Clickbait title, however, it points to the real problem - instead of being an intern, it becomes a boss who offloads findings on the real intern in this situation - the user.
My idea is:
- Intern-like features available from the device level
- Those intern-like features should involve creating a project with a plan for its realisation.
Here is how I see it.
User input is: [do something complex, i.e,] I want to learn Java [could be: wake up early in the morning, lose weight, etc.]
Rabbit: I can create a plan to take you there. Do you want to proceed?
User: yes
Rabbit: I will assess variables to take into consideration [your programming skill level, daily routine, or weight and health condition] (then he asks relevant questions that are put into memory for later)
Rabbit: When you are available for this to happen
User: I have some free time daily for about an hour, usually around 6 PM, and on weekends, I can spend longer, like two hours, at 4 PM.
Rabbit: Here is your plan…
[Creates in memory a structure of that plan, so main topics, after that, for each topic decomposes it into smaller topics and smaller until there are no candidates to decompose to make it reasonable to decompose. It’s like a multi-level list, but when you ask AI to do it, it will go shortcut; therefore, the list of candidates should be asked and expanded one by one.
Every one of those topics is like a TODO task; however, Rabbit, at the given time, i.e., 6 PM, will prompt the user like an alarm clock to complete the task. By the alarm clock, I mean notifications about it will appear in engag way every 4 or 9 minutes]
Rabbit: You wanted to learn Java. The first topic according to the plan is …
Tell me when you are ready
[9 minutes pass]
Rabbit: You wanted to learn Java. The first topic according to the plan is …
Tell me when you are ready
User: I’m ready
Rabbit explains the chunk to learn [that’s why decomposition should be so detailed that basic topics could be explained in a couple of sentences]
Rabbit: Time to put that to work. Do… [activity for a user, i.e. do exercises if it’s a weight loss plan, write some simple code if it’s about learning Java, make a call or write a paragraph in a document if it is some business project.
Rabbit: Are we there yet?
User: done
or if 9 min passes
Rabbit: You wanted to learn Java. The first topic according to the plan is …
Tell me when you are ready
…
Rabbit: Do you want to explain something in more detail, to you, or do you need to do some more exercise, or do we tick off this topic as done?
User: I’m good
Rabbit: Congratulations you completed task
Rabbit: You wanted to learn Java. The first topic according to the plan is …
Tell me when you are ready
…
or when the given time block exceeds
Rabbit: Good work [or You missed today’s practice relating to …] Today you have completed … see you [next appointed time for this project]
My thoughts on it:
That’s the idea. The rabbit becomes, in this way, a real assistant, not my boss. You can test it by a simple task: pretend I’m the kid and use a project feature to help me get through the process of making coffee for a parent, for instance, for Mother’s Day, or sandwiches, or cake, and I want to start it now.
Implementation:
Some of the implementation is strictly rigid programming, but part of it involves AI. As we do not have general AI, mixing both in this way as an internal rabbit app makes it reasonable.