November 24, 2024

Testing with GPT-4: Update ChatGPT understands images, creates recipes, and tells jokes

2 min read
Testing with GPT-4: Update ChatGPT understands images, creates recipes, and tells jokes

ChatGPT’s new AI language model, GPT-4, was released on Tuesday (14). According to OpenAI, the tool is “more creative and collaborative than ever before,” with the ability to “solve challenging problems with greater precision.”

GPT-4 is currently available on ChatGPT Plus and Microsoft’s Bing search engine.

Those who don’t always follow chatbot news may not be aware of the changes in the update. So let’s list everything for you!

Read more:

More accurate answers

GPT-4 has seen a huge improvement in answering specific questions. As reported by The New York Times, researcher and AI professor Oren Etzioni asked the GPT-3.5 model “What is the relationship between Oren Etzioni and Elie Etzioni?”. The bot always answers wrong

By rolling out the GPT-4 model, the AI ​​got the right relationship between the two. paying off:

(Photo: Reproduction/The New York Times)

He can see and interpret pictures

One major novelty is its “multimedia” capability which allows understanding and interpretation of images in questions and interactions.

In the GPT-4 demo post, OpenAI presented an image with eggs, milk, and flour and asked, “What can I do with these ingredients?”

The robot interpreted the picture and replied: “There are many options for what you can do with these ingredients! Some possibilities include: – Pancakes or waffles – Crepes – French toast – Omelette or frittata – Quiche – Custard or pudding – Cake or cupcake – Muffin or bread – Biscuits Or biscuits. Those are just a few examples, but the possibilities are endless!”

(Photo: Clone/OpenAI)

OpenAI also provided an image and asked the bot to identify what was funny:

(Photo: Clone/OpenAI)

More creativity

OpenAI said it has made improvements to the advanced logic of the language model. The new technology can perform creative and technical writing tasks, compose music, write screenplays, and learn a user’s writing style.

In the demo, OpenAI asked a chatbot to create a Cinderella story in a sentence containing words starting with all letters of the alphabet but without repeating any.

Check the answer in English corresponding to the request:

(Image: Clone/ChatGPT)

Oren Etzioni, the aforementioned AI researcher, asked the robot to tell a joke about Madonna:

(Photo: Reproduction/The New York Times)

The best explanations

OpenAI has improved the concept of Steerability, which provides more options for personal interactions and adapts better to each user’s contexts.

During our audition, we asked ChatGPT 4 to play a Hollywood actor thanking the audience after winning the Academy Award for Best Actor. paying off:

ChatGPT using the GPT-4 language model. (Image: Clone/ChatGPT)

Compared to GPT-3.5, the new language model performs much better at interpreting:

ChatGPT using the GPT-3.5 language model. (Image: Clone/ChatGPT)

with information from New York times that it Open AI.

Featured image: Ebru-Omer / Shutterstock.com.

Have you seen the new videos on Youtube digital outlook? Subscribe in the channel!

Leave a Reply

Your email address will not be published. Required fields are marked *