OpenAI’s sensational AI-powered chatbot, ChatGPT, has been creating waves on the internet ever since its launch in November 2022. Its super abilities have set off an AI race among the biggest tech companies in the world. Even as tech giants scramble to create the most innovative AI technology, developers around the world are exploring new ways to use ChatGPT, which is currently powered by OpenAI’s highly potent large language model, GPT-4. In March, Sam Altman’s company announced that it was endowing ChatGPT with the ability to browse the internet with the help of plugins. Plugins seem to have opened a floodgate of possibilities for millions of developers, and the chatbot has witnessed a flurry of new plugins in the last few weeks. Recently, OpenAI announced its latest plugin, the ChatGPT code interpreter, along with an array of other plugins. What is the ChatGPT code interpreter? In essence, the ChatGPT code interpreter plugin comes with a range of functions. The plugin offers ChatGPT a working Python interpreter in a sandboxed environment. According to OpenAI, the plugin can effectively solve mathematical problems, both qualitative and quantitative, perform data analysis and data visualisation, and convert files across formats. While ChatGPT is known to produce codes from inputs in natural language, this plugin enhances that ability with an altered format of response. The new plugin allows ChatGPT to use Python and offer results inside a dialogue box. “We provide our models with a working Python interpreter in a sandboxed, firewalled execution environment, along with some ephemeral disk space. Code run by our interpreter plugin is evaluated in a persistent session that is alive for the duration of a chat conversation (with an upper-bound timeout), and subsequent calls can build on top of each other. We support uploading files to the current conversation workspace and downloading the results of your work,” reads OpenAI’s blog demonstrating the plugin. The demo showed ChatGPT doing a range of tasks. For instance, it showed the uploading of large data sets, and the chatbot disseminating to identify trends and comparing variables. What can a ChatGPT code interpreter do? Ever since the plugin came to the fore, there have been numerous use cases discussed online. The code interpreter can allow ChatGPT to do basic video editing and convert GIFs into longer MP4 videos into slow zoom. Another interesting feature is that uploading location data (i.e., CSV file of every lighthouse location in the US) creates a visualised map. Ethan Mollick, a Professor at The Wharton School, used a prompt - “Create a gif of a map of the lighthouse locations, where the map is very dark but each lighthouse twinkles." A few seconds later, it showed a map that had illuminated dots showing the lighthouses. He also did a similar experiment on lighthouses in Europe. The Code Interpreter plugin is proving to be an incredible tool for its file uploading and downloading capabilities. According to designer Pietro Schirano, the plugin also extracts colour from an image to create a palette.png, and it also automatically compresses large images when it runs out of memory. Another user on Twitter also shared that the code interpreter can analyse the Bitcoin dataset that he had downloaded from his iPhone. As if its game-changing data analysis wasn’t enough, this plugin can also turn images into text files and render animated GIFs from scratch within seconds. The Code interpreter is also capable of analysing a user’s music taste from their Spotify playlist. Based on all that is being done with ChatGPT’s Code Interpreter, it can be said that the plugin is a boon to developers and all those testing new use cases of the AI chatbot. The biggest advantage of the plugin is perhaps it can make entirely new workflows effortless and efficient. So far OpenAI has not announced a date for the wider rollout of ChatGPT plugins, but it seems that it will happen sooner.
