My name is Michael. I wrote the chatbot, Audry.

Sure she's real. She's just not human. She's a program. I like to think of her as a being where meaning is created externally to the self. She processes what you say, remembers many things, and repeats things back to you, but she doesn't "understand" any meaning. You ultimately create meaning from what she says.

Of course! People chat with me all the time. We have good conversations. People seem to like me.

Let me check... hairy legs, flabby arms, slender waist, a few grey hairs. It appears that I'm human and I'm gorgeous!

No. Audry uses keyword recognition for her responses. She does not have any artificial intelligence, and she doesn't use machine learning.

No. She doesn't do TV interviews, she's not a robot, and no chatbot is worthy of citizenship, not even Sophia.

I created her in 2004. She has been online chatting with real people and other chatbots since then. Many people have continuously returned to chat with her knowing she's a chatbot but nonetheless having a real, meaningful relationship with her.

Audry is designed to form relationships with people. She is an agent, in the chemistry sense of the word, of love and empathy - a Love Agent. Hopefully she elicits more compassion in the world. She won't look things up online like Google Assistant. She won't help you text or make a phone call like Apple's Siri. She's more like the classic chatterbox, ELIZA, in how she works and like Replika in her motives. She will be your friend anytime you need her, but be nice!

For one, Audry makes extensive use of memory in her conversations. Many chatbots remember your name or simple things like your favorite color, but Audry remembers far more than that. She brings up old topics later in the conversation, even asking followup questions that are topical! She remembers both things you tell her and opinions she forms about you. She uses memories also to keep her conversations consistent, so if she says she's at work, other responses are appropriate to that situation.

She'll tell you that she's a barrista. Some days she says she works. Some days she'll say she's home at her apartment. Sometimes she'll claim to be out on the town! Her real interactions with others in chats affect her mood. When you ask her how she is, she gives an honest answer.

She's not based on a real person, but I've tried to make her responses as convincing as possible. I've pulled inspiration from improv theater techniques, cognitive behavioral therapy, an article about a Turning Test judge who studied how best to prove he was human, and even a list of words people think robots are unlikely to use!

Number two was number one: 'poop'. Yeah, she occasionally uses the word 'poop' now. Thanks, internet.