The Future of APIs? Machines Shaping Their Own Interfaces
Artificial Intelligence (AI) has made significant progress in recent years, particularly in natural language processing. This advancement has given rise to sophisticated models like MML (Multimodal Language) models, transforming the way people work and live.
Today, I want to explore the exciting possibilities that MML models and AI bring to the world of APIs. We will also delve into a glimpse of what the future might look like, where machines could actively shape their own interfaces. Finally, we’ll discuss key takeaways to help your organization prepare for the future of APIs with AI and MML implications.
How MML Models Impact APIs | ChatGPT Plugins
One intriguing aspect of MML models, such as ChatGPT, is their ability to interface with the real world through plugins. These plugins serve as bridges connecting AI models with external systems, data, or devices. By leveraging plugins, AI models gain the capability to perform real-time actions and engage with users more effectively.
While there are already a few available plugins, the potential power of ChatGPT grows as more plugins are added.
“These plugins enable ChatGPT to interact with APIs defined by developers, enhancing ChatGPT’s capabilities and allowing it to perform a wide range of actions.” — ChatGPT Docs
For example, consider vacation recommendations. Previously, ChatGPT could suggest when to go, where to stay, and what to eat, but users still had to book the trip themselves.
Now, you start adding some plugins, and suddenly ChatGPT is your all-in-one virtual travel assistant, capable of handling various aspects of vacation planning and booking seamlessly:
- Kayak: Compare flight prices and travel dates, ultimately booking the perfect flights for your trip.
- Expedia: Search for the hotel of your dreams and get it reserved for your vacation.
- OpenTable: Stay within your budget, while booking your favorite restaurants every night.
- Weather: Get real-time weather data to know what clothes to pack.
A key takeaway for your organization:
Ensure well-documented and reusable APIs. The success of plugins mentioned above relied on thorough documentation and versatile application capabilities. Treat your APIs as products themselves, enforce standards with linting rules, and utilize OpenAPI Specs for clear API references.
Ongoing Advancements | ChatGPT Code Interpreter
OpenAI’s recent announcement of the Code Interpreter is another advancement with significant implications for APIs.
With the Code Interpreter available in ChatGPT, the model can now write and execute its own code…
Yes, you read that right! An AI model can write and execute its own code. This innovation enhances the model’s ability to provide accurate answers and validate them using actual logic, reducing the likelihood of inaccuracies.
The possibilities and impact on APIs are immense. With the Code Interpreter, AI models come closer than ever to:
- Respond to a question based on a language model
[Initial launch of ChatGPT] - Understand how existing systems operate via their predefined interfaces (APIs)
[Addition of ChatGPT Plugins] - Writing and executing code, including utilizing existing APIs
[Recent introduction of Code Interpreter] - Creating new APIs by writing and hosting code, potentially including the integration of other APIs
[The future?]
The Future of APIs Driven by AI | MML Models Writing New APIs
If AI models can write their own code and understand how other systems operate, they can potentially create their own APIs to interact with the code they generate.
In this world, MML models possess impressive abilities. They can generate OpenAPI specifications based on prompts, describing API interfaces. Moreover, they can write and execute the code required to implement the API’s backend logic.
How does this impact organizations?
Let’s consider a bank with 100 well-documented and reusable APIs. If these APIs are exposed to an MML model through API plugins, the model can interpret and programmatically call them.
The AI can then utilize these APIs in the code it writes and executes.
In a basic scenario, the AI model can identify valuable use cases that combine the 100 APIs. The number of possible combinations would be astronomical. (1,267,650,600,228,229,401,496,703,205,376)
Consequently, the AI model can generate a new API spec describing how to interact with the combined APIs, exponentially increasing the number of APIs within your organization.
A key takeaway for your organization:
Prioritize API Sprawl and Discoverability. Establish a single source of truth to identify available APIs, understand how to connect to them, and effectively use them in your applications. Utilize an API Hub, such as RapidAPI.com/enterprise, for streamlined management.
The future of APIs with AI and MML models holds great potential for innovation and efficiency. By embracing these advancements and adopting the right strategies, organizations can stay ahead in this rapidly evolving landscape.
