by
CNN
—
On the first day after its revelation, GPT-4 surprised many users in early tests and a company demo with its ability to write lawsuits, pass standardized tests, and create a working website from a hand-drawn sketch.
On Tuesday, OpenAI announced the next-generation version of its AI technology based on its viral chatbot tool, ChatGPT. The most powerful GPT-4 promises to blow previous iterations out of the water, potentially changing the way we use the Internet to work, play and create. But it could also add to challenging questions about how AI tools can upend professions, enable students to cheat and change our relationship with technology.
GPT-4 is an updated version of the company’s large language model, which is trained on massive amounts of online data to create complex responses to user prompts. It’s now available via a waiting list and has already made its way into some third-party products, including Microsoft’s new AI-powered Bing search motor. Some early access users of the tool share their experiences and highlight some of its most exciting use cases.
Here’s a closer look at GPT-4’s capabilities:
At its core, the biggest change in GPT-4 is its ability to work with user-uploaded photos.
One of the most impressive use cases so far came from an OpenAI video demonstration that showed how a the design could be turned into a functional website within a few minutes. The demoer uploaded the image to GPT-4 and then pasted the resulting code into a preview that showed what a working website might look like.
In its announcement, OpenAI also showed how GPT-4 was asked to explain a joke from a series of images – featuring a smartphone with the wrong charger – and described why it was funny. Although it may sound obvious, analyzing a joke it’s more complicated for AI tools to use because of the framework needed.
In another test, The New York Times showed GPT-4 a picture of the inside of a refrigerator and prompted it to come up with a meal based on the ingredients.
The photo feature isn’t live yet, but OpenAI is expected to release it in the coming weeks.
Some early adopters of GPT-4 with very little to no coding knowledge have also used it recreate virtual games like Pong, Tetris or Snake after following the step-by-step instructions provided by the tool on how to do it. Others have 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 to character creation to game content creation,” said Arun Chandrasekaran, an analyst at Gartner Research. “This could lead to more independent gaming providers 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 moderate game discussion boards.”
Similar to games, GPT-4 could change the way people develop applications. One Twitter user said that made a simple drawing app in minutes, while another claimed he had coded an app that recommends five new movies each day, along with providing trailers and details on where to watch them.
“Coding is like learning to drive – as long as the beginner gets some guidance, anyone can code,” said Lian Jye Su, an analyst at ABI Research. “AI can be a good teacher.”
Although OpenAI said the update is “less competent” than humans in many real-world scenarios, it shows “human-like performance” in various professional and academic tests. The company said the GPT-4 recently passed a mock law school bar exam scoring in the top 10% of examinees. In contrast, the previous version, GPT-3.5, scored around the bottom 10%. The latest version it also performed strongly on the LSAT, GRE, SAT, and many AP exams, according to OpenAI.
In January, ChatGPT made headlines for its ability to pass prestigious graduate-level exams, such as one from the University of Pennsylvania’s Wharton School of Business, but not with particularly high scores. The company said it spent months using lessons from its testing program and ChatGPT to improve the system’s accuracy and ability to stay on topic.
Compared to the previous version, GPT-4 is able to produce longer, more detailed and more reliable written responses, according to the company.
The latest version can now provide answers of up to 25,000 words, up from around 4,000 previously, and can provide detailed instructions for even the most unique scenarios, ranging from how to clean a piranha’s fish tank to extracting the DNA of a strawberry . One early user said it provided in-depth pick-up line suggestions based on a question mentioned in a dating profile.
Joshua Browder, CEO of legal services chatbot DoNotPay, said his company is already work on using the tool to create “one-click lawsuits” to sue motorists, in an early indication of GPT-4’s huge potential to change the way people work across industries.
“Imagine receiving a call, clicking a button, [the] the call is transcribed and a 1,000 word lawsuit is created. GPT-3.5 wasn’t good enough, but GPT-4 handles the job extremely well,” Browder tweeted.
Meanwhile, Jake Kozloski, CEO of dating site Keeper, said his company is using the tool to best suited to its users.
According to Su at ABI Research, we are also likely to see major developments in the “connected car [dashboards]remote diagnosis in healthcare and other AI applications previously not possible.”
Although the company has made huge improvements to the AI model, GPT-4 has similar limitations to previous versions. OpenAI said the technology he lacks knowledge of events that occurred before his data set was cut off (September 2021) and does not learn from his experience. It can also make “simple errors of reasoning” or be “overly naive in accepting obvious false statements from a user” and not double-check, the company said.
Gartner’s Chandrasekaran said this also reflects many AI models today. “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 black box in nature.”
For now, OpenAI said GPT-4 users should be cautious and exercise “extreme caution” especially “in high-stakes environments.”
