ChatGPT's New Era: Live Web Data & Third Party Services from Plug-ins
Significant Implications for Large Language Models & How Users Interact with AI
The Gist
OpenAI introduced plug-ins for ChatGPT to access up-to-date information, run computations, or use third-party services.
For a simple name, plug-ins are a massive upgrade
Companies creating plug-ins include Expedia, FiscalNote, Instacart, KAYAK, Klarna, Milo, OpenTable, Shopify, Slack, Speak, Wolfram, and Zapier.
OpenAI is gradually rolling out plug-ins to study real-world use, impact, and safety/alignment challenges.
ChatGPT plug-ins help overcome the static nature of AI models by providing real-time information from curated sources.
More Detail
OpenAI has introduced the first plug-ins for ChatGPT, which can access real-time information, perform computations, and use third-party services. As David Sacks said, this is a pretty big deal.
Companies like Expedia, Instacart, KAYAK, OpenTable, and Shopify have already created plug-ins. These plug-ins help ChatGPT access live web data for the first time, expanding its capabilities beyond pre-trained information.
Users can easily upload plug-ins for various purposes, and we’ll use groceries as an example.
Imagine ordering groceries, but before ordering through Instacart, you ask OpenTable for restaurant recommendations. With those recommendations, you then ask ChatGPT for specific recipes at those restaurants, along with the total calorie count of each meal. Once you decide which meals you want, you then order those ingredients through the Instacart plug-in within ChatGPT.
Not only is this saving you a ton of steps bouncing around the web, the data is current, and the information is curated from various sources.
Previously, using the restaurant example, had you asked ChatGPT “what are the top Laotian restaurants in San Francisco?” ChatGPT might have suggested Hawker Fare, which a Michelin-star chef started in the Mission district.
Unfortunately, Hawker Fare closed earlier this year, but ChatGPT would not have known due to the stale data they were using. Now with OpenTable’s plug-in, ChatGPT will save users a wasted trip by telling them whether restaurants are still in business.
OpenAI plans to develop open standards for AI-facing interfaces, and seeks feedback from developers interested in collaborating.
Implications for Large Language Models
As mentioned earlier, plug-ins hold significant implications for the future of large language models (LLMs) and the way users interact with AI systems.
Providing real-time, up-to-date information
As the Hawker Fare example showed, one of the major limitations of LLMs like ChatGPT is their reliance on pre-trained information.
With plug-ins, ChatGPT can access live web data and provide users with more accurate and relevant responses based on current information, enhancing the user experience and utility of AI chatbots.
Expanding capabilities and use cases
Plug-ins can perform computations and integrate with third-party services, unlocking a vast range of potential applications. This could foster widespread adoption of AI-powered chatbots in various industries, such as customer support, personal assistants, content creation, and more.
Customizing and ease of use
Plug-ins help users customize and deploy ChatGPT for their specific needs. This lowers AI’s barrier to entry and helps more organizations and individuals adopt AI solutions.
Collaborating and providing open standards
Plug-ins pave the way for open standards in AI-facing interfaces, promoting collaboration between developers and enhancing the interoperability of AI systems. This enables faster innovation and more powerful AI applications.
By providing real-time information access, expanded capabilities, and ease of use, plug-ins can revolutionize the way users interact with AI chatbots, while accelerating the adoption of AI solutions across various industries.