March 14 was a big day in the world of AI. First, Google announced it will introduce AI capabilities in Gmail, Google Docs and much more.
On the same day (and perhaps not coincidentally), OpenAI stole Google’s limelight by announcing that it is releasing its updated AI model, GPT-4.
It’s an improvement on ChatGPT’s model (GPT-3.5) that took the world by storm in the past few months, and is OpenAI’s most advanced model made available to the public.
What can you do with GPT-4?
Yesterday, OpenAI gave a demo on GPT-4’s capabilities. Some, but not all, of these features have already been made accessible to the public. Here’s what you can expect from the new model:
- It’s a lot smarter: GPT-4 scores in the top 90th percentile for the Uniform Bar Exam, while ChatGPT scored in the 10th percentile. That’s a considerable boost in IQ. This means the quality of its answers are likely to be much more accurate than ChatGPT’s (which often gives misleading and factually incorrect answers).
- It can handle bigger prompts: GPT-4 is able to handle over 25,000 words of text. That means you could ask it to summarize a small book. Alternatively, you could give it a link to a website and ask it to summarize all the text on that page. However, GPT-4 is not yet linked to the internet, so can’t summarize other website pages yet.
- It accepts images as inputs: In the GPT-4 demo, the presenter draws a rough sketch of a website he would like to build. He takes a picture of it, and asks GPT-4 to write the code for a website that looks like his sketch. As promised, GPT–4 writes the code for the website in seconds. Another interesting use case was taking a picture of some ingredients, and asking what you could cook with those ingredients. Unfortunately, this image recognition feature is not yet publicly available at the time of writing.
How to use GPT-4?
If you can’t wait to get your hands on the new model, then we have some good news. GPT-4 is available to all ChatGPT Plus members ($20/month) on ChatGPT right now. All you have to do is to use ChatGPT in the usual way, and make sure the GPT-4 model is selected. Although not all of its features are available, it is ready and waiting for you to test out some of its new capabilities.
For developers who want to use the GPT-4 API to build your own AI app, there is currently a waitlist. You can apply here.
With new capabilities, come new opportunities. To make sure you stay up to date with the latest tech advancements, and how to use them in your own life, subscribe to the WGMI newsletter. Every week, we discuss new AI tools and how to use them to boost productivity and make your first million.