife. And no, we don't talk about marketing funnels.