Does AI Need a Brain or a Body?
Thoughts after researching Echo, Google Home, and Stack-chan
I realized something while researching Google Home last time.
What I wanted wasn't the new smart device itself, but an AI that understands my goals and works with me to achieve them.
So, where should that AI be?
Inside a smart speaker?
Inside a small robot?
Or perhaps it should reside on a PC, to be called upon from another device only when needed?
This time, I will think about that 'AI brain and body'.
I already have an Amazon Echo in my home.
Turning on the lights.
Operating the air conditioner.
Setting a timer.
Playing music.
For controlling things around the house, the Echo is more than enough.
In other words, the Echo is sufficiently convenient as a 'voice portal for operating the home'.
On the other hand, what caught my attention while researching Google Home was not home appliance control.
If conversations with AI become natural, and it can be used as a partner for organizing information or consulting, wouldn't there be a point in placing it in my room?
That was the perspective.
However, I thought about this a little.
If what I really want to do is 'autonomous AI design' or 'long-term task management', is a smart speaker alone enough?
For example, this is the kind of AI I want.
It remembers where we left off yesterday.
It organizes the tasks I am currently working on.
It suggests priorities for me.
It thinks together with me about what I should research next to build an autonomous AI.
If necessary, it breaks down how to achieve it.
When I think about this, just being able to converse is not enough.
It needs memory.
It also needs the ability to see the entire workflow.
It needs a mechanism to make decisions based on long-term goals, not just immediate responses.
That is why I became interested in small robots like Stack-chan.
It shows expressions on its screen.
It reacts when you talk to it.
It moves its head.
It feels like it is really there.
This is different from a smart speaker.
Just by making it look like the AI has a body, the relationship changes.
It has the potential to feel like a partner on your desk rather than just a voice assistant.
On the other hand, it seems difficult for Stack-chan to do everything on its own.
Since it is a small device, it is more natural to think of it as an entry point for conversation or a way to display expressions, rather than leaving all the heavy processing to it.
Similarly, one could consider a DIY terminal using a Raspberry Pi and a small display.
Displaying a screen.
Listening with a microphone.
Speaking through a speaker.
Attaching a camera if necessary.
You can adjust the appearance and movement yourself.
It is easier to understand such terminals not as the AI itself, but as a container for connecting the AI to humans.
So, where should the brain be placed?
Thinking about my current environment, I felt that using a new PC for AI as the brain would be the most natural approach.
Heavy processing is performed on the PC.
Memory is also managed on the PC side.
Designing the autonomous AI and prioritizing tasks are also done on the PC side.
On top of that, terminals like Echo, Google Home, M5Stack-chan, and small displays serve as gateways to converse with that AI.
In other words, it means not trapping the AI's core inside a device, but thinking of the brain and body separately.
The brain is the place for thinking.
The body is the place for interacting with humans.
By separating them like that, the confusion about which device to buy can be somewhat cleared up.
Echo is a gateway for controlling the home.
Google Home is a gateway for conversation.
M5Stack-chan is a companion that provides a sense of presence.
Raspberry Pi and small displays are terminals you can build yourself.
And a new PC for AI is a place to nurture a brain.
Thinking about it this way, perhaps there is no need to complete the ideal AI terminal all at once.
First, nurture the brain.
Next, test which interface feels natural to talk to.
Finally, choose a body that suits you.
That order seems to suit me better right now.
When we talk about creating an AI personality, we tend to think about the character's appearance and way of speaking first.
But I think what is truly important, before the appearance, is what that AI understands, what it remembers, and what it will help you with.
The body can be changed later.
But the design of the brain is a process of accumulation.
Therefore, what I should think about next is not which terminal to buy, but what to teach this AI and how much to entrust to it.
I think I will start from there.