Snap first launched its My AI chatbot last year, using generative AI models from OpenAI, the maker of ChatGPT. But the company recently started using Gemini as well because Google’s large language model can process video, audio and text simultaneously, capabilities collectively known as multimodal AI.
That feature is important because Snapchat users often communicate through videos and images, in addition to text, said Snap chief executive officer Evan Spiegel.
“The chat with large language models has always been quite engaging,” Spiegel said, though he also noted that the experience when sending photos and videos to a chatbot has lagged behind. Since sending photos and videos is the main way Snapchat users communicate, this “held back our vision” for what the product could ultimately become, he added.