Many users in early testing and a business demo the day after it was introduced were astounded by GPT-4’s ability to file lawsuits, pass exams, and create a working website from a hand-drawn design.
The artificial intelligence engine that powers ChatGPT, OpenAI’s popular chatbot tool, was updated recently. With its incredible features, GPT-4 is expected to perform better than previous iterations, which could change the way we work, play, and create online. Yet, it might also prompt more challenging questions about how AI technologies might affect our relationship with technology and transform vocations, make it simpler for students to cheat, and change how we engage with it.
The company’s huge language model, known as GPT-4, was enhanced and is trained on a sizable quantity of web data to deliver smart user speedy replies. It is currently on a waitlist and has already been seen in a few applications from other developers, like the recently launched Bing AI-powered search engine from Microsoft. Early adopter users are highlighting some of the most alluring GPT-4 features and application cases while also sharing their insights. This article will examine the top 5 amazing GPT-4 features that outperform ChatGPT.
Analyzing More Than Text
Essentially, the biggest change is GPT-4’s capacity to function with submitted images. One of the most remarkable use cases to date came from an OpenAI video demo that demonstrated how a sketch could be swiftly converted into a working website. When the image was uploaded to GPT-4, the demonstration incorporated the generated code into a preview to show how a functional website may appear. In addition, OpenAI showed how GPT-4 was tasked with expressing the humor of a series of images that included a smartphone being charged improperly in its introduction. Despite the seeming simplicity, the required context makes it more difficult for artificial intelligence programs to recognize jokes.
In a different test, The New York Times gave GPT-4 a picture of the contents of a refrigerator and asked it to make a meal using those ingredients. The images feature has not yet gone online, but OpenAI plans to do so in the coming weeks.
Making Coding Even Simpler
Some early GPT-4 users with little to no prior experience with coding have also used it to reproduce classic games like Pong, Tetris, or Snake by carefully following the tool’s step-by-step instructions. A few people have developed their own original games. (According to OpenAI, GPT-4 can write code in all the most widely used programming languages.)
The powerful linguistic abilities of GPT-4 will be used for everything from character development to storyboarding to the creation of video game content, according to Arun Chandrasekaran, a Gartner Research analyst. In the future, this may lead to the formation of other independent game suppliers. GPT-4 and comparable models, however, can also be used to write news articles, moderate gaming forums, and create marketing materials for game trailers.
Similar to how gaming has changed, GPT-4 may also alter how applications are created. One person claimed to have programmed a simple sketching app in a matter of minutes on Twitter, while another claimed to have created an app that recommends five new movies every day, along with information on where to watch them and trailers.
Everyone can learn to code, claims Lian Jye Su, an analyst at ABI Research, provided they receive some initial coaching. To learn to drive, she compared it to learning to code. “AI is a capable teacher.”
Passing Exams with Flying Colors
It is “less competent” than humans in many real-world situations, despite OpenAI’s claims that the update performs at “human-level performance” on a variety of professional and academic assessments. The company claims that GPT-4 recently passed a law school’s mock bar exam with a score in the top 10% of test takers. Comparatively, GPT-3.5, the earlier version, scored in the bottom 10%. The most recent version, according to OpenAI, also performed well on the LSAT, GRE, SAT, and several AP exams.
One of the prestigious graduate-level exams that ChatGPT was able to pass in January was the Wharton School of Business test at the University of Pennsylvania, though not with particularly high marks. Using the knowledge gained through its testing program and ChatGPT, the company claimed to have spent months improving the system’s accuracy and capacity to stay on topic.
Providing More Accurate Answers
The company claims that GPT-4 can produce written responses that are longer, more in-depth, and more reliable than the prior version.
The most recent iteration can now react with up to 25,000 words, up from the previous 4,000, and can provide comprehensive instructions for even the most peculiar circumstances, including how to clean a piranha’s fish tank or extract a strawberry’s DNA. Based on a query on a dating profile, one early user reported that it supplied comprehensive pick-up line ideas.
Streamlining Operations Across a Variety of Industries
Joshua Browder, CEO of legal services chatbot DoNotPay, stated that his company is already considering using the tool to draught “one-click lawsuits” to combat robocallers as an early illustration of the immense potential for GPT-4 to revolutionize how people interact across sectors.
“Imagine receiving a call, clicking a button, and the call’s transcript being used to write a 1,000-word complaint. According to a tweet from Browder, GPT-3.5 was insufficient, but GPT-4 is excellent at handling the task.
According to Keeper CEO Jake Kozloski, the company is using the method to better match its clients.
Significant advancements in “connected automobile [dashboards], remote diagnostics in healthcare, and other AI applications that were previously not possible,” according to Su of ABI Research, may also be on the horizon.
