ChatGPT encourages users to call GPTs, third-party apps powered by OpenAI’s AI models, in any chat, promoting their utilization. Paying ChatGPT users can include GPTs in conversations by entering ‘@’ and choosing from a list in OpenAI’s chatbot interface. The chosen GPT understands the discussion, and users can tag in multiple GPTs for various purposes, entering with prior information.
OpenAI tweeted, “This allows you to add relevant GPTs with the full context of the conversation.“
In the future, OpenAI intends to offer revenue for developers who would like to charge for users’ access to their GPTs. However, the business may need to increase traffic first. As of right now, custom GPTs account for only 2.7% of ChatGPT’s global web traffic, according to data from web analytics firm Similarweb. Since November, custom GPT traffic has been steadily decreasing month over month.
Another difficulty is turning out to be moderation. In its first week, the GPT Store was flooded with ‘romantic’ chatbot apps, some violating OpenAI’s service conditions with sexually suggestive content. Another issue arose with the rapid creation of political campaigning bots, like one impersonating US presidential candidate Dean Phillips.
OpenAI has since removed some of the offending apps. OpenAI promises to flag GPTs using a combination of human and machine evaluation. However, if the number of GPTs increases as the firm evidently hopes, the issue would likely only worsen.