ChatGPT: what a neural network can do for working with texts

[sumaiyakhatun27]

Kateryna Kushnir , editor, content marketer, author of the Telegram channel "Not Just an Editor" , played with ChatGPT . What can I say: it's too early to expect the disappearance of the profession of authors and editors, but some people should already be worried .

I generated texts on medical topics, questions for interviews with doctors, south africa rcs data and blog posts. In some places, the results are even frightening: the neural network writes better than many people, but it also has its weaknesses.

What artificial intelligence can't do
Correctly work with information in all cases. It must be said that it copes with many requests. For example, the answer to the question of what antioxidants are and whether it is necessary to take supplements with them, ChatGPT generated almost like a real author — everything strictly according to modern recommendations. It turned out better and more correct than 90% of articles that the search engine produces.

However, with ambiguous questions like whether wisdom teeth should be treated or should they always be removed, the algorithm does not cope well. Here, a person is needed who can analyze several sources, read research and opinions of authoritative organizations, and understand where the truth lies. Even the quality of the text in terms of readability is worse in such cases - apparently, it is difficult for AI to generate something understandable due to the lack of data on the Internet.

Take into account the context. The neural network cannot always take into account the characteristics of the audience, the platform where the text will be placed, and other similar nuances. However, for many issues, it generates readable and understandable text without factual errors, which can be modified for the desired situation.
Find interesting texture. Examples, details, cases, etc. Sometimes the tool does the job, but often they are not enough - the result is a dry presentation of facts. And, of course, the neural network will not be able to talk to an expert and add to the material the texture that is not on the Internet.
Use your own experience and expertise. A neural network does not have its own experience, but a person does. You can generate a post about how to come up with a title for an article — but it will not be as deep and useful as the text of a person who summarizes his experience, knowledge, and cases. The result is readable but superficial material — the tool is suitable for generating posts, however, they will have to be greatly refined.