Development of neural networks

Development of neural networks

If you delve even a little into neural networks for design, video, photo, automation. Subscribed to a dozen thematic channels and started following the news. Set up some agents to ease routine processes. You experience the same gambling addiction as stock traders and crypto enthusiasts. This is FOMO in its pure concentrated form. As if we missed everything and are not keeping up.

In reality, 99% of all the neurohype is outright slop and has no practical application. I really like how guys sit with shining eyes doing market analytics through a neural network or working out key conversion points on a site. Without immersion, without understanding the distribution model, without specific numbers. But beautifully, abundantly, visually, interactively. And pseudo-useful, in fact. But if it's for skill training — then please, no questions.

I am not against automation of routine, or coding. Debugging processes, working with arrays, developing parts where you understand what you are doing. Otherwise, it's not development, but intellectual impotence. You can't become a programmer or vibe coder if you don't understand a thing about it and have no clue about architecture. A neural network won't draw you a good and breakthrough design. A neural network won't write a selling text without your knowledge of audience triggers. It won't even compose a soul-touching song (and this is confirmed by all the crap that has filled playlists). You should do this, and the neural network is just an auxiliary tool.

Moreover, the Massachusetts Institute of Technology conducted an experiment and found that users of ChatGPT have 55% weaker brain connections compared to those who did not use it. And 83% of the ChatGPT user group could not quote a single line from their essay.

So we sleep peacefully. We are not missing anything critical. Let them develop by leaps and bounds. But for practical application of these technologies, experience and a human mind are still needed.