The AI requires orders. On how it selects the right data.

The main problem. With the AI. It is that. It doesn’t distinguish between old and new sources. This causes a negative effect on its trustworthiness. These kinds of things. They can be fatal. If the AI uses old sources that don’t match the newest information. And one of the solutions. For this problem. Is that. The AI should have orders to use only the secured and trusted databases. 

This means that the person who gives orders can give an order. The AI to use. things like governmental homepages. In cases where AI must generate texts that should resonate with the law. In that case, the official homepages involve trusted information about laws. Another way to solve that problem is the interactive AI. 

1) The system can use the date as the priority. 

2)The system can introduce a source list and its dates to the user, who can accept or deny some of those sources. 

3) The system can introduce how those sources differ from each other. And then. The user can read them and accept or deny their use. 

4) The user can limit the sources that the AI uses only to trusted homepages. 

5) The user can give orders to use data that a certain system collects. The user can. As an example, give the main priority for information. That. The JWST telescope is collected. 

The system communicates with the user. It can give a list of the sources that it found. The system can search the data when the source is downloaded to the network. Then it can search for differences from those pages. Then it can tell the user. What type of differences does it find? And if there are some differences, the system can introduce those differences. In an easier way. It can introduce the dates when those sources are published. 

And. The user can allow the use of sources. Or deny their use for the text generation process. This method is easier. But it's less effective. If we take focus from error-detection. The third way is to order the AI to use a certain type of source. The order can be. The system uses mainly data. That's what the JWST telescope collects. This means that the date when the newest tool entered service can be the limit that the AI uses. Those things should guarantee that the AI uses the latest and most trusted data.