OpenAI introduced its GPT Store this week, a platform where paid users can discover, share, and potentially profit from specialized chatbots tailored for specific uses based on the company’s versatile large language model (LLM).

Deemed an “app store” for artificial intelligence, the GPT Store is a marketplace for custom chatbots dubbed Generative Pre-trained Transformers, or GPTs. Since OpenAI opened up the tools to let anyone build unique GPT chatbots two months ago, over 3 million of these have already been created.

Now, through the GPT Store portal, creators can choose to publicly list their customized AI-powered assistants for others to find, similar to how mobile apps are showcased to users.

How Do OpenAI’s Customized Chatbots Work?

Customized AI-powered assistants are built on top of large language models like OpenAI’s GPT-4. These foundation models are trained on huge datasets to understand natural language.

Developers can then take these base models and “fine-tune” them by providing more data, instructions, and examples for a specific domain or use case. This helps specialize the model, making them significantly better at certain tasks.

For example, one of the most popular custom GPTs at the time of writing, Consensus, is built to be an AI research assistant. It can search through 200 million academic papers to drastically reduce the time and energy costs that come with scientific research.

A bot could also be customized to have conversations about recipe ideas or to provide textbook study help. The fine-tuning makes them better at responding to those particular topics.

These customized bots are essentially AI-powered assistants that can have natural conversations and respond to text prompts associated with their specific area of expertise. Their responses continue to improve the more they are used and trained.

Practical Uses and Applications of Customized Chatbots

  • Help desks and customer support: Bots customized for different products/services can automate replies to common questions and issues.
  • Research and analytics assistants: Customized AI assistants can be fine-tuned to understand technical papers or data, and generate summaries that help researchers and data analysts automate some of their work.
  • Educational tutors: They can be used for tutoring students and reduce the need to expand teaching staff.
  • Creative aids: Bots trained in design, writing, coding, and other similar tasks can collaboratively exchange ideas with humans to increase their productivity and expand creative output.
  • Internal corporate tools: They can be tailored to improve business processes like data entry, document review, and e-mail automation.

Also read: Bill Gates Says AI Personal Assistants Will Wipe Out Amazon and Search Engines

Upon launch, OpenAI is highlighting select GPTs as featured picks spanning categories like research, programming, lifestyle, and content writing assistance.

New GPT store additions will rotate weekly to showcase novel and practical customizations that demonstrate the technology’s breadth.

Visitors can also browse popular bots identified on community leaderboards based on engagement and trending posts. Some current highly-shared GPTs cover niche interests from cocktail recipes and trailhead guides to textbook study aids.

Monetization System for Creators Coming Soon

While not immediately in place, OpenAI confirmed plans to let builders eventually profit when users interact with their published GPTs.

A specific business model and compensation structure based on engagement metrics will be detailed sometime in Q1 2023. Initially, the program will only apply to creators based in the United States.

“To start, US builders will receive payments based on how users engage with their GPTs,” the company stated in the official press release. They promised to share more details about this monetization program in a few months. “We will clarify the payment criteria as the launch date approaches,” the development team noted.

The postponed rollout comes on the heels of internal shakeups in November 2022 that saw the swift ouster and subsequent reinstatement of CEO Sam Altman by OpenAI’s board of directors. Insiders partially blamed GPT Store delays for the turmoil that ultimately caused the executive’s ousting.

OpenAI Implements Review System to Ensure Policy Compliance

With the platform now live, OpenAI said that a formal submission review process is in place to confirm that custom GPTs are meeting the company’s evolving usage guidelines and community standards before they are visible and accessible to the public.

Both automated filters and human evaluations will assess aspects like safety, quality, and adherence to norms as dictated by OpenAI’s content policy. Users can also flag concerning bot behaviors for follow-up oversight.

Also read: 100+ Artificial Intelligence Statistics You Need to Know

The review workflow intends to mimic how major app stores govern listed offerings to align with regulations and prevent abuse like scams and misinformation.

Since GPT personalization remains an exclusive perk for paying subscribers. Only ChatGPT Plus and Enterprise customers will have access to leverage and publish tailored GPTs from the store’s inventory.

On the same day as the GPT Store introduction, OpenAI announced ChatGPT Team – a new intermediate subscription plan between Plus and Enterprise targeted at smaller groups.

Team accounts provide collaborative benefits like shared bot development and deployment that keep custom models private within an organization. Pricing starts at $25 per assigned seat when billed monthly.

Customers have praised the impact of ChatGPT in increasing the efficacy of their internal systems. OpenAI showcased the success of its product via a couple of testimonials featured in the blog post that announced the launch of the Team tier.

“With ChatGPT Team, we’ve been able to pilot innovative GPTs that enhance our team’s productivity and collaboration,” commented Dr. John Brownstein, Chief Innovation Officer at the Boston Children’s Hospital.