She says that even if it is far from perfect (for making sometimes odd mistakes), it neverheless can correct your mistakes if asked. I found her statements pretty confusing. Firstly, as a general remark, how could I rely on its quality if it makes mistakes itself. For instance, once per 3 sentences.
Aye, (as Hamlet would say), there's the rub. It may be able to correct itself if asked, but as you point out, how does one know which parts of an answer to ask it to check?
It is nonsense as I could try and enter a text general by the AI and ask it to check its own result. Please let me know whether under such circustance, it makes a sentce to deal with ChatGPT?
That's long been a simple test for translation software - ask it to translate from one language to another, and then retranslate the translation back to the original language. How different the 2nd version is from the original version is a gauge of how effective the translation software is.
Please let me know whether under such circustance, it makes a sentce to deal with ChatGPT?
That's entirely up to you, but we've had similar discussions about this in the past. The consensus is generally 'don't use it for learning English (or any language, presumably)'. The problem with it correcting your mistakes is that depends how you phrase the questions.
See the first suggested thread below, where ChatGPT suggested that a particular sentence was 'less natural' when it was flat out ungrammatical. When the OP rephrased their question, it then stated it was ungrammatical.
AI at this point is still just reactive, not proactive. It can only respond to what you ask, not volunteer information or corrections. To me, that's just a crapshoot on how well you phrase your questions - something difficult enough for a native speaker, but compounded more when you're a language learner. Ask good questions, get good answers. Ask the wrong questions, and get wrong information. AI is also notorious for making up examples of something when it can't find any real-life examples in its database.
You see examples of this all the time on this forum. Somebody posts a question asking about X, but members chime in with information about how they're asking the wrong question, or point out additional problems to the original question itself. Take your post, for example - you had several typos, which emsr2d2 pointed out before addressing your actual question at hand. AI can't do that, at least yet.
My two cents is that AI is a potentially useful tool for both learners and educators, but like any tool it's only effective when used correctly. We're still learning the correct way to use the AI tools.
I mainly uses it to look up in synonyms and antonyms.
Also, if all you want is synonyms and antonyms, there are several decent online thesauruses. I like
thesaurus.com, as it lists both synonyms and antonyms, and it gives a brief definition, all in one click. I'm sure there are probably others, but that's my go-to.