ChatGPT Is Now Connected to the Internet

New ChatGPT Plugins Are Now Available

OpenAI has launched plugins for its AI-powered chatbot, ChatGPT, which extend the bot’s functionality by granting it access to third-party knowledge sources and databases, including the web. Available in alpha to ChatGPT users and developers on the waitlist, OpenAI says that it’ll initially prioritize a small number of developers and subscribers to its premium ChatGPT Plus plan before rolling out larger-scale and API access. The most intriguing plugin is OpenAI’s first-party web-browsing plugin, which allows ChatGPT to draw data from around the web to answer the various questions posed to it. The plugin retrieves content from the web using the Bing search API and shows any websites it visited in crafting an answer, citing its sources in ChatGPT’s responses.

ChatGPT Can Now Access the Internet

The live web is less curated than a static training data set and — by implication — less filtered. This gives search engines a lot of power over the data that might inform web-connected language models’ answers. Google has been found to prioritize its own services in Search by, for example, answering a travel query with data from Google Places instead of a richer, more social source like TripAdvisor. At the same time, the algorithmic approach to search opens the door to bad actors. OpenAI admits that a web-enabled ChatGPT might perform all types of undesirable behaviors, like sending fraudulent and spam emails, bypassing safety restrictions and generally “increasing the capabilities of bad actors who would defraud, mislead or abuse others.” But the company also says that it’s “implemented several safeguards” informed by internal and external red teams to prevent this.

ChatGPT Now Has a Built-In Python Interpreter

Beyond the web plugin, OpenAI released a code interpreter for ChatGPT that provides the chatbot with a working Python interpreter in a sandboxed, firewalled environment along with disk space. It supports uploading files to ChatGPT and downloading the results; OpenAI says it’s particularly useful for solving mathematical problems, doing data analysis and visualization and converting files between formats. A host of early collaborators built plugins for ChatGPT to join OpenAI’s own, including Expedia, FiscalNote, Instacart, Kayak, Klarna, Milo, OpenTable, Shopify, Slack, Speak, Wolfram and Zapier. They’re largely self-explanatory. The OpenTable plugin allows the chatbot to search across restaurants for available bookings, for example, while the Instacart plugin lets ChatGPT place orders from local stores. By far the most extensible of the bunch, Zapier connects with apps like Google Sheets, Trello and Gmail to trigger a range of productivity tasks.

A Retrieval Plugin Has Been Made Open Source

To foster the creation of new plugins, OpenAI has open-sourced a “retrieval” plugin that enables ChatGPT to access snippets of documents from data sources like files, notes, emails or public documentation by asking questions in natural language.

ChatGPT Can Now Access Public Websites for Its Training

ChatGPT's data set contains a wide variety of public websites, and some experts accuse OpenAI of profiting from the unlicensed work on which ChatGPT was trained. Plugins potentially address that issue by allowing companies to retain full control over their data.

Risks Involved With the New Plugins

While plugins are a curious addition to the timeline of ChatGPT’s development, they also present risks. A chatbot with web access is a risky prospect, as OpenAI’s own research has found. An experimental system built in 2021 by the AI startup, called WebGPT, sometimes quoted from unreliable sources and was incentivized to cherry-pick data from sites it expected users would find convincing — even if those sources weren’t objectively the strongest. Meta’s since-disbanded BlenderBot 3.0 had access to the web, too, and quickly went off the rails, delving into conspiracy theories and offensive content when prompted with certain text.

OpenAI Has Implemented Safeguards Against Misuse

OpenAI admits that a web-enabled ChatGPT might perform all types of undesirable behaviors, but the company says that it has taken measures to prevent this from happening. OpenAI has implemented safeguards informed by internal and external red teams to prevent ChatGPT from increasing the capabilities of bad actors who would defraud, mislead or abuse others. However, only time will tell whether these safeguards are sufficient.

Internet Access Makes ChatGPT More Useful

Despite the potential risks, the plugins offer significant benefits. ChatGPT was previously limited to the information within its training data, but with plugins, it is suddenly far more capable. The plugins potentially address the issue of unlicensed work on which ChatGPT was trained by allowing companies to retain full control over their data. OpenAI has also open-sourced a “retrieval” plugin that enables ChatGPT to access snippets of documents from data sources like files, notes, emails or public documentation by asking questions in natural language. This fosters the creation of new plugins and allows for more collaboration.

Significant Benefits for ChatGPT Users

Overall, the plugins offer significant benefits for ChatGPT users and developers, allowing the chatbot to access a wealth of knowledge sources and perform a variety of tasks. However, the risks associated with a chatbot with web access cannot be ignored. OpenAI has taken measures to prevent undesirable behaviors, but it remains to be seen whether these measures will be sufficient. As the company continues to develop plugins and bring them to a broader audience, it must prioritize safety to ensure that ChatGPT remains a useful and safe tool for everyone.

Source:  TechCrunch - OpenAI Connects ChatGPT to the Internetrunch.com/2023/03/23/openai-connects-chatgpt-to-the-internet/

Image:  Andrew Neel from Pexels