OpenAI’s Greg Brockman Demonstrates GPT-4’s Capabilities in Developer Livestream
Over the past few weeks, there has been a lot of hype surrounding the next iteration of OpenAI’s language model, the GPT-4. People have been talking about how it would take AI’s capabilities to a whole new level, taking in and providing a level of intelligent automation never seen before (for consumers).
However, initially only paid ChatGPT Plus users will have access to the new features that come with GPT-4. These include the ability to edit, upload, and generate images on ChatGPT, and the standout feature, the ability to create AI videos from text.
OpenAI Showcases GPT-4’s Capabilities in Developer Livestream
OpenAI’s president and co-founder, Greg Brockman, has showcased the language model GPT-4 in a developer livestream, highlighting its strengths and weaknesses.
According to OpenAI, GPT-4 can respond using up to 25,000 words, which is eight times more than the free version of ChatGPT. This expanded word limit allows the chatbot to provide greater context in its responses.
Ability to Provide Image Contexts
In addition to this, GPT-4 can now also process image inputs, which is a significant upgrade from ChatGPT’s text-only capabilities.
In a demo during the developer livestream, OpenAI used GPT-4 to build a Discord bot that could handle both text and image inputs. While the bot could not handle the image input instantly and had some bugs, it was able to recognize the image and provide context.
This would surely be a major step up as the image processing functionality is not available in the current free version of ChatGPT.
Chat Completions Playground for Developers
GPT-4 will be rolled out soon to ChatGPT Plus users and developers using the ChatGPT API. Developers will have access to the Chat Completions Playground feature, which allows them to train and steer the GPT model towards their goals.
While GPT-4’s capabilities are still being developed, there is already talk of using the model to create a tax tool that allows people to handle complicated tax returns themselves, without the need for a professional.
In addition to this, OpenAI was able to use the model to build a website from a handwritten drawing, which could be a game-changer for those without the expertise to code their own websites.
GPT-4 would surely open doors for infinite possibilities and may as well change the way we interact with technology. The features shown off in the stream are really exciting, particularly the ability to handle images and create AI videos from text.
While we all expected more from the livestream, this is all that was showcased. We will make sure to keep you updated on all developments as new information becomes available.