Artificial Intelligence + Artificial Emotions
May 17, 2008 in Computers, Programming, Python, news
Back in 2003 I was fascinated by a story published in the Reader’s Digest book called “How did it Really Happen?”. It was of the Russian Grand Duchess, Anastasia Romanova. In short, princess Ana. When I researched on the Internet about who she was like, a lot of references came up, some documentation was available and a lot of fan sites were ready to tell how it must have been to know the young princess in person.
An idea took root - I was already creating small chat-bots, software which imitates human conversations, after learning a few tricks from ALICE. My addition to the trick was simple - instead of complex parsing mechanisms was using simple SQL statements to emulate the basic functioning of an AI chat-bot.
I wanted to go a step further and add emotion to my software; however artificial it may be. My attempts either failed or I gave up a bit too soon. This idea was a skeleton in my closet that just wouldn’t keep quiet. It was feeling sad that I had neglected it. Machines could be much more interesting if they had some emotions… I always wondered how it would feel to make a machine happy, angry or sad… and live with the consequences!
Fortunately inspiration came again when a bunch of Indian techies decided to gather up in an IRC channel to discuss things. We wanted a ‘bot’ to manage user permissions and to log parts of the conversation. That was fulfilled pretty soon, but I was getting the itch to write a bot that listened to people’s conversations and would chip in with helpful hints if someone asked a question that was frequently answered.
I made the software and was testing it… but I wasn’t happy… wanted to do something else with it. I realized that my old project was taking over me again. I decided to give in and challenged myself to start and create the first prototype of the software within 32 hours. Technically I failed again - When I was about 8 hours into the project, another project that I had worked on needed some urgent modifications. That resulted in 24 hours lost. Continued after that and managed to finish the project within next 24 hours.
Finally after almost five years of thinking about an idea, I had it in my hands! I was searching for a name fo rthe software, soemthing to identify it with. After going through baby-names lists, and checking username availability and trying to find the easiest name to spell and type, I finally settled for “Hema”, its an Indian name which means “golden”.
Now, Hema can understand when you are praising her, or when you call her stupid… and her mood changes accordingly. She even stops saying the cheerful “Hey!” if you say “Hi!” when she’s in a bad mood. Its fun to cheer her up and get her talking again.
I’m working on some documentation and introduction/demo that I can present to all of you. I currently have a lot of commitments that demand priority treatement. Like always, the pet project must wait a bit more.
Updates about Hema will be posted on this blog and on my twitter account, so if you are interested, watch these spaces.
Catch ya’ll soon…
