As soon as GPT-4 was released, there was access to the mobile APP. You can upload photos and videos for one-click interpretation. It also served as a Q&A bot for the American version of Zhihu.

This new feature based on GPT-4 is called Virtual Volunteer and comes from the application Be My Eyes.

GPT-4 has just been released, and there is already a mobile application to access it!

Just upload the image and make a request using voice, GPT-4 can help the visually impaired "see clearly" the world in front of them.

Real-time interpretation anytime, anywhere, as natural as chatting.

For example, if you want to change your clothes, but don’t know the color of the clothes in your hand:

Just take a photo and upload it to GPT-4, it will be fast You can describe the texture of the clothes, and the material, color and shape are clear:

##△Translated by Youdao

in Previously, in addition to touching with their hands, visually impaired people often relied on the help of people around them or volunteers to identify objects.

Some netizens praised after seeing it: This is one of the most amazing GPT-4 applications I have ever seen!

So, what capabilities does GPT-4, which has added the image understanding function, show?

Built based on GPT-4 image reading ability

This new function based on GPT-4 is called

Virtual Volunteer, from the application Be My Eyes .

Currently, Virtual Volunteer connected to GPT-4 can help users complete functions including asking for directions, navigating, reading menus, and searching.

For example,

ask for directions.

You only need to take a photo of the location, and then ask GPT-4 how to get to the designated location through voice, and it will output a complete route map and "read" it to The user listens:

△Translation by Youdao

Another example is Reading Picture

Translation.

Just show GPT-4 a description of the item, or even just the shape of the object, and it will be able to search and output the function, specific instructions and usage of the product.

△Translation by Youdao

Another example is

Public place navigation.

You want to go to the gym, but you can't know the location of the free equipment. Just take a photo of the scene and GPT-4 can guide you to the unoccupied equipment.

△Translation by Youdao

Of course, there are also ways to order food, use vending machines to buy drinks, and search for certain plants Name, give fashion suggestions...

As long as you tell GPT-4 your needs, it can help the visually impaired solve the corresponding problems.

However, this feature is still under testing. You can join the waiting list in the Apple Store. The Android version is coming soon.

The application equipped with it, Be My Eyes, is a charity application developed for blind people.

It was first established as a mutual aid community in 2012, and launched an iOS version of the application in 2015, and then launched an Android version of the application.

This application is divided into two groups: volunteers and blind people. Volunteers will accept photos or videos from blind people and help them solve their difficulties through voice communication (phone calls). Currently, there are nearly 450,000 visually impaired people. and 630w for volunteer use.

If you become a volunteer, you only need to stay online to ensure that calls from visually impaired people can be answered at any time.

If it is used by the visually impaired, you can call the volunteer phone number when you need help, or seek help from some professionals:

Join now After Virtual Volunteer, blind people can also call the "virtual volunteer" GPT-4 for help, without having to worry about no one answering late at night.

The official also played a homophonic meme, AI→Eyes, "Let AI be your eyes":

It also serves as the American version of Zhihu Q&A bot

Of course, in addition to the Be My Eyes app, many other apps are also vying to access GPT-4.

For example, on Poe published by the US version of Zhihu Quora, you can now chat with GPT-4 (limited to one sentence):

Like the AI ​​lawyer software DoNotPay, it has also been connected to GPT-4 and plans to use it to launch a “one-click litigation” service.

#Based on this service, you can report complaints about phone scams that you don’t like with just one click.

If you receive a fraudulent phone call, you only need to click a button, and the corresponding phone call will be recorded in its entirety, and a 1,000-word lawsuit will be generated, claiming a claim for US$1,500. (However, it is currently only applicable in the United States)

It is worth mentioning that DoNotPay CEO Joshua Browder said that they had used GPT-3.5 to do similar functions before, but the effect was not very good, and GPT-4 It has been able to hold this requirement very well.

Some netizens have even tried to use GPT-4 for drug discovery...

In addition to the above applications and In addition to its functions, some netizens have noticed the prospect of GPT-4 developing small games.

The new GPT-4 seems to have improved the reliability of programming, whether it is making a table tennis game in 60 seconds:

Still make a complete greedy snake within 20 minutes:

can be said to be very easy to do, and can be completed in time when making changes. The simple program written is basically There are no bugs that need to be fixed.

The above is the detailed content of As soon as GPT-4 was released, there was access to the mobile APP. You can upload photos and videos for one-click interpretation. It also served as a Q&A bot for the American version of Zhihu.. For more information, please follow other related articles on the PHP Chinese website!