OpenAI’s latest feature in ChatGPT has brought AI personalities into a single chat, allowing for a level of teamwork among different AI agents. This feature, announced on Tuesday, enables users to bring in custom personalities called GPTs into any ChatGPT conversation by using the @ symbol. This development makes collaboration with a team of AI agents within OpenAI’s platform more practical and brings us one step closer to a future where GPTs can work together as a team.
Previously, if users wanted to share information between GPT profiles, they had to copy the text, select a new chat with the GPT, and explain the context of the information. However, with the latest feature, ChatGPT users can now stay within the default ChatGPT window and bring in GPTs as needed, without losing the conversation history.
OpenAI introduced GPTs in November to allow users to create custom personalities or roles for ChatGPT. For instance, users can develop their own GPTs to focus on specific topics or skills. Additionally, paid ChatGPT subscribers can freely download various GPTs developed by other ChatGPT users through the GPT Store.
To illustrate this feature, we created a Wellness Guide GPT, an expert in human health conditions, and a Canine Health Advisor for dog-related health questions. By simply using the @ symbol and typing the first few letters of the required GPT, users can easily switch between different GPTs in the same conversation and ask specific questions.
This new functionality opens up possibilities for additional scenarios where users can consult different GPTs within the same chat. One might ask an ad copywriter GPT to write some text and then invite an editor GPT to review it from a different perspective. Different system prompts that define a GPT’s personality contribute to significant behavior differences.
Interestingly, when switching between GPT profiles, ChatGPT does not differentiate between them as different personalities. It will still refer to previous GPTs’ output as if it were its own, indicating that from its point of view, it is still just ChatGPT and not multiple agents.
From a broader perspective, this feature represents a step toward a future where GPTs, as independent agents, can collaborate as a team to undertake more complex tasks as directed by the user. While similar experiments have been conducted outside of OpenAI by using API access, OpenAI has been cautious in implementing a more agentic model for ChatGPT. However, it seems OpenAI is gradually moving towards that goal, as observed with the introduction of GPTs and now with this latest feature. It remains to be seen when we will witness true agentic teamwork in a shipping service.
With OpenAI’s continuous development and improvement of ChatGPT, the potential for AI agents to work together and provide new solutions emerges. By bringing different AI personalities into a single chat, OpenAI aims to enhance user experience and expand the possibilities of AI collaboration in various fields.
