Many AI speakers are on the rise these days. The main function of the AI speaker is to convert the voice into data and the data can be used to execute the program.
A running program is used to get weather information or to play music. You don’t have to think of it as hard because AI speakers currently are not involved with artificial intelligence.
I would like to inform you the WIZnet products that can be played music or recognized the voice.

WIZwiki-W7500 WebRadio player (streaming music player)

This project describes that connect to the server and play the song by streaming. It is the same as the AI speaker function that is nowadays. The file extension can be up to midi, wav, mp3, aac.

FTP Music Box (WIZ MP3)

FTP Music Box connects to the FTP server and saves the song to the SD card. It plays a song from the stored on SD card and information of song appears on the LCD.

Christmas Station

Christmas Station works with Arduino Uno and Ethernet Shield. Carol song(Jingle bell) comes out with a buzzer, and Santa is dancing according to the sound, unlike other music playbacks.

The Chimaera looks like a guitar, but it makes a sound differently through the touch, and metal should be worn with your finger for the Chimaera to recognize. The DSP device is a Digital and Analog signal board that handles sensor reading, event detection, and host communication.
The advantage of AI is voice recognition.

Android voice commands to arduino
Nowadays, it is possible to show up weather information and play music with voice by mobile phone.
It is a project to demonstrate that it is possible to turn on and off the LED by using the mobile phone.
You can see that it is not difficult to implement the current AI speaker features.
How about making your own AI speakers?