Hugging Face, the deep learning-based chatbot acts primarily as a digital friend allowing people to text back and forth, trade selfies and jokes as if it were a real friend.
“There are many people working on artificial intelligence for productivity or utility applications,” co-founder and CEO Clement Delangue told TechCrunch. “We’re building an AI so that you’re having fun talking with it. When you’re chatting with it, you’re going to laugh and smile — it’s going to be entertaining.”
Using NVIDIA Tesla V100 GPUs on the Google Cloud and the cuDNN-accelerated PyTorch and TensorFlow deep learning frameworks, the team trained their system on multiple models that analyze sentiment, emotion, sarcasm, and image recognition.
The startup recently raised $4 million for building out the technology for their chatbot that receives around 1 million messages per day, and to this day over 100 million messages have been exchanged through the app.
According to the company, 820,000 of the 100 million messages are what they describe as “laughs” or users sending messages like “lol,” “haha,” etc. 455,000 are thank you notes, or people thanking the AI for cheering them up.
“This demonstrates that an AI can help people feel better, boost their confidence and make them a better version of themselves,” Delangue wrote in a medium post.
Hugging Face is available on the Apple App store, Kik, and Facebook Messenger.