How to reduce the “Spaminess” of text?

High “spam” of articles is a common problem for many AIs (including GPT). In ClipAI we have done our best to automatically eliminate this problem. However, this may sometimes not be enough.

If you encounter the above problem (both in ClipAI and in any other program), we recommend that you follow these steps, checking the result after completing each step:

1. Change the AI you use. For some neural networks, this problem occurs more often. For example, the Google Gemini 1.5 Pro is usually less spammy than the GPT-3.5.

2. Use connecting the neuron to the Internet. The network you are using may not have enough knowledge on some topics. Because of this, not only logical, but also direct repetitions occur in the text. When connected to the Internet, AI will know relevant information even on the most obscure topic.

3. Try reducing the text size. The logic is simple: less text means less repetition, even as a percentage. ClipAI automatically changes the number of subheadings in the article, queries, and even the logic of work depending on the text size. We also recommend limiting text size to characters rather than words.

4. Turn on “Priority on...” - “High text quality” in the “Options” tab. In this case, more information will be sent in the context of requests, and there will be much less repetition (both logical and direct) in the text. In third-party programs, this can be achieved by sending the context of previous requests and responses via the API. However, not all developers of other programs add this feature in principle.

5. Try a little change advanced options “Spammy text” and “Variety of text”. We recommend that you first try increasing “Text Variety” by 1 point -> check the result -> decreasing “Text Spaminess” by 1 point. If you have problems with the text (switching to other languages or the appearance of random characters), you should return the values of these parameters back to the “Default” position. In third-party programs, the closest (but not complete) analogue of these parameters is “Temperature” and “Frequency penalty”.

6. Change text style in the "Processing" tab. For example, Google Gemini often writes much better if you enable the First Person style.

7. If you are using keywords — check whether you indicate them correctly. Then try changing their number. In the “Processing” -> “Keyword Repetition Frequency” tab, select “Normal”.

8. Use the “Writing an Article (Auto Detect)” mode. And if you still use short article writing mode, or Custom template, RAW mode or similar - enable the “Check for matches when continuing text” functionality in the “Processing” tab.

9. Try changing the query input language in the “Processing” tab. In some cases this may help. This will not affect the language of the output text.