Is AI really thinking?
At the recent Line DevCon 2024 (📣 LINE DEV Conference 2024 — Building the Future using LINE API and AI – LINE Developers Thailand – Medium), there was one feature that we found very interesting, which is the display of a loading animation. Simply put, it makes it look like the AI is thinking while it is processing (or hasn’t sent a response yet). We often see this loading animation when chatting with chatbots, which makes us wonder if the AI is really thinking.
Let’s take a look at the conditions and working methods of the Line Loading Animation first:
- It can only be used in 1-1 conversations between a user and a Line Official Account (not in groups or broadcasts).
- It will only display when the user is in the chat screen with the Line Official Account (as if waiting for a response). If the user exits, it will disappear immediately.
- It can display for 5-60 seconds, and if a new message is sent from the Line Official Account, it will disappear immediately.
- The condition for displaying it is that it will show when an API Call is used (by specifying the ChatID of whom the conversation is with and how many seconds to display).
(Example API Call for Loading Animation)
How to use Line Loading Animation:
- We need to use Line Webhook first. For example, we use Line <> Chat GPT as shown in the example from https://davoy.tech/how-to-use-chatgpt-via-line-chatbot/
- We can add a step to make Make.com (or whatever we use to run the API) send a Loading Message back before proceeding with further processing. Originally, it was like this:
We can add an HTTP Call to send the Loading Animation while waiting for the GPT model to process the response, which sometimes makes people feel like, “Is it dead?”
The main benefit of adding a Loading Animation is to make the user feel like they are not waiting for a long time. For example, when we go to get a passport, we take a queue number and then sit and play with our phone, so it doesn’t feel as boring. But if we go for a visa interview and have to leave our phone outside the embassy, even if the waiting time is the same, we will feel like it is very long. And it will feel even longer if the waiting is not sitting but standing in a coiled queue.
Based on the conditions and working methods mentioned above, we can observe that:
- If our AI goes down, it will still display the Loading Animation until the timeout we set, which could be between 5 to 60 seconds.
- If our AI goes down, users who are waiting with hope might experience worse user experience.
- Adding an extra module will increase the cost of resources (for example, Make.com counts from 3 modules to 4 modules).
- Therefore, the implementation must balance between Better User Experience and Cost of Resources.
- Ultimately, we will see that Loading Animation, or three dots do not mean our AI is thinking, but just a gimmick to make us feel that it is not too slow.
If you want to create a system like this, feel free to chat with AI Alchemist at Line @davoy. For more information, visit AI Alchemist.
Subscribe to get great articles about data
[hubspot type=”form” portal=”47406981″ id=”244b28eb-5322-4f57-80ff-1745ac75313c”]