A new marketplace for bespoke AI-powered chatbots has been launched by OpenAI

Following last week’s hinting to developers, OpenAI has now opened a marketplace for GPTs, which are bespoke chatbot applications driven by its AI models that can generate text and images (e.g., DALL-E 3 and GPT-4).

Located in a new tab inside the web-based ChatGPT client, the so-called GPT Store showcases a variety of GPTs created by OpenAI’s collaborators and the broader development community. Popular and trending GPTs may be found on the community leaderboard, which is sorted into sections like lifestyle, writing, research, programming, and education, so GPT users can easily find what they’re looking for.

Users with a paid subscription to ChatGPT Plus, ChatGPT Enterprise, or the brand new ChatGPT Team from OpenAI are able to visit the GPT Store.

An AllTrails trail recommender, a Khan Academy coding teacher, and a Canva content creator are just a few of the GPTs that will be accessible at launch. Currently, you may use them all for free. (We will get into it more in the second part.)

Developers don’t need coding skills to build GPTs, and developers may make GPTs as basic or complicated as they like. Developers may use OpenAI’s GPT-building tool, GPT Builder, to try to create an AI-powered chatbot that can do the skills that developers specify in plain English.

One way to teach a GPT to answer queries about ingredients for a particular recipe is to use a library of cookbooks. On the other hand, a GPT may consume a company’s secret codebases, allowing programmers to verify their syntax or produce code that adheres to industry standards.

By submitting their GPTs to OpenAI’s new review system, which combines human and automatic assessment, developers may confirm that their GPTs adhere to the company’s conditions of use and have them listed in the GPT Store. (If any malicious GPTs manage to evade detection, users have the option to report them.)

I emailed OpenAI a number of questions regarding the “human” component of the GPT Store review process, such as the reviewers’ compensation and access to mental health resources, in light of the company’s history of hiring underpaid, overworked contractors in developing nations to moderate and enhance its GenAI systems. As of the time of writing, the firm has not answered; however, I will update this piece if they do react.

There will be no way for developers to monetize GPTs at launch. U.S. builders will be able to earn money from “user engagement” with their GPTs when OpenAI launches its “GPT builder revenue program” in the first quarter. OpenAI has promised to clarify the payment requirements in due course.

Although the GPT Store was announced last year at OpenAI’s first developer conference, DevDay, it was postponed until December. The leadership change that took place right after the announcement in November is most likely what caused this. (To cut a long tale short, OpenAI’s board of directors fired CEO Sam Altman, only to reinstate him with a new board after investor and staff fear.)

To a certain extent, GPTs make it easier for more people to build generative AI applications, at least for the models developed by OpenAI. The consulting firms whose core competency is creating GPTs for clients may very well go out of business if GPTs become widely used. That may not be a good thing in and of itself, but the outcome will determine that.