Why does all AI-generated content sound the same

Why does all AI generated content sound the same.

The SEJ came out funny article «Why AI Content All Sounds the Same & How SEO Pros Can Fix It» — созвучно заголовку этого поста. Обязательно почитайте оный опус и поразитесь, насколько отстали в промптоведении наши западные коллеги.

Actually the whole cycle is shown on the screen and it does not even reach the initial blocks of our promptology.))))) That's good to see.

By the way, all sorts of LLMs have started to artificially degrade the quality of generation on free plans. All our mighty prompts do not work even one third of the time. Google AI Studio has become especially guilty of this. Well, either change LLM or take paid plans. Or you can strain the LLM itself.

For the same Google AI Studio, in which we shove our prompts into system instructions, and in the body of the chat only the initial data and generation results. Therefore, after generating the result, we write this prompt

Reread the results of your work. Now reread the system instructions. Now make a complete list of all the flaws and omissions you made while writing the slot, starting from unassembled entities, intents, key phrases, insufficiently elaborated overview taxonomy, not covering all entities, missing points in the system instructions and so on. Make as big a list as possible of all deficiencies and required fixes.

And after its execution, we start the generation again, taking into account the identified deficiencies.

Test it. You'll be amazed how much LLM underperforms and how it ignores half of the prompt, even despite the built-in check blocks.

Let's get the LLM to labor at 100% percent!