OpenAI has announced the long-awaited GPT-4. This is a follow up to GPT-3.5 which powers the very popular ChatGPT platform. The company says GPT-4 is their “most advanced system, producing safer and more useful responses.” Apparently, even Microsoft’s new Bing has been powered by GPT-4 all this while.
Earlier in the year, OpenAI CEO had said they hype around GPT-4 was quite high, and that people were begging to be disappointed once GPT-4 was announced. Well, it is now live. And it is impressive and also quite okay/ordinary depending on what expectations you had in place.
The new GPT-4 is available for users on ChatGPT Plus, and with OpenAI’s API through a waitlist. In countries like Kenya, where ChatGPT Plus is available for $20 a month, you can jump on the bandwagon to try out the new generator. If not, you can test it out on the new Microsoft Bing.
What’s new with GPT-4:
- Text plus Images
- Human level performance on both academic tests
- More reliable and factual
- More creative
- OpenAI is partnering with more companies
Text Plus Images on GPT-4:
This is a big leap given GPT-3.5 was text only. The new generator understands both text and images, but the image functionality is currently locked to a single partner – Be My Eyes, a Danish Startup that connects people who are blind or have low vision with volunteers for help with hundreds of daily life tasks like identifying a product or navigating an airport.
Be My Eyes is using OpenAI’s new model’s visual input capability to develop an app that can generate the same level of context and understanding as a human volunteer would.
GPT-4 can now caption, and even offer interpretation to relatively complex images.
Human level performance and reliability :
OpenAI says “GPT-4 outperforms ChatGPT by scoring in higher approximate percentiles among test-takers.” The company says they spent 6 months making GPT-4 safer and more aligned. It will now answer with better facts, more reliable information, and is 82% less likely to respond to requests for disallowed content.
GPT-4 is also more creative. It can not only understand input better, but it can also do more. OpenAI says, it can generate, edit, and iterate with users on creative and technical writing tasks, such as composing songs, writing screenplays, or learning a user’s writing style. So if you’re say working on something, you can give it your writing style to enable it come up with stuff for you.
The company is bold that the new GPT-4 surpasses ChatGPT in its advanced reasoning capabilities, thanks to feedback and research from GPT, GPT-2, and GPT-3.
We’ve already explained the partnership with Be My Eyes. Here are other companies OpenAI is partnering with to bring GPT-4 to more people:
- Duolingo – Where the new model helps deepen conversations with two new features: Role Play – to help with conversations, and Explain my Answer – which helps users understand the mistakes they’ve made grammatically.
- Khan Academy – where GPT-4 is helping each student have a customised tutor for their courses thanks to Khanmigo, an AI-powered assistant that functions as both a virtual tutor for students and a classroom assistant for teachers.
- Iceland Government – where OpenAI will help Iceland use GPT-4 in the preservation effort of the Icelandic language.
- Morgan Stanley – where the new model will power an internal-facing chatbot that performs a comprehensive search of wealth management content and “effectively unlocks the cumulative knowledge of Morgan Stanley Wealth Management.“
- Stripe – where GPT-4 will help streamline user experience and combat fraud.