(CNN) On the first day after its disclosure, GPT-4 surprised many users in early tests and showcased the company with its ability to draft lawsuits, pass standardized tests, and build a working website from a hand-drawn sketch.
On Tuesday, OpenAI announce The release of the next generation of AI technology that powers the viral chatbot tool, ChatGPT. GPT-4’s strongest promises To get previous iterations out of the water, which could change the way we use the Internet for work, play, and creativity. But it could also add to difficult questions about how AI tools can upend careers, enable students to cheat, and transform our relationship with technology.
GPT-4 is an updated version of the company’s Big Language Model, which is trained on huge amounts of online data to generate complex Responses to user prompts. It’s available now via a waitlist and has already made its way into some third-party products, including Microsoft’s new AI-powered Bing search. engine. Some users who have early access to the tool share their experiences and highlight some of their most compelling use cases.
Here’s a closer look at the capabilities of GPT-4:
Analyze more than one text
In essence, the biggest change to GPT-4 is its ability to work with images that users upload.
One of the most amazing use cases to date has come from View OpenAI video that showed how a The graphic can be turned into a functional website within minutes. The viewer uploaded the image in GPT-4 and then pasted the resulting code into a preview showing how it could be a working website.
In its announcement, OpenAI too show up How GPT-4 was asked to explain a joke from a series of photos – which showed a smartphone with a faulty charger – and describe why it was so funny. While it may seem obvious, anatomy is a joke It is more complex for AI tools to capture because of the context required.
In another test, the New York Times show up GPT-4 Image of the interior of a refrigerator and its drive to create an ingredient-based meal.
The photo feature hasn’t been published yet, but it’s expected to roll out to OpenAI in the coming weeks.
Make coding easier
It was also used by some early GPT-4 users with very little or no knowledge of code Recreate iconic games like PongOr Tetris or Snake after following the step by step instructions provided by the tool on how to do it. Others made their own original games. (GPT-4 can write code in all major programming languages, according to OpenAI.)
“GPT-4’s powerful language capabilities will be used for everything from storyboarding, character creation to game content creation,” said Arun Chandrasekaran, an analyst at Gartner Research. “This may lead to more independent game providers emerging in the future. But beyond the game itself, GPT-4 and similar models can be used to create marketing content around game previews, create news articles and even edit game discussion boards.”
Similar to games, GPT-4 can change the way people develop applications. One user on Twitter said they Made a simple drawing app Within minutes, while another claimed to have encrypted An app that recommends five new movies each day, along with providing trailers and details on where to watch them.
“Programming is like learning how to drive – as long as a novice gets some guidance, anyone can code,” said Lian Jye Su, an analyst at ABI Research. “Artificial intelligence can be a good teacher.”
Pass tests with flying colors
Although OpenAI He said Update is “less capable” than humans in many real-world scenarios, displaying “human-level performance” on various professional and academic tests. The company said that GPT-4 recently passed the law school’s bar simulation exam with a score of nearly 10% of the test takers. By contrast, the previous version, GPT-3.5, scored around 10%. The latest version He also performed strongly on the LSAT, GRE, SAT, and several AP tests, according to OpenAI.
In January, ChatGPT made headlines for its ability to Pass the prestigious postgraduate exams, such as one from the Wharton School of Business at the University of Pennsylvania, but not with a particularly high score. The company said it spent months using lessons from its testing software and ChatGPT to improve the system’s accuracy and ability to stay on topic.
provide more accurate responses
Compared to the previous version, GPT-4 can produce longer, more detailed, and more reliable typing responses, according to the company.
The latest version can now provide responses of up to 25,000 words, up from about 4,000 words previously, and can provide detailed instructions for even the most unique scenarios, from how to clean a piranha tank to extracting the DNA of a strawberry. One early user said he made in-depth suggestions for pickup lines based on a question included on his dating profile.
Simplify work in various industries
Joshua Browder, CEO of legal services chatbot DoNotPay, said his company is already in the works Work on using the tool to create “one-click lawsuits” to sue robo-dialers, an early indication of the huge potential of GPT-4 to change the way people work across industries.
Imagine receiving a call, clicking a button, [the] The call was transcribed and a 1,000-word lawsuit was created. GPT-3.5 wasn’t good enough, but GPT-4 handles the job very well,” Browder tweeted.
Meanwhile, Jake Kozlowski, CEO of dating site Keeper, said his company is in action using the tool to better match with their users.
According to Su at ABI Research, it’s also possible that we’ll see big developments in the “connected car [dashboards]remote diagnostics in healthcare and other AI applications that were not previously possible.”
Work in progress
Although the company has made significant improvements to its AI model, GPT-4 has similar limitations to previous versions. OpenAI technology said He lacks knowledge of the events that occurred before his data set went out (September 2021) and does not learn from his experience. The company said it could also make “minor logical errors” or be “extremely naive in accepting obvious false statements from a user,” not checking back to work.
This also reflects many models of artificial intelligence today, Gartner’s Chandrasekaran said. “Let’s not forget that these AI models are not perfect,” Chandrasekaran said. “They can produce inaccurate information from time to time and can be nature’s black box.”
For now, OpenAI has said that GPT-4 users should exercise caution and use “extreme caution” especially “in high-risk contexts.”
“Certified music scholar. Freelance analyst. Social mediaholic. Hipster-friendly web nerd. Zombie buff.”