ChatGPT knowledge cutoff has been extended beyond 2021. At its first developer event, OpenAI made some improvements to its AI program, including a more powerful language model called GPT-4 Turbo and the ability for anyone to develop custom chatbots. The company also updated ChatGPT’s knowledge base to include information up to April 2023 and added a new information retrieval function to improve the chatbot’s ability to talk about current topics. Here are the details…
ChatGPT knowledge cutoff: What is GPT-4 Turbo?
GPT-4 Turbo is a new native language model designed to meet the demands of developers and users. This model updates ChatGPT‘s knowledge base until April 2023. With the ability to process up to 128,000 tokens of information, GPT-4 Turbo can handle longer and more complex text than before.
Customizing chatbots with GPTs
In addition to GPT-4 Turbo, OpenAI is focusing on allowing anyone to create custom chatbot models known as “GPT.” These bots can be assigned to domains or tasks beyond the standard ChatGPT. For example, you could create a GPT focused on teaching computer science in schools or a GPT for consumers to create marketing logos and posters using a text command. GPTs can be used privately, shared publicly, or used only within a company.
To create private GPTs, OpenAI is offering a new GPT generator application. This effort builds on the plugin functionality the company introduced for ChatGPT in March. The goal is to offer GPTs in a digital store and share revenue with the developers who make them. The so-called “GPT Store” will launch later this month. According to the company’s blog post, users can create their GPTs today.
During the developer conference, CEO Sam Altman showed how users can create custom GPTs without programming skills. Users only need to write the requirements for the bot, and GPT will do its best to comply. “You can program a GPT with language just by talking to it,” Altman said.
Altman thinks that chatbots will one day become smart enough to act as “agents” that can perform any task a human requests. He believes GPTs and assistants are the precursors to the even more powerful AI programs the company is already working on. “Over time, GPTs and assistants, the precursors to agents, will be able to do much more,” Altman said. “Gradually, they will be able to plan and perform more complex actions on your behalf.”
OpenAI is committed to gradual, iterative deployment, ensuring that people can start building and using these agents now to get a feel for what the world will be like when they become more capable. With these improvements, OpenAI is taking another step towards creating a smarter and personalized user AI experience.
Featured image credit: Jonathan Kemper / Unsplash