Snapchat has announced that it is launching an OpenAI-powered chatbot similar to ChatGPT called “My AI”. The feature will be available initially for Snapchat+ subscribers for $3.99 per month. However, Snap plans to make it available to all users in the future. The chatbot will be available starting this week.
Snapchat’s CEO, Evan Spiegel, explained to The Verge that their goal is to integrate daily communication with artificial intelligence (AI) in addition to talking to friends and family. As a messaging service, they are in a good position to do this. The AI chatbot, called “My AI,” will be accessible in the chat section of the Snapchat app, where it will be pinned above conversations with friends.
Snap’s announcement states that the chatbot runs on the latest version of OpenAI’s GPT technology, which has been customized specifically for Snapchat. According to reports, this likely refers to OpenAI’s Foundry, an exclusive developer program available to wealthy developers.
Snapchat now lets you share youtube videos as a sticker on your snap
Foundry provides access to a more advanced model, GPT-3.5, which is the basis for the ChatGPT model. OpenAI’s publicly available API currently only supports up to GPT-3, which is an older and less intelligent model. We have reached out to Snap for clarification on the exact model used and will update this article if we receive a response.
Snap’s chatbot will have certain restrictions to ensure that it adheres to the platform’s trust and safety guidelines. To avoid issues seen in previous AI experiments, such as CNET’s AI-written articles and the AI Seinfeld experiment, Snap’s chatbot, called “My AI,” will avoid swearing, violence, sexually explicit content, and political opinions.
The company plans to continue to improve the model based on user feedback and reports of inaccurate or inappropriate answers. Users can submit feedback by holding down on a troublesome message. However, even with these protections, Snap’s chatbot may still encounter issues with misinformation and offensive content. The company acknowledged this in its announcement and cautioned that, while My AI is designed to:
- avoid biased
- or misleading information, mistakes can occur.