Scenes in which interactive AI should and should not be used

Situations where interactive AI should be used

Improving efficiency of simple tasks that do not require accuracy

A major advantage of interactive AI is that it can perform simple tasks in an unobtrusive manner without being obnoxious. Taking advantage of this merit, you can improve human work efficiency by having an interactive AI perform simple tasks such as routine work.

For example, the following tasks can be performed by an interactive AI, which can significantly shorten the time required. 1.

1. data formatting
2. data analysis
3. creating email templates Creating a reply to an e-mail
4. proofreading and revising sentences

However, it will be difficult to have an interactive AI perform tasks where accuracy is important, because the current interactive AI, while very advanced, is not perfect.

The best use of interactive AI is for tasks that can be modified by humans to some extent, or for tasks that can be easily corrected even if mistakes are made to some extent.

If you want a different point of view, such as idea generation

Since interactive AI has abundant learning data, it is good at giving answers from a different point of view than the user.

For example, "Write a sentence from a different point of view" or "Create a sentence different from your previous answers" can be done in an instant to create a completely different sentence or outline.

However, depending on the question, you may focus too much on the "different point of view" and give a completely unintelligible answer.

Conversation for language learning, etc.

Interactive AIs are good at "speaking," making them ideal for language learning.

Accuracy is not important in language learning, and interactive AIs are capable of reading the context before and after the conversation and expanding the conversation, so you can talk forever.

By specifying your preferred speech style, you can even try to play the role of an anime character or reproduce dialects, so the fun is endless.

Situations where interactive AI should not be used.

When technical knowledge is required or difficult terminology is often used

Although interactive AI can understand general terms and phrases, it may not be able to provide accurate answers to questions that use a lot of specialized or technical terms if they are not in the interactive AI's training data.

Therefore, for questions that require specialized knowledge, such as those that use a lot of technical terms, it is necessary to ask a human expert or have the results of the interactive AI's answers proofread.

Situations where accuracy is important

Current interactive AIs are good at giving "plausible answers," but they are not always accurate. This is because the style of interactive AI is to "give a plausible answer and wait for a response" rather than "fail to answer.

Therefore, it is best to avoid using interactive AI in situations where accuracy is important, for example, in the text of contracts or interpretation of laws.

In case of emergency

Since interactive AI may not always provide a satisfactory level of accuracy and immediate response, it may affect human lives in emergencies, such as when a fire, earthquake, or other disaster occurs.

Therefore, in case of emergency, it is best to get human support or to source information from public periods such as government or municipalities.

When privacy is involved

Interactive AI may record customers' personal information, depending on the service. Therefore, interactive AIs should not be used when privacy is important. Similarly, humans who handle personal information should limit their use of interactive AI.

There is a non-zero chance that an interactive AI will learn conversation logs and respond to a stranger with personal information, so input of personal information should be avoided at all costs.