GPT-4: how to use the AI chatbot that puts ChatGPT to shame

People were delighted when ChatGPT came along, impressed by its natural language capabilities as an AI chatbot. But when the long-awaited GPT-4 large language model appeared, it surpassed what we thought was possible with AI, with some calling it the first glimpses of AGI (Artificial General Intelligence).

The model’s creator, OpenAI, calls it the company’s “most advanced system that produces safer and more useful responses.” Here’s everything you need to know about it, including how to use it and what you can do.

Availability

Open laptop on ChatGPT website.Shutterstock

GPT-4 was officially announced on March 13, which Microsoft previously confirmed, though the exact day was not known. However, as of now it is only available on the paid subscription of ChatGPT Plus. The current free version of ChatGPT will still be based on GPT-3.5, which is less accurate and capable by comparison.

GPT-4 is also available as an API “for developers to build applications and services”. Some of the companies that have already integrated GPT-4 include Duolingo, Be My Eyes, Stripe, and Khan Academy. The first public demo of GPT-4 was also streamed live on YouTube, showcasing some of its new capabilities.

What’s new in GPT-4?

GPT-4 is a new language model created by OpenAI that can generate text similar to human speech. Improves the technology used by ChatGPT, which is currently based on GPT-3.5. GPT is an acronym for Generative Pre-trained Transformer, a deep learning technology that uses artificial neural networks to type like humans.

According to OpenAI, this next-generation language model is more advanced than ChatGPT in three key areas: creativity, visual information, and broader context. In terms of creativity, OpenAI says that GPT-4 is much better for creating and collaborating with users on creative projects. Examples of this include music, scripts, technical writing, and even “learning the user’s writing style.”

GPT-4 Developer Live Stream

The longest context also plays a role in this. GPT-4 can now process up to 25,000 words of user text. You can even send a web link to GPT-4 and ask it to interact with the text on that page. OpenAI says this can be useful for creating long-form content, as well as “extended conversations.”

GPT-4 can now also accept images as the basis for interaction. In the example provided on the GPT-4 website, the chatbot is given a picture of various baking ingredients and asked what can be made with them. It is currently unknown if the video can be used in the same way.

Finally, OpenAI also says that GPT-4 is significantly more secure to use than the previous generation. It can reportedly produce 40% more factual responses in OpenAI’s own internal tests, while being 82% less likely to “respond to requests for objectionable content.”

OpenAI says it trained with human feedback to take those steps, claiming to have worked with “more than 50 early feedback experts in domains including AI safety and security.”

In the weeks since its release, users have posted some of the amazing things they’ve done with it, like inventing new languages, detailing how to escape into the real world, and creating complex animations for apps from scratch. While early adopters flocked to get their hands on it, we’re just starting to learn what it’s capable of. One user reportedly got GPT-4 to create a working version of Pong in just sixty seconds, using a combination of HTML and JavaScript.

Where is the visual input in GPT-4?

One of the most anticipated features of GPT-4 is visual input, which allows ChatGPT Plus to interact with images instead of just text. The ability to analyze images would have been a great help for GPT-4, but the feature has been kept because of that. mitigate security challenges, according to OpenAI CEO Sam Altman.

You can get an idea of ​​what visual input can do on Bing Chat, which recently opened up the visual input feature for some users. It can also be tested with a different application called MiniGPT-4. The open source project was created by some PhD students, and while it’s a bit slow for image processing, it shows the types of tasks you’ll be able to perform with visual input once it’s officially introduced to GPT-4 in ChatGPT Plus. .

What are the best GPT-4 plugins?

A Wolfram plugin that performs calculations in ChatGPT.

Plugins are one of the best reasons to pay for ChatGPT Plus, as they open up the possibilities of what the model can do. The two most powerful examples were created by OpenAI itself: Code interpreter and web browser plugins.

By using these plugins in ChatGPT Plus, you can greatly expand the capabilities of GPT-4. ChatGPT Code Interpreter can use Python in a persistent session, and can even handle uploads and downloads. A web browser plugin, on the other hand, gives GPT-4 access to the entire Internet, allowing it to bypass model restrictions and get live information directly from the Internet on your behalf.

Some of the other best GPT-4 plugins include Zapier, Wolfram, and Speak, all of which allow you to take AI in new directions.

How to use GPT-4

Bing Chat displayed on a laptop.Jacob Roach / Digital Trends

The main way to access GPT-4 currently is to upgrade to ChatGPT Plus. To jump to the $20 paid subscription, simply click “Upgrade to Plus” in the ChatGPT sidebar. After entering your credit card information, you will be able to switch between GPT-4 and earlier versions of LLM. You can even double check that you’re getting GPT-4 responses, since they use the black logo instead of the green logo used on older models.

From there, using GPT-4 is identical to using ChatGPT Plus with GPT-3.5. It’s more capable than ChatGPT and allows you to do things like tune your dataset to get custom results that suit your needs.

If you don’t want to pay, there are other ways to experience just how powerful GPT-4 is. For one, you can try it as part of Microsoft’s Bing Chat. Microsoft has revealed that it uses GPT-4 in Bing Chat, which is completely free to use. However, some features of GPT-4 are missing from Bing Chat and it is clear that it is combined with some proprietary technology from Microsoft. But you’ll still have access to that extended LLM (big language model) and the advanced intelligence that comes with it. It should be noted that although Bing Chat is free, it is limited to 15 conversations per session and 150 sessions per day.

there are many others applications that currently use GPT-4 as well, such as the question and answer site, Quora.

What are the limitations of GPT-4?

While discussing the new capabilities of GPT-4, OpenAI also points out some of the limitations of the new language model. Like previous versions of GPT, OpenAI says the latest model still suffers from “social biases, hallucinations, and conflicting claims.”

In other words, it’s not perfect. You’ll still get answers wrong, and there are plenty of examples listed online that show its limitations. But OpenAI says these are all problems the company is working on, and overall GPT-4 is “less creative” with answers and therefore less likely to make up facts.

Another main limitation is that the GPT-4 model is trained to stop internet data in 2021. However, a web browser plugin helps to work around this limitation.

Does Bing Chat use GPT-4?

Microsoft originally claims that the new Bing, or Bing Chat was more powerful than ChatGPT. Since OpenAI Chat uses GPT-3.5, there was an implication at the time that Bing Chat might use GPT-4. And now Microsoft has confirmed that Bing Chat is actually built on top of GPT-4.

However, features like visual input weren’t available in Bing Chat, so it’s still unclear exactly which features are built-in and which aren’t.

Anyway, Bing Chat has apparently been updated with the ability to access current information over the Internet, which is a huge improvement over the current version of ChatGPT, which you can only rely on the training you’ve acquired through 2021. .

Aside from Internet access, the AI ​​model used for Bing Chat is much faster, which is extremely important when taken out of the lab and added to a search engine.

Evolution, not revolution?

StrictlyVC in conversation with Sam Altman, part 2 (OpenAI)

We haven’t tested GPT-4 in ChatGPT Plus yet, but it will surely be more impressive, building on the success of ChatGPT. In fact, if you’ve tried the new Bing Chat, you’ve obviously already tried it. Just don’t expect it to be something completely new.

Prior to the release of GPT-4, OpenAI CEO Sam Altman said in an interview with StrictlyVC that he posted Connie Loizos on YouTube that “people beg to be let down and they will be.”

There, Altman recognized the potential of AGI (artificial general intelligence) to wreak havoc on the world’s economies, saying that the rapid introduction of a few small changes is better than a shocking breakthrough that gives little opportunity for the world to adapt to changes. changes. As impressive as GPT-4 looks, it’s certainly more of a careful evolution than a complete revolution.

editor’s recommendations

Categories: GAMING
Source: newstars.edu.vn

Leave a Comment